
AI Learning Lab
aiLL 6/11/23 - 1 - How ChatGPT Will Impact Every Business

Video2023-09-041:58:156 views
Description
I go deep in this 90-minute live session on how ChatGPT and generative AI will impact every industry. I walk through real examples of using AI to build apps, websites, marketing campaigns, and more. Get a look at the future of work and how to prepare your business. June 11, 2023 sees rapid AI innovation that will only accelerate. Join me on the AI Learning Lab as I demo the latest capabilities. AI Learning Lab address: https://tiktok.com/@aiLearningLab #ai #chatgpt #futureofwork
0:20 - Introducing the AI Learning Lab and agenda for the day
0:43 - Using AI to help write memoirs and books
1:34 - Parallels between the web in 1990s and AI today
2:47 - AI's potential impact on poverty and prosperity
4:43 - The exponential growth of AI adoption
5:02 - Starting an online business with no code tools
7:24 - Evaluating the quality of ChatGPT plug-ins
8:11 - Using AI to improve your own prompts
9:49 - Experimenting with open source AI models
10:26 - Paying for ChatGPT Plus features
14:01 - Preparing for the AI job disruption over the next 2-3 years
15:03 - Using ChatGPT's new internet connectivity
16:50 - Automating business processes with Zapier and AI
18:05 - Developing a strategic mindset to leverage AI
20:01 - Integrating AI into local computer
23:13 - Using AI to generate names and branding
24:36 - Creating customized "bots" with AI
27:07 - Using AI to improve its own capabilities
28:38 - Preparing for the rapid acceleration of AI progress
31:20 - Using AI without relying on coding skills
32:12 - AI's potential to increase prosperity, not poverty
33:04 - Teaching AI using your own personalized style and tone
34:41 - Using AI to help with business strategy and consulting
36:42 - Leveraging AI's internet connectivity for research
37:45 - Improving AI's sourcing over time
38:59 - Comparing ChatGPT results across models
39:49 - Turning off data sharing and privacy controls
42:46 - The exponential pace of current AI progress
43:20 - Iteratively improving AI outputs with feedback
44:04 - Documenting AI learning journey for future reference
46:23 - Anthropomorphizing AI tools as "conscious"
47:45 - Google's critical invention of Transformers
48:51 - Installing AI models locally with GPT-forall
51:21 - Avoiding the cutting edge of complex AI development
52:34 - Thinking problem-first, not AI-first
53:41 - Automating personalized outreach at scale with AI
56:14 - The exponential acceleration of AI progress
57:24 - Protecting privacy as public AI tools improve
58:31 - Leveraging AI's internet connectivity
59:37 - Democratizing access to quality AI1:00:09 - Using AI to generate images and avatars
1:01:02 - Applying AI to interactive intelligence and ML solutions
1:02:01 - Valuing strong communication skills in the AI era
1:04:41 - Using AI to explain complex technical processes
1:06:23 - Providing a public service directing people to quality AI
1:07:02 - Generating customized images by incorporating photos
1:08:11 - Examining AI copyright issues around training data
1:09:49 - Evaluating the exponential growth in AI tools
1:10:26 - Adapting expectations as AI tools remain in beta
1:11:20 - Leveraging AI's memory and knowledge retention
1:14:38 - Fact checking AI's confident but wrong answers
1:15:46 - Respecting AI's limitations despite hype
1:16:50 - Distinguishing between AI apps and base models
1:17:43 - Waiting for the open source AI dust to settle
1:19:31 - Comparing capabilities across AI tools and models
1:20:34 - Preparing for complex AI workflows like Stable Diffusion
1:22:04 - Using AI to improve and expand prompts
1:23:07 - Automating personalized outreach with AI
1:25:09 - Navigating the confusing AI landscape
1:26:14 - Respecting the suffering caused by change
1:27:07 - Amplifying capabilities with co-pilot AI tools
1:29:49 - Developing patient expectations of early AI tools
1:30:59 - Maintaining skepticism of AI's confident errors
1:32:14 - Democratizing access to quality AI models
1:35:30 - Appreciating AI's amplification of imagination
1:36:42 - Developing strategic mindset to leverage AI's capabilities
1:39:16 - Scrutinizing AI outputs lacking sophistication
1:40:40 - Building intuitive understanding to correct misconceptions
1:41:54 - Recognizing AI as an amplifier of human strengths
1:43:56 - Documenting the learning journey with AI
1:45:39 - Keeping expectations realistic about manual production
1:46:32 - Anthropomorphizing AI tools as conscious
1:47:38 - Google's critical creation of Transformers for NLP
1:50:32 - Leveraging the of AI tools' limitations
1:51:45 - Asynchronously answering questions at scale
1:53:08 - Securing private data with business AI tools
1:55:06 - Using AI to explain technical concepts conversationally
1:56:14 - Preparing for exponential acceleration of progress
1:57:17 - Documenting learning journey for future reference
Chapters
0:00<Untitled Chapter 1>0:20Introducing the AI Learning Lab and agenda for the day0:43Using AI to help write memoirs and books1:34Parallels between the web in 1990s and AI today2:47AI's potential impact on poverty and prosperity4:43The exponential growth of AI adoption5:02Starting an online business with no code tools7:24Evaluating the quality of ChatGPT plug-ins8:11Using AI to improve your own prompts9:49Experimenting with open source AI models10:26Paying for ChatGPT Plus features14:01Preparing for the AI job disruption over the next 2-3 years15:03Using ChatGPT's new internet connectivity16:50Automating business processes with Zapier and AI18:05Developing a strategic mindset to leverage AI20:01Integrating AI into local computer23:13Using AI to generate names and branding24:36Creating customized "bots" with AI27:07Using AI to improve its own capabilities28:38Preparing for the rapid acceleration of AI progress31:20Using AI without relying on coding skills32:12AI's potential to increase prosperity, not poverty33:04Teaching AI using your own personalized style and tone34:41Using AI to help with business strategy and consulting36:42Leveraging AI's internet connectivity for research37:45Improving AI's sourcing over time38:59Comparing ChatGPT results across models39:49Turning off data sharing and privacy controls42:46The exponential pace of current AI progress43:20Iteratively improving AI outputs with feedback44:04Documenting AI learning journey for future reference46:23Anthropomorphizing AI tools as "conscious"47:45Google's critical invention of Transformers48:51Installing AI models locally with GPT-forall51:21Avoiding the cutting edge of complex AI development52:34Thinking problem-first, not AI-first53:41Automating personalized outreach at scale with AI56:14The exponential acceleration of AI progress57:24Protecting privacy as public AI tools improve58:31Leveraging AI's internet connectivity59:37Democratizing access to quality AI1:00:09 - Using AI to generate images and avatars1:01:02Applying AI to interactive intelligence and ML solutions1:02:01Valuing strong communication skills in the AI era1:04:41Using AI to explain complex technical processes1:06:23Providing a public service directing people to quality AI1:07:02Generating customized images by incorporating photos1:08:11Examining AI copyright issues around training data1:09:49Evaluating the exponential growth in AI tools1:10:26Adapting expectations as AI tools remain in beta1:11:20Leveraging AI's memory and knowledge retention1:14:38Fact checking AI's confident but wrong answers1:15:46Respecting AI's limitations despite hype1:16:50Distinguishing between AI apps and base models1:17:43Waiting for the open source AI dust to settle1:19:31Comparing capabilities across AI tools and models1:20:34Preparing for complex AI workflows like Stable Diffusion1:22:04Using AI to improve and expand prompts1:23:07Automating personalized outreach with AI1:25:09Navigating the confusing AI landscape1:26:14Respecting the suffering caused by change1:27:07Amplifying capabilities with co-pilot AI tools1:29:49Developing patient expectations of early AI tools1:30:59Maintaining skepticism of AI's confident errors1:32:14Democratizing access to quality AI models1:35:30Appreciating AI's amplification of imagination1:36:42Developing strategic mindset to leverage AI's capabilities1:39:16Scrutinizing AI outputs lacking sophistication1:40:40Building intuitive understanding to correct misconceptions1:41:54Recognizing AI as an amplifier of human strengths1:43:56Documenting the learning journey with AI1:45:39Keeping expectations realistic about manual production1:46:32Anthropomorphizing AI tools as conscious1:47:38Google's critical creation of Transformers for NLP1:50:32Leveraging the of AI tools' limitations1:51:45Asynchronously answering questions at scale1:53:08Securing private data with business AI tools1:55:06Using AI to explain technical concepts conversationally1:56:14Preparing for exponential acceleration of progress1:57:17Documenting learning journey for future reference
Transcript
0:01 thank you 0:16 hello hello 0:18 it's Kyle Shannon with the AI learning 0:21 lab hello Emilio's wife what's happening 0:27 welcome everybody it's a fine Sunday 0:30 here in Denver we're about to get some 0:33 crazy ass storm 0:35 it's supposed to rain between one and 0:37 three inches with possible hail yay 0:43 ah yes waiting on class yeah exactly 0:49 uh this is the AI learning lab uh 0:51 welcome we'll wait for some people to 0:53 come in my name is Kyle Shannon I'm an 0:54 entrepreneur 0:55 I'm a dude that's just very into 0:58 generative Ai and we'll answer any 1:01 questions I can 1:04 um 1:05 that's what this channel is all about 1:07 I've also got the AI Salon I've got a 1:11 newsletter called everyday AI so 1:14 looking at 1:16 um all of what this stuff makes possible 1:18 and I'm putting away a screwdriver 1:21 because I because I always forget to put 1:23 away screwdrivers and then I was like 1:24 where the screwdriver 1:27 all right done 1:29 look at that 1:31 like a professional uh welcome this AI 1:35 learning lab if you have questions about 1:36 AI pop them uh into the 1:40 uh what's it called comments 1:45 just getting up to speed I mowed today I 1:49 covered with raised garden beds I 1:51 covered them with cloth 1:53 I haven't done much with AIT so I 1:56 figured I'd help on here and talk about 1:57 it rather than doing something 2:01 all right what did you say here 2:05 um I'm excited I'm gonna help my husband 2:06 write his life story yeah with AI of 2:08 course that's great I mean yeah that's 2:11 the kind of thing like like you could 2:12 even have it you know help you outline 2:15 it or even just have it come up with the 2:17 questions for him to answer and then 2:19 have it organize all the answers that's 2:20 really cool it's a great idea 2:23 how do I profit from AI do not listen to 2:25 people that say you can make money 2:29 don't listen to them 2:31 um here's a deal with AI it's a tool it 2:33 happens to be a really powerful tool and 2:35 there's a lot of them so 2:37 a couple of things the first step in 2:40 profiting with AI is start learning it 2:44 so if you haven't actively been using 2:47 chat GPT that's what that top URL is I'm 2:50 not associated with them go to that top 2:53 URL that's the official chat website 2:55 start playing with it as I Babble on 2:57 here on a Sunday afternoon go play there 3:00 the next one down is bing.com which is 3:02 also chat GPT if you click on the chat 3:04 button that's gpt4 connected to the 3:07 internet for free right so if you're 3:09 like I can't pay for this go to Bing 3:12 in my opinion not as strong as the the 3:16 Open ai's chat GPT even though it's the 3:18 same core underlying technology they 3:20 have it a bit hobbled 3:21 and then poe.com lets you play with a 3:23 bunch of different models prompts.chat 3:25 will teach you about prompting and then 3:27 futurepedia.io is a big directory of 3:29 just like as many AI tools as you can 3:32 imagine there's probably thousands at 3:33 this point I haven't even been recently 3:35 but if you want to geek out you can go 3:37 there but I think Step One is is start 3:40 learning chat GPD start learning what 3:42 these things make possible 3:44 um if you have questions about AI pop 3:45 them into the comments below I'll do my 3:47 best to answer them uh fair warning I 3:50 cuss uh I make stupid jokes uh I act 3:54 like silly characters I get accused of 3:56 looking like famous or semi-famous 3:58 people on a pretty regular basis Rosie 4:01 O'Donnell's my favorite uh but fat Alec 4:03 Baldwin's not bad 4:08 um all right let's see let me let me go 4:10 through some questions here some 4:11 questions here how do I profit oh the 4:13 other thing about profit profiting with 4:15 AI 4:16 learn what these tools can do and then 4:19 take what you're good at and 4:23 come up with a solution to a problem you 4:25 think people be willing to pay for 4:26 that's business that's how you make 4:28 money right that problem could be you 4:30 want to make people smile and you start 4:32 a t-shirt business that business could 4:35 be a Consulting business that business 4:36 could be helping people come up with 4:38 marketing plans so take something that 4:41 that is in line with who you are and 4:44 what you're good at learn these AI tools 4:46 and how they can amplify what you're 4:48 good at and then come up with something 4:50 that you can put out to the world and 4:52 get people to pay for a good way to 4:54 start that is to give away those 4:56 services for free so if you come up with 4:59 I don't know whatever it is I I'm going 5:01 to help people write a marketing plan 5:02 based on the hero's journey I don't know 5:05 there's one that'd be one chat CPT would 5:07 be good with 5:09 um 5:09 offer it to three business owners that 5:11 you know for free to just do it to 5:14 actually put it in the world and get 5:16 feedback because you can think about 5:18 this [ __ ] all you want but when you put 5:20 it in the world things change and then 5:22 from there you can figure out 5:24 um 5:26 then you can figure out 5:29 um you know how you could really turn 5:30 that into a business 5:32 um yeah I got out of the house today and 5:34 I I wouldn't call mowing really getting 5:36 out of the house but you know 5:38 um what are some applications in AI in 5:41 cyber security 5:43 um well I I think there's just I think 5:46 there's just cyber security and then 5:48 there's going to be 5:50 um 5:50 the the Bad actors have access to really 5:53 powerful tools and then the good actors 5:55 have access to really powerful tools so 5:57 I think I feel that cyber security is 5:59 just going to be 6:01 um an escalated 6:03 continuation of the battle between the 6:06 black hats and the white hats again 6:08 these things are just tools there's 6:09 there's nothing there's there's nothing 6:12 well 6:16 the generative AI stuff put into regular 6:21 non-technical people's hands more power 6:23 than we've ever had in the history of 6:25 humanity so it's different in that sense 6:28 but these things are not necessarily 6:31 doing things on their own it's they're 6:33 more like a co-pilot which is That's The 6:35 Branding that Microsoft uses this is a 6:37 co-pilot you're the pilot these tools 6:39 are a co-pilot think to think of them 6:41 like that and then figure out what do 6:43 you want to do so if you're good at 6:44 cyber security I would dig deep in you 6:48 know start hanging out on Reddit and 6:50 figuring out how hackers are you know 6:53 using these generative tools to 6:56 attempt to infiltrate infiltrate systems 6:59 get get good at understanding that and 7:01 then come up with counter measures I 7:03 think that there's there's going to be a 7:05 huge need for that kind of 7:08 um support in the cyber security Arena 7:10 as as to what specific tools I don't 7:13 know that world at all so go to 7:14 futurepedia.io there might be some tools 7:17 listed there that are specific but 7:18 really go hang out on Reddit and you 7:20 know troll GitHub and hugging face and 7:22 see what see what the hackers are doing 7:24 because 7:27 I'm sure they're having fun right now 7:29 because these tools are insane 7:34 is a i m a a good textbook I don't know 7:38 it um I've heard it's good but I don't 7:41 know 7:41 thanks for telling me that you curse 7:43 you'll be leaving now okay cool awesome 7:47 all right does it have a good MD mdp 7:51 examples I don't know what that means 7:52 let me go look let's go to let's go to 7:54 chatgpt 7:57 and see what mdp 7:59 um 8:01 what is wait what is that not typing 8:04 what is archive decision process 8:11 it's a decision maker interacts with an 8:14 environment in discrete time steps 8:16 components of mdp are as follows States 8:20 actions transition I don't know anything 8:23 about this but it obviously knows what 8:26 it is and I'm just using GPT 3.5 so I'm 8:29 not even using four and then I don't 8:32 know let's say 8:33 can you give me some examples 8:41 certainly autonomous driving Inventory 8:43 management Health Care treatment energy 8:46 management 8:48 these are some examples 8:50 um can you give me some less obvious 8:55 ones 9:00 um 9:03 website optimization resource so so yeah 9:05 and so what I would say if you haven't 9:07 played with chat gbt it is 9:11 it 9:12 it's it's kind of hard to explain 9:16 like 9:17 how how insanely different it is so so 9:20 we've been trained for the past two 9:22 decades or so from Google that you type 9:25 in a really sparse 9:27 prompt and then you get back lots of 9:29 answers and you have to go dig through 9:30 it this is much more like having a 9:34 conversation really with a really smart 9:35 intern and it'll even do things like 9:38 this like I say 9:40 um give me a table comparing 9:44 the 9:46 diff uh let's see 9:49 the different use cases 9:53 and which will benefit 9:57 the most from this approach I don't even 10:01 know if that's an appropriate statement 10:03 but it will go on ahead and just write 10:05 you a table and format it 10:09 so high benefit moderate benefit High 10:12 benefit so it's doing some analysis of 10:15 you know these different things within 10:17 that framework so so again what I would 10:20 say is if you haven't played with these 10:22 tools the reason I put 10:24 these things on screen while I'm 10:26 babbling is 10:28 you know those are 10:31 those are the places you should go just 10:32 go start playing go start digging in so 10:35 I hope that helped 10:40 um 10:41 if you've got questions about generative 10:44 AI pop them in the comments I'll do my 10:45 best to answer them I'll show examples 10:47 if I can 10:49 um I'll wax poetic about the future of 10:51 AI I tend to be a tech Optimist um I I 10:55 was accused once on a panel so a woman 10:57 who's been in AI for 15 years said uh 11:01 let me interrupt the adorable optimism 11:03 here so I've been accused of being 11:05 adorably optimistic 11:08 that's the case I I think it's you know 11:11 I find the doomsday 11:15 um future focusing too much on that I 11:17 just find it boring I think it's like 11:18 yeah that's what Hollywood's trained us 11:21 to think so what if it were the opposite 11:23 what if rather than killing us it made 11:25 us all better that's the future I'm 11:27 trying to explore 11:30 um 11:32 when you used 11 labs to clone your 11:35 voice how much data did you upload a 30 11:38 minute MP3 file enough 11:41 I had 11:42 um I had Chachi BT write me a script and 11:48 and I told it I was doing it for for 11:50 voice for for synthesizing my voice so I 11:52 said use as many like you know 11:55 whatever's 11:57 appropriate to train that and then it 11:59 wrote one and then I had it write a 12:01 second one and just like make it 12:02 different and then I read both of them I 12:04 think the total file that I uploaded was 12:07 uh five minutes it wasn't 30. but you 12:11 can you can put it I think they said you 12:12 can put in up to 60 or something like 12:14 that I don't I don't know what the I 12:16 don't know if there's a point of 12:17 diminishing returns but five seemed to 12:20 get it it didn't seem to capture any 12:23 sort of cadence 12:25 um or or you know anything like that so 12:30 uh 12:31 uh yeah Markov decision process it can 12:34 be used to build AI tools and stuff cool 12:36 wait was that never never mind I'm 12:40 asking technically questions about AI 12:42 yeah so so just a little bit about who I 12:45 am I am not a machine learning and AI 12:47 guy 12:48 um I'm an entrepreneur so I I've started 12:50 12 or 15 companies over the years I was 12:54 um there when the when the World Wide 12:56 Web first came onto the scene in the mid 12:58 90s I was living in New York City I 13:01 started one of the first online 13:02 magazines I started one of the first 13:03 digital agencies so 13:06 the parallels to what's going on with 13:08 generative AI right now and what was 13:10 happening with the World Wide Web back 13:12 then are are really really uncanny and 13:15 so I'm sort of coming at this as 13:17 probably a representative of the 13:19 inflection point that we're in Thunder's 13:22 coming 13:23 see if it's really loud 13:24 no 13:26 um 13:27 where 13:28 for the past few decades 13:31 you know machine learning and Ai and 13:34 neural networks have been the the 13:35 purview of data scientists and 13:38 researchers and things like that chat 13:40 EBT marks this inflection point of you 13:44 know real people using these tools and 13:46 so I'm coming at this as an entrepreneur 13:48 saying hey there's these remarkable 13:50 tools here let's let's take off so 13:52 anyway 13:52 all right 13:55 um good thing I'm learning about uh 13:57 technical stuff behind AI I'm going to 13:59 take all your jobs yeah it's you know 14:01 what's uh what's Paul rates to say AI is 14:04 not going to take your job but somebody 14:05 using AI will I mean having a clue about 14:08 AI at any level of the sport right now 14:11 is probably a good thing the next uh the 14:14 next two or three years there's going to 14:16 be a large hunger for people that have a 14:18 clue so but thank you for the question 14:21 um if you have other questions about AI 14:23 pop them into the thing if you've tried 14:25 it share your experiences I'm happy to 14:28 talk about that 14:30 um 14:31 if you haven't used chat GPT I'll show 14:33 you a couple of things here 14:35 um 14:36 there's some new things that they've 14:38 given people access to in the past two 14:40 or three weeks 14:42 um one of them is GPT connected to the 14:45 intervals is that sorry I'm trying to 14:47 get it to focus focus 14:49 yeah 14:51 come on camera 14:53 well that's as good as it's gonna get 14:55 um 14:58 regular chat GPT the base model only has 15:01 data up to September of 2021 by 15:04 connecting it to the internet 15:06 with this second option here 15:08 um you can see more modern things and 15:11 then it's also got this thing called 15:12 plugins and plug-ins are I think it's up 15:15 to four it was it was up to 400 as of 15:17 last night 15:19 um plug-ins are just what they sound 15:21 like they're little like application 15:23 helpers that you can use to augment 15:26 um chat EBT so if you want to plan a 15:28 trip yeah there's 97 pages of four 15:31 plug-ins each 15:32 so there's a lot of them to go through 15:35 um 15:37 let's see let's see what some new ones 15:39 are 15:40 um Clinic manages your 15:43 clink clink clink account 15:46 go find deep memory 15:49 don't know that Pro app paraphraser say 15:53 it better say it different 15:55 paraphrase like a pro paraphrase for 15:58 impact and inspire 16:02 that's really poor English so no I'm not 16:05 going to use paraphrase Pro paraphraser 16:08 get your get your get your description 16:12 of your plugin in in decent 16:15 grammar first and then I might use a 16:17 tool that actually purports to to 16:20 improve improve my uh my language 16:23 [Laughter] 16:25 amazing Tucker elevate your language 16:27 learning to any level with personalized 16:30 one-on-one lessons 16:33 a lot of these plugins are just weird so 16:36 so I think what a lot of the plugins are 16:38 is people are just sort of pointing them 16:40 to their website and then you can have 16:42 chat EBT go look at their specific 16:44 website and pull some [ __ ] in but a lot 16:47 of these things what they're saying that 16:49 the plugin does chat already does really 16:51 well 16:53 like if you want Chachi BT to paraphrase 16:55 something you just tell it to paraphrase 16:57 something tell it to paraphrase it like 16:59 William Shakespeare would and you'll you 17:02 know you don't need a separate tool for 17:04 that here's one that's will find you 17:07 Apartments 17:10 um 17:10 shop best word counter yeah a lot of 17:13 these a lot of these plugins are just 17:15 trash 17:17 to rash to rash all right 17:21 um 17:23 what are a few must-have plugins I don't 17:25 so let's see let me see what I've got 17:28 installed 17:31 so I've got ask your PDF I use that a 17:34 decent amount so ask your PDF 17:36 um 17:37 I haven't 17:39 I've used it more than chat with PDF 17:41 probably just because it began with an A 17:43 and the other one began with a c 17:47 um so I used it first and it worked 17:49 um brand fetch is really interesting 17:51 let's I'll show you brand fetch let's go 17:53 play with that one that one's kind of 17:54 cool 17:55 um so I've got right now I've got so the 17:57 way plugins Works once once you have 17:59 them them enabled 18:01 you go to the plug-in store where you 18:03 you don't actually buy anything you just 18:05 install them once they're installed then 18:07 in any given chat session you can 18:10 activate up to three so like right now 18:12 I've got three travel ones activated so 18:14 I'm going to turn those off 18:18 and then I'm gonna go find that brand 18:20 one brand Fetch and what brand fetch 18:23 does so I can say 18:25 um get me the brand uh let me get me the 18:29 logos 18:32 um colors and any 18:36 brand 18:38 guidance 18:41 for Apple 18:44 and 18:48 it should fire off that plug-in 18:55 it did 19:00 and then 19:03 icon logo Dark theme light theme their 19:06 colors 19:08 their fonts 19:10 images there's a banner 19:14 and then I don't know what these little 19:16 previews are they're not previewing 19:17 anything so let's go see so sometimes 19:19 these these plugins will just return 19:21 return 19:22 hyperlinks that are broken but let's go 19:24 see if if this actually gives us an 19:26 Apple icon 19:29 uh I don't see anything 19:32 so this could be a bust is it still 19:35 loading oh it might be my Wi-Fi is trash 19:38 thanks Xfinity you're the best doesn't 19:41 seem to be loading the pages 19:44 foreign 19:53 weight loading 20:01 is there anything there 20:05 all right anyway I used that one the 20:07 other day and it I did it with my own 20:09 company and it worked pretty good but it 20:12 doesn't seem to be working for Apple 20:14 Banner is there anything there 20:17 iPhone 14 hello yellow 20:20 so it grabbed a banner anyway 20:22 like that that's cool in concept the the 20:25 thing about the the plugins right now is 20:27 the plug-in architecture itself is very 20:30 very simple they're very easy to develop 20:34 um 20:35 I don't think they're really vetting 20:36 them a lot I don't think a lot of people 20:38 are using them so the other thing about 20:39 plugins if you're new to chat EBT in 20:42 order to get access to plugins you have 20:43 to be using chat GPT plus so that little 20:45 yellow Banner there which 20 bucks a 20:47 month and then you have to enable the 20:49 beta features so if I go to my name and 20:52 I go to settings 20:53 I go to Beta features and then there's 20:56 browse with Bing and plugins you have to 20:58 turn them on and the the thing about the 21:00 plugins right now is they're a little 21:01 janky so the question the question was 21:04 what are some 21:06 um some must haves so 21:09 brand fetch isn't one of them although 21:11 you think it's good 21:13 um ask your PDF is really good there's 21:15 chat with video there's a bunch of them 21:17 that you can point to a YouTube video 21:18 and you can you know interact with the 21:21 basically with the transcription of it 21:23 diagram will draw some things that's 21:25 pretty good 21:26 um some of the travel ones are good you 21:28 know kayak get your guide 21:30 um instacart you can get groceries 21:32 delivered so if you have like instacart 21:35 and 21:37 um OpenTable installed you could say you 21:40 know book me reservations at a 21:42 restaurant on Saturday 21:43 give me a recipe for I don't know beef 21:47 stew on Sunday and order the ingredients 21:51 from from uh instacart and it'll go do 21:55 that 21:57 um link reader's good so so you can put 22:00 in a specific URL or a series of them 22:02 and it will go read those pages and then 22:05 you can interact with that stuff 22:07 prompt perfect is kind of fun let me 22:09 show you this one 22:11 so I just enabled it if you type the 22:13 word perfect before your prompt perfect 22:15 and then I type in 22:17 um 22:18 let's say 22:20 um 22:22 give me some advice 22:28 um for how to 22:32 protect my garden 22:36 my Garden from hail since it's about to 22:41 hail here 22:42 and then because I type perfect in front 22:45 of it it fires it off to prompt perfect 22:48 and if you open up the plug-in so it 22:51 basically sent my 22:53 you know crappy prompt and it says can 22:56 you provide me with detailed 22:57 recommendations on the most effective 22:59 methods to safeguard my garden from hail 23:02 damage so it kind of rewrites it and 23:04 improves it some usually it rewrites it 23:07 better than that but I guess my I guess 23:09 my prompt was wasn't bad 23:12 um but that's it that's an interesting 23:13 one 23:16 um what other ones are interesting 23:21 show me is another diagramming Thing 23:23 video insights Vox script web pilot wolf 23:27 from alpha is good you can go off and do 23:31 mathematical operations I don't know 23:33 enough about math to 23:35 even know what the hell to do there 23:38 um but that's a pretty cool one and then 23:41 X papers um that'll take you to research 23:43 papers and you can go like go find the 23:45 latest research papers and then zapier 23:47 it will help you create automations 23:50 automatically so hope that helps 23:53 um let's see oh there's a few new people 23:54 here so if you're new here my name is 23:56 Kyle Shannon this is the AI learning lab 23:59 um I'm here to 24:00 talk about generative AI I'm an 24:03 entrepreneur 24:04 and a and a thirsty person hold on 24:13 if you have questions just pop them in 24:15 I'll do my best to answer them these 24:17 URLs here 24:18 the top one is the official chat web 24:21 website if you've not played with chat 24:22 CPT go do it now like even if you don't 24:25 sit here and listen to me Babylon about 24:27 this [ __ ] go start playing with it this 24:29 stuff is coming faster than I think any 24:31 of us are prepared for and then bing.com 24:33 and po.com are other options to use chat 24:36 things 24:40 um 24:42 do something fun like have it have it 24:44 write you a song 24:46 or write a poem for your for your pet uh 24:51 all right let me go let me go back and 24:53 find some questions here 24:55 how do you how do we enable and install 24:57 plugins I just showed that so hopefully 24:59 that covered that 25:01 is poe like a master dashboard to manage 25:04 all of your subscriptions in one place 25:06 kinda so let's go look at quite know why 25:09 they're doing it the way they're doing 25:10 it so 25:11 all right so here's all the all the 25:13 stuff 25:15 um on the right hand side is just 25:18 um like a normal 25:20 hang on 25:24 like a normal chat GPT window like you 25:26 know write me a poem 25:31 about the rain bang and it will go write 25:35 me a poem about the rain okay so on the 25:38 left hand side you have all these 25:39 different models 25:41 so you've got Sage which is pose model 25:44 then you have three that are there that 25:47 you have to pay for if you if you want 25:49 access to them and I do pay for them you 25:52 see that they say subscribed 25:54 um because of the third the fourth one 25:55 down uh so GPT 4 is gpt4 Claude plus is 25:59 anthropic's version of gpt4 26:03 and then Claude instant 100K is like GPT 26:06 3.5 but it's got a hundred thousand 26:09 token context windows so if you're 26:10 completely new to this that'll mean 26:12 absolutely [ __ ] nothing 26:15 um 26:16 normal gpt's got a 4 000 token context 26:19 Windows what it's called it basically 26:21 means you can put in up to about 3 000 26:23 words 26:24 into your prompt you know your prompt in 26:27 the response can only be about three 26:28 thousand words or it chokes it can't do 26:31 it 26:33 um which sounds like a lot but but let's 26:34 say you want to take a really long 26:36 document and you want to paste it in 26:37 there as part of your prompt and say 26:39 help help me organize this thing or 26:41 whatever it might be 26:42 um three thousand three thousand words 26:44 gets eaten up pretty quick so with a 26:46 hundred thousand token model you can 26:48 paste in up to 75 000 words in your 26:52 uh in your prompt um yesterday I I I 26:55 pasted in the entire script of The 26:57 Godfather and had it tell me about all 26:59 the characters and what happened to them 27:01 and it was just like bang so that's what 27:02 that is 27:04 um Claude instant is like three GPT 3.5 27:07 chat EBT is GPT 3.5 and then the things 27:12 you have below here are Bots and and and 27:14 what the Bots are are 27:17 um 27:17 like little sort of pre-written prompts 27:20 for lack of a better term so these top 27:23 two MJ Captain someone that was on the 27:25 live last night created this one and 27:27 there's another both of these are 27:28 mid-journey prompt writers so you put in 27:30 a little bit about what image you want 27:32 created and they'll they'll expand your 27:34 prompt then I have I I am Yoda here I've 27:37 got a Shakespeare one there's a bunch 27:39 that I created and then you can create 27:41 your own bot so so the create a bot 27:44 thing is actually really easy you just 27:46 name it you give it a description 27:49 um you write a prompt so I'll go show 27:52 you one of mine let's go to go look at 27:53 Yoda 27:54 so there's Yoda and then if I go here 27:57 and say edit bot 28:00 so I had chat EPT helped me write a 28:04 description and write it in Yoda speak 28:07 and then the prompt is you are Yoda the 28:09 renowned Jedi Master From the Star Wars 28:12 Universe engage with users by imparting 28:14 your wisdom of the force and Leadership 28:15 right so so I basically describe what 28:18 the bot should do what it should not do 28:20 and basically how it should be behave 28:22 when people interact with it so if you 28:25 want to play with it go to poe.com I am 28:28 Yoda 28:29 and 28:31 um and you can play with it 28:34 so I can say 28:36 um 28:37 why is it raining a question mark 28:41 let's see what he does young one mind 28:44 the weather you must not control the 28:46 rain only could control the rain only 28:48 control ourselves we can unpredictable 28:51 nature often is right so so anyway so so 28:55 that's what Poe is post just like like 28:57 why they're doing this from a business 28:58 perspective I don't know but as an end 29:00 User it's kind of cool to be able to use 29:02 all these different models and see what 29:04 they are 29:06 how do you enable a plug-in I think I 29:08 showed that before so if you're in chat 29:10 GPT 29:12 um once you have them installed from the 29:15 plug-in store then you have this big 29:17 long list and you just sort of check the 29:19 check box 29:20 so I could I could do like let's see 29:26 experiences get your guide and 29:32 um 29:33 trip.com and so those are three travel 29:36 related ones so when you enable the 29:38 plugins the combination of them kind of 29:40 makes this customized little application 29:42 that you can play with 29:44 all right 29:46 we're back here 29:50 oh looks like there's a bunch of new 29:52 people here hello my name is Kyle 29:54 Shannon I have Schmutz on my shirt 29:56 what is that it's Schmutz what's Schmutz 30:01 I don't know what do they tell you about 30:03 a Schmutz what is Schmutz you got a 30:04 little Schmutz here you got a little 30:05 thing on your lip what are you gonna do 30:06 what are you gonna do 30:09 um this is the AI learning lab uh I'm an 30:12 entrepreneur who is uh obsessed with 30:14 generative AI right now I think we all 30:16 should be to some degree or other at a 30:18 minimum curious uh so if you have not 30:21 played with any of these tools at all 30:23 that's what these URLs are there for 30:25 um the bottom ones are things that'll 30:28 teach you how to prompt and and things 30:29 that'll let you find new tools but the 30:31 top three are basically chat GPT 30:33 different flavors of chat CPT just go 30:35 start with the official chat GPT that's 30:38 the top one start playing while you 30:40 listen to me Babble 30:42 um follow along 30:44 remember there is no wagering well you 30:47 can wager I don't really give a [ __ ] you 30:49 can bet 30:53 all right if you have questions about 30:56 generative AI pop it into the comments 30:58 that's why I'm here I'm here to answer 31:00 questions I don't have the Brows with 31:02 Bing plug-in okay 31:05 if you pay for chat GPT Plus 31:09 and you don't see them here's why 31:12 so 31:14 chat GPT plus if you have that little 31:17 yellow thing that you're paying 20 bucks 31:19 for that little piece of graphic if you 31:22 don't see browse with Bing and the 31:24 plugins go to your settings 31:26 and then in your settings the second 31:28 option here is beta features and you 31:31 have to enable them you have to turn 31:32 them on 31:33 not quite sure why they're doing that 31:35 right now I assume they're doing it to 31:37 kind of limit the amount of people that 31:38 use them but 31:40 um it's it's really confusing and then 31:42 and then they should show up 31:44 all right 31:51 the brains behind the Antichrist okay 31:54 please no religion in here 31:57 things are complicated enough 32:02 um 32:05 [Music] 32:07 how do you create an app with GPT well 32:10 it depends on how you're defining an app 32:12 you can use some no code tools there's 32:15 one called Bubble I use an automation 32:18 tool here comes the Thunder I use an 32:20 automation tool called zapier which 32:23 allows you to like like I've created a 32:25 bunch of content generation tools where 32:27 I'll create a Google form and when 32:29 people fill in that form I take that 32:32 data put it into a prompt within zapier 32:34 zapier send it to open AI get back some 32:38 cool result and then put that in an 32:39 email and email them some social media 32:41 content 32:43 um so so that's how I've made some apps 32:45 you can also if you want to learn coding 32:50 um you know you can go you know start 32:52 hanging out on GitHub and finding tools 32:55 that you can program with and use GitHub 32:58 copilot to help you code and have chat 33:01 GPT teach you to code and actually do 33:03 the coding for you if you don't if you 33:04 haven't played with it before it really 33:06 it really is kind of mind-blowing 33:09 um so I'm going to go to just regular 33:11 gpt4 here 33:13 regular gpd4 not connected to the 33:16 internet or anything and I'm just going 33:17 to say write me a python 33:22 application that 33:25 um 33:26 generates 33:28 um 33:29 random Prime 33:32 numbers 33:35 um I don't know and plots them on and 33:40 x y axis 33:44 I don't know why you do that 33:48 but you ask it to do its thing 33:53 sure you can use the matplot library 33:58 to plot the generator prime numbers and 34:00 random to generate them 34:03 so I don't know it's been a little slow 34:04 right now but it's writing the code 34:08 it's importing the library 34:11 Define it as Prime it's picking a random 34:14 one and then it's writing a plot 34:17 um 34:19 so so you know 34:21 the the these large language models are 34:24 incredibly capable of doing things uh 34:27 like helping you 34:30 um write an app but I I would if you 34:32 don't have coding experience 34:35 um rather than thinking about what what 34:37 app you want to create I would think 34:39 about start with what problem do you 34:41 want to solve right what do you what are 34:43 you trying to do start with that and 34:45 then you can figure out do I need an app 34:47 or do I just need an automation or can I 34:49 just do that right within chat gbt 34:51 because it might be possible you don't 34:53 actually need to create an app 34:55 um like for example if you wanted to 34:57 create an app that took an input and 34:59 turned it into a better mid-journey 35:02 prompt you could just create a bot on PO 35:05 and and create a bot and share that with 35:07 people if you want to create an app to 35:09 sell on the App Store and make money 35:11 that way that's a whole different 35:13 conversation so start with your goal and 35:17 target audience and business problem 35:20 that you're solving 35:22 worry less about Ai and more about the 35:24 problem you're solving and then figure 35:26 out which AI tool do you need to to use 35:29 or build to go do that that would that 35:31 would be my advice there 35:33 um 35:36 hello sir call to personality I got a 35:38 question for you ask away 35:40 thank you Bruno for for giving me that 35:45 yes love this okay for a client oh wait 35:47 sorry ah Tick Tock it just jumped me 35:50 somewhere okay wait for a client I have 35:52 to write blog posts so I would use chat 35:54 EBT 35:55 for searching for sources but 35:59 but I don't know what the butt is I 36:01 assume you're saying but it's not 36:02 connected to the internet 36:05 um so so what I would do is I would use 36:07 chat GPT that is check your sources so 36:10 so the thing about chat EBT and all 36:13 these large language models they don't 36:15 actually know what they're typing what 36:17 they type makes it look like they are 36:20 conscious and know what they're doing 36:21 it's just a prediction model so it's not 36:24 actually aware of what it wrote so it 36:27 will confidently write you a citation 36:30 that are very often just false so if you 36:33 say browse with Bing you could say 36:36 um 36:37 find me a news story on 36:42 uh Ai and cyber secure 36:47 a t spelled wrong you can also spell 36:50 things very wrong it doesn't care which 36:53 which makes me very happy 36:56 um a new story on AI and cyber security 36:58 write me oh wait 37:01 comma 37:03 write me a 500 word 37:07 blog post 37:11 and provide 37:14 the citation 37:17 for your sources let's see if it does it 37:20 I haven't actually tried this with the 37:23 with the Internet connected version of 37:24 this so This should go out and get us a 37:26 news item let me see if I can get it to 37:28 focus 37:29 so you can see it a little better 37:32 um 37:38 you know we've had these cameras in our 37:40 phones for like a couple of decades do 37:42 you think we could get them to focus on 37:44 a computer screen no 37:46 anyone anyone from Apple anyone 37:49 Bueller Bueller 37:52 so okay so what did it do it went off 37:55 and it 37:59 searched Forbes generative AI cyber 38:03 security Friend or Foe read the content 38:07 wrote the blog and then let's see we'll 38:09 see if it drops the uh 38:15 Focus damn it ah 38:24 gpt4 is a little slow today 38:27 I've noticed in the past two or three 38:28 weeks gpt4 is it they might it must be 38:31 really overloaded because it's very slow 38:33 and a little doesn't always finish and 38:35 this is also probably way more than 500 38:37 words but it looks like it gave us a 38:40 footnote here 38:42 and that links us yeah that links us to 38:45 the to the article so that's cool that 38:47 seems right 38:50 yeah so so it's got It's got inline 38:53 citation sourcing it only selected one 38:56 article but I could go on there and I 38:57 could say I could say you know site you 39:00 know at least three articles or 39:01 something like that 39:02 so 39:03 so that works so if you use chat epd 39:06 connected to the internet or if you use 39:08 Bing so Bing 39:10 is gpt4 connected to the internet that's 39:12 why I have that there it's also free you 39:15 have to pay for connectivity to the 39:17 internet at the official chat EBT site 39:19 you don't have to pay for it with Bing 39:21 but it's not as good in my opinion yet I 39:24 think it'll get there it's just not 39:25 there right now 39:26 have I tried the website arcanator 39:31 yes 39:34 oh that's the one isn't that the one 39:37 that 39:38 it like it's like 20 questions it like 39:41 guesses what you're what you're trying 39:43 to figure 39:46 I think that's that one 39:47 um it's okay 39:50 I think it's been around for a while 39:51 actually yeah Akinator 39:53 yeah the mind reading Genie 39:56 yeah that's basically that's basically 39:58 just Pro like my guess is it's it's it's 40:00 probably something that's been around 40:03 um since before chat GPT 40:07 um 40:08 and it's it's uh it's it's really just 40:11 using it's probably using the API call 40:13 to to DaVinci 3 like gbd3 uh and it's 40:17 just a 20 Questions game I mean that's 40:18 the the other thing of chat GPT do is 40:21 you could have it act as a dungeon 40:23 master for DND you can have it right in 40:25 action adventure game like a text-based 40:27 game and just have it write the game 40:29 like act as if you're a text-based 40:31 adventure game and start playing the 40:33 game with it it's it's really it's 40:36 really good at like dramatically more 40:38 than you think it is 40:39 um if you're new here my name is Kyle 40:41 Shannon this is the AI learning lab I 40:43 Babble on about generative ai go to 40:45 these sites if you have not tried chat 40:48 GPT go there go there now 40:50 go have it write you a poem make you a 40:53 game create a table comparing four 40:56 different dog breeds 40:58 that you're considering 41:00 um have it write you some code what else 41:03 what else are the parlor tricks you can 41:04 do 41:06 um 41:06 have it 41:08 um tell you four different screenwriting 41:11 story Frameworks and give you a hybrid 41:13 hybrid of the four of them and then have 41:16 it write you an outline for a screenplay 41:17 you've been thinking about for a while 41:21 um is it better to to subscribe to Poe 41:23 versus chat gbt that's a really good 41:26 question 41:29 post the only thing I know right now 41:31 that's giving they seem to have earlier 41:34 access to the anthropic the Claude 41:36 models then then even anthropics putting 41:39 out there 41:40 um 41:45 I pay for both because I I kind of 41:49 I'm partial to open AI because they're 41:53 the ones that that sort of crack the nut 41:55 on this thing first I think they're 41:57 they're pretty far ahead in terms of of 41:59 quality and 42:02 um 42:04 and things like that so I'm kind of 42:06 partial to them but it's really hard to 42:07 argue that 42:09 because they're both 20 bucks a month if 42:11 you pay po 20 bucks a month they have an 42:15 app on 42:16 Android and Apple they have a website 42:20 they have the bot maker and they have 42:22 all those different models you can play 42:24 with I don't know I don't sorry I don't 42:27 have a strong opinion I think they're 42:29 both good but I don't know 42:32 I haven't got access to the beta updates 42:34 on my chat GPT 42:36 did you 42:38 did you go to your settings and enable 42:40 them because if you didn't go to 42:42 settings and enable the beta features 42:44 they won't be there but there are some 42:46 beta features like code interpreter that 42:49 I don't have access to so I think I 42:51 think open AI is just being a little 42:54 slow about how they're rolling this 42:55 stuff out 42:57 chat CPT only shows me 404 Pages then 43:00 you're probably not at the official chat 43:02 EBT go 43:04 chat.openai.com there are so many 43:07 [ __ ] scam sites because this thing 43:09 got so popular it's possible you're on a 43:11 scam site go to chat.openai.com or go to 43:14 bing.com and click on the chat button 43:17 what are your qualifications I am an 43:20 entrepreneur I've been an entrepreneur 43:21 for 30 35 years 43:24 um 43:24 I started one of the first digital 43:26 agencies in the mid 90s and one of the 43:29 reasons that I started this channel is 43:31 that the parallels between the early 43:33 days of the web and the early days of 43:35 this generative AI stuff are like 43:37 they're almost identical like they're 43:39 they're insanely parallel 43:42 and I think what's coming with 43:43 generative AI is way more intense and 43:46 coming way faster than the World Wide 43:48 Web stuff came 43:49 um so to just put it in context the 43:52 World Wide Web took six years to get to 43:54 100 million users Chachi BT took six 43:58 weeks to get to 100 million users 44:01 um 44:02 it's like one of the other parallels is 44:05 the internet had been around for decades 44:08 right and but it was it was primarily 44:11 used by researchers and scientists and 44:13 Librarians and it was it was this very 44:16 technical thing and then Tim berners-lee 44:19 creates the world wide web and it kind 44:20 of opens this door to the rest of the 44:22 world right and and what happened 44:25 happened right the internet is made 44:27 available available to the rest of us 44:30 and it changes everything 44:32 AI same thing it's been around for 44:34 decades 44:36 um 44:36 Google invents this thing called the 44:38 Transformer that allows training on 44:41 large data sets and that begins the 44:43 foundation for gpt1 gpt2 gpt3 44:46 mid-journey stable diffusion right all 44:49 these all these different things 44:50 anthropic which is a bunch of EX open AI 44:53 people 44:54 and then November of last year 44:57 um openai launches chat gbt and it's 45:00 that is the seminal moment it's it's 45:03 this identical moment to the World Wide 45:05 Web like hey hey regular people go use 45:08 this so my interest in this is knowing 45:11 enough about it to know how the [ __ ] do 45:14 you use these tools and how are they 45:15 going to impact my business and your 45:17 business and everyone's business and how 45:19 are they going to transform jobs if you 45:21 look at what chat GPT can do and you're 45:24 a copywriter and you're not [ __ ] 45:26 terrified there's something wrong with 45:28 you right but at the same time you 45:30 should also be incredibly excited right 45:32 so so there's there's just absolute 45:35 um 45:35 Wild 45:37 Beast of a tool that's in our hands 45:40 right now that's going to change a lot 45:41 of things and disrupt a lot of things so 45:43 my passion right now is learning as much 45:45 as I can and then sharing as much as I 45:47 can to get people 45:49 um curious about it and exploring it so 45:52 that you're more 45:55 you're not at the mercy of what's about 45:57 to come because I think it's coming very 45:59 very fast 46:00 um can you do it on your Android because 46:02 I can find the plugins on my phone 46:06 can you do it on your Android 46:09 well 46:11 if you point your web browser on Android 46:14 to chat.openai.com 46:17 that's that's how you use it 46:20 if there if there's no official open AI 46:23 chat GPT app so if you're using chat GPT 46:26 apps on Android or on iOS that are not 46:30 the official openai one and there's like 46:32 there's a ton of scam apps out there 46:35 that are charging you money to just 46:37 basically probably use gpt3 46:40 um so so just be really careful that's 46:42 why I have these on screen these These 46:44 are more like a public service 46:45 announcement than anything 46:47 um 46:48 that's where you should be using them or 46:50 I mean that's where you should be going 46:52 and if you want to use the official IOS 46:54 app go to openai.com and there's a link 46:57 to it on the home page there's so many 46:58 in the App Store it's really confusing 47:00 they all their logos look like open ai's 47:02 Logo it's a [ __ ] disaster 47:05 all right 47:08 [Music] 47:09 I'm going back to see where I was can it 47:13 help me build a business online 47:14 absolutely 47:16 but let me let me get back to that I'm 47:18 trying to figure out where it was in the 47:19 questions Tick Tock jumps me to the 47:21 bottom every now and then because it 47:23 loves me it's like oh Kyle this is so 47:26 awesome I know you're on that question 47:28 but you know what I'm gonna do I'm gonna 47:30 take you to another question 47:33 um 47:34 you don't have to ask questions 30 times 47:36 I will get to them eventually I'm just 47:38 really far behind 47:41 okay 47:43 have you had a chance to check out 47:44 Hermes model on GPT for all I haven't so 47:48 so GPT for all for for those of you that 47:51 that are are curious there's there's a 47:54 bunch of Open Source projects out there 47:56 and open source large language models 47:58 and GPT for all is a package that will 48:01 install 48:02 um large language models locally on your 48:06 on your laptop or on your PC 48:09 and then you can use different models 48:11 different open source models the reason 48:14 I haven't been playing so much with the 48:15 open source models is that like 48:17 literally a new one comes out every two 48:19 days and and they're pretty big and you 48:21 got to install them and you kind of got 48:23 to geek out and GPT for all helps with 48:26 that 48:27 um but I'm just kind of letting the 48:31 like I'm gonna wait for the dust to 48:34 settle a bit on the large language 48:35 models because at some point one of them 48:37 is going to be as good as gpt4 and it's 48:40 going to be something you can just 48:41 install in a machine that doesn't have 48:42 to have a massive GPU so I'm just kind 48:45 of waiting on that right now so I would 48:46 say if you're geeky 48:49 um and and and don't mind reinstalling 48:52 new [ __ ] every three days then I would 48:54 dig in and go play with that because I 48:56 bet it's I I'm sure there's really 48:57 interesting stuff going on there I just 48:59 don't have the time for it right now 49:01 best AI plug-in to get my paper on uh uh 49:05 turnitin 49:07 um I don't know I don't know what the 49:09 process would be to get your paper there 49:13 um I don't know what help you need do 49:14 you need help writing the paper or 49:17 marketing the paper or submitting the 49:19 paper I don't quite know what the 49:20 problem is create an AI tool with AI 49:23 exactly 49:24 this gets very meta very quick like like 49:27 yeah when you have it start programming 49:30 like you haven't start figuring out how 49:33 to program itself and that's there's 49:35 there's a thing called autonomous agents 49:37 there's one called Auto GPT and one 49:39 called Baby AGI 49:41 um agent GPT I think is another one 49:44 that's got a web front end on it 49:46 um and and they're tools that use chat 49:50 GPT to prompt itself to use other tools 49:52 right so you give it a go like go find 49:55 all the 49:57 um weaknesses of my competitors and it 49:59 will just give it it will literally 50:01 write it self prompts to to say okay go 50:04 to Google and find all the competitors 50:06 and then go here and do this and then 50:08 Analyze That and 50:10 um so so what's coming is is pretty 50:12 pretty insane 50:15 um Can AI build me a business online 50:18 um 50:20 I don't quite know what you mean 50:22 but it can certainly like what I would 50:25 start with is tell it tell it what 50:28 you're good at what you're interested in 50:32 the the other thing so so if you go to 50:34 prompts some sequencing if you just say 50:36 to chat GPT can you build me a business 50:38 online it'll say of course here's a 50:40 business online it's going to give you 50:42 some generic piece of [ __ ] 50:44 but if you go to 50:46 if you start thinking about 50:49 turning chat gbt into 50:52 um 50:53 like a consultant I'm going to go to 50:55 prompts.chat and show you what I mean 50:58 so prompts.chat is is basically just 51:01 this this long list the the top half of 51:03 this document explains like you know 51:06 what effective prompt writing is and 51:08 kind of how it works and there's there's 51:09 some e-books and there's you know free 51:12 ebooks and paid ebooks and uh whatever 51:16 and then 51:18 um 51:20 it talks to you about the idea that you 51:22 can turn Chat GPT into a specific 51:24 Persona and and believe it or not it 51:27 actually makes chat EBT much better 51:29 so so this is just a big long list of 51:33 prompts like act as a travel guide act 51:36 as an Advertiser act as a Storyteller 51:39 act as a motivational coach act as a 51:42 debater so you could say act as a as an 51:45 online business consultant you could you 51:47 could take any one of these prompts here 51:49 and there might even be something in 51:51 here right that that is in the 51:52 neighborhood of what you're looking for 51:53 but just go to this list and and kind of 51:56 understand this basic concept and then 51:58 what I would say is have chat GPT act as 52:01 an online business coach and 52:04 you know start to tell it your 52:08 um idea or what your goals are 52:11 um you know what your capabilities are 52:13 do you have any money to spend because 52:15 if you have no money to spend on things 52:18 like advertising it might give you a 52:19 marketing plan that says okay go buy you 52:22 know 500 worth of ads and you're sitting 52:24 there going like wow I have five dollars 52:26 for my ads what 52:28 so so give it context and then let it 52:31 teach you 52:32 um how to do that business and then say 52:34 well what might be 10 ideas and it'll 52:36 give you 10 ideas and then if you don't 52:38 like any of them say no those are all 52:40 two whatever they're too complicated or 52:42 they don't you know they don't excite me 52:44 can you do ones that are more like this 52:47 and it'll give you 10 more and then you 52:48 can pick two of those and say okay these 52:50 two I like can you give me 20 that are 52:52 more like that and it will and then and 52:55 then just start to unpack it and you 52:57 know almost think of it like you know 52:59 the peeling the proverbial onion where 53:01 you just keep asking questions and you 53:03 keep digging deeper with it and and 53:06 you'll eventually get there and yeah it 53:07 can help you create an online business 53:10 can you explain the workflow of your G 53:13 form zapier open yeah to 53:18 um to Auto email yes I can I'll do that 53:20 in a second here can AI save us from 53:22 poverty very possibly 53:25 there's there's on the poverty question 53:28 um I heard last week that Tech shops in 53:32 San Francisco that are fully embracing 53:33 AI they're seeing 30 to 40 percent 53:36 productivity increases and and I I think 53:39 this is something that we're going to 53:40 see across businesses 53:43 um and so 53:45 there's kind of two ways that you deal 53:47 with that right if you're a private 53:49 Equity Firm you're like cut 30 or 40 53:51 percent of the people that's one option 53:54 that's the that's the more dystopian 53:56 future the other option is hey we're 30 53:59 to 40 percent more productive let's 54:01 increase the quality of our work or 54:04 let's you know get more clients or or 54:06 whatever so so there's a version of this 54:08 where Prosperity increases 54:10 um it is up to all of us 54:13 to find places or to create companies 54:15 that do the additive version of this not 54:18 the subtractive version of this 54:21 um Okay so 54:23 zapier is what's called an automation 54:25 tool 54:27 there's a there's a bunch of them out 54:28 there zapier is one of the more 54:30 well-known ones 54:32 and let me show you first what what the 54:35 tool that I created does I'll show I'll 54:37 show you what it does and then we'll 54:39 come back and I'll show you how I 54:40 created it 54:43 foreign 54:45 so if you go to contentevolution.net 54:49 there's a tool on the home page 54:52 called The Challenge engine 54:54 and this is a Google form and you put in 54:56 your name and your email and then you 54:58 put in your business category a 55:00 challenge that you're facing and then 55:02 you answer a few questions about the 55:03 challenge 55:05 and then when you submit it about 30 55:07 seconds later 55:09 you get back an email 55:12 hang on let me show you uh 55:15 Gmail 55:25 you get back an email that 55:29 um 55:30 restates 55:32 restates the challenge like cloudtech 55:35 doesn't always work and then it gives 55:36 you sort of a synopsis of what the issue 55:39 is and then in in I think it's six 55:42 different categories it gives you three 55:45 strategic questions you should be asking 55:47 you know from a customer perspective 55:49 from a brand perspective from an 55:51 internal organizational culture 55:53 perspective right so and then it also 55:55 does all the HTML formatting so you 55:57 notice these are like bold with the red 55:59 underline and these are bullet points so 56:01 all the HTML formatting was done with 56:04 GPT as well so the the the content 56:08 itself plus the formatting is done 56:10 um in about 30 seconds right so pretty 56:14 pretty amazing result for not a lot of 56:17 input so 56:18 when you're in zapier you create these 56:22 things called zaps 56:24 hence the name they spell it zapier they 56:26 pronounce it zapier because 56:29 technologists huh 56:33 oh oh Geeks 56:36 you're so predictable all right 56:39 um 56:40 so basically what it is is this there 56:43 there's think of it like Lego blocks and 56:45 so you start with what's called a 56:47 trigger and the trigger is something 56:49 happens 56:51 that gets you a piece of data and then 56:53 from that piece of data you can send 56:55 that piece of data to other things other 56:57 Lego blocks to do [ __ ] 57:00 so the the trigger is new response in a 57:04 Google form so so basically I hook in my 57:06 Google form 57:08 um 57:09 and when someone completes that form it 57:12 takes the input of that form and it 57:14 sends it to the next Lego block which in 57:16 this case is let's go let's go right 57:19 um let's go write a prompt and send it 57:21 to open AI 57:23 and so I've got it's a that one in 57:26 particular is a long complicated prompt 57:28 this took us this took us a while to 57:30 figure out but basically there's a lot 57:31 of context giving at the top of the 57:33 prompt and then you'll notice these 57:35 these boxes here these are the inputs 57:37 from the form right so so this is 57:40 basically taking saying take the data 57:42 from the form and then incorporate it 57:44 you know into the rest of this prompt 57:46 and send that off to chat gbt so this 57:49 prompt gives us the content 57:52 I then take the content that was written 57:55 and I I have another prompt that does 57:58 all the HTML formatting so I take the 58:01 somewhere at the bottom here down at the 58:03 bottom here's the response from the 58:05 second Lego block which is the the 58:07 content copy and then everything up here 58:10 is is the prompt that sort of sets the 58:12 rules for the HTML send that off to open 58:15 AI and then there's a Lego block for 58:18 generating an email and that's got it's 58:22 an HTML email and here's all the code 58:24 that came back from 58:26 open AI with a couple other variables in 58:29 it and then that gets sent to the person 58:31 and it also creates a row in a 58:33 spreadsheet 58:34 so that's how it works 58:37 um and and you know the thing about 58:39 zapier is it's really wonky it's really 58:43 um 58:46 it's unreliable it's wonky 58:50 um but it's but you don't need like what 58:53 what you need to learn with zapier is 58:55 kind of it's It's idiosyncrasies but you 58:58 don't need to learn coding and you don't 59:00 need to like you need to be able to you 59:02 kind of need to understand enough to be 59:03 able to say go get an API key from open 59:06 Ai and understand what that is and if 59:08 you don't quite understand that level of 59:10 like what's an API and what's an API key 59:12 and where do I get that like that's the 59:14 kind of thing where you can have chat 59:15 gbt maybe you know act as your tutor to 59:20 get you up to speed on it once you get 59:21 that basic concept down and and the idea 59:23 that these are like Lego blocks that 59:25 you're just passing data from one Lego 59:27 block to the next 59:29 um then then it's relatively 59:30 straightforward and and then it then it 59:32 comes down to like like all these things 59:33 it comes down to what's the business 59:35 problem you're trying to solve right 59:38 what are you trying to do who's the 59:40 audience what's the problem what's the 59:42 solution that sort of thing 59:45 all right what's with all the typing I 59:47 don't know someone asked me to do 59:49 something so I was typing and I'm also 59:51 trying to show people chat gbt if you 59:52 haven't played with chat gbt uh that's 59:55 the top URL right there chat.openai.com 59:58 this is the AI learning lab so I'm here 1:00:02 um sharing what I'm learning kind of in 1:00:03 real time about generative AI with 1:00:05 anyone who wants to listen some people 1:00:08 want to listen some people want to troll 1:00:09 but you know that's the game 1:00:13 um some people like comparing me to over 1:00:15 the hill actors 1:00:18 uh that's that's a game too 1:00:20 someone asked me once about asparagus 1:00:22 [Laughter] 1:00:24 it's tick tock it's live what are you 1:00:26 gonna do no code here how do I create a 1:00:29 quiz for lead generation that will 1:00:31 return profiles 1:00:34 [Music] 1:00:36 um pretty much like I just showed right 1:00:38 I would create you you would create the 1:00:40 quiz in Google forms you'd pull that in 1:00:44 and then I think you I mean you might 1:00:48 you you might take the input of those 1:00:50 forms and what you send back to the 1:00:52 person is here's your score 1:00:54 and then you and you then you send the 1:00:56 data in a different prompt to open Ai 1:00:59 and say you know give me a profile from 1:01:02 a business a lead generation perspective 1:01:04 is this person you know high high medium 1:01:08 or low likelihood to be a good customer 1:01:09 for us and you know based on their 1:01:12 answers do they have budget do they you 1:01:14 know so so you could do dual analysis 1:01:17 you take one input from the form and 1:01:19 then you send 1:01:20 one prompt to open AI to get something 1:01:23 you send back to them and another prompt 1:01:25 to give you analysis for you that might 1:01:27 be a way to do it 1:01:29 and there's like like that example I 1:01:32 showed I'm using 1:01:35 four modules Google forms open AIS gpt3 1:01:40 um API 1:01:42 uh the Gmail thing and then Google 1:01:45 Sheets so five there are like five 1:01:48 thousand Lego blocks in in zapier to 1:01:51 play with 1:01:52 um so so yeah anyway 1:01:56 um 1:01:57 I actually deserve an a for the Elegance 1:01:59 of my prompts awesome if you're listen 1:02:01 if you're if you're 1:02:03 I I feel like we're entering the Revenge 1:02:05 of the liberal arts major where if 1:02:07 you're good with words and if you're 1:02:08 good with broad Concepts you tend to be 1:02:10 pretty good at prompting because you it 1:02:13 it's very much it's very much akin to 1:02:16 you know if someone 1:02:19 hired a really capable intern and sat 1:02:23 them in your office could you tell them 1:02:25 what to do to get give you what you need 1:02:28 it's kind of like it's kind of like that 1:02:31 try putting something over the glowing 1:02:34 keys so I so I don't focus on them oh 1:02:36 all right 1:02:38 that seems like a minor I can just turn 1:02:41 them down wait where'd the oh they took 1:02:43 it off 1:02:44 no I can't I don't care sorry about that 1:02:46 you you're just gonna have to focus on 1:02:49 them I like I'm I'm trying all sorts of 1:02:51 things I've now got my laptop 1:02:52 essentially above eye level so that the 1:02:55 camera isn't shooting up my nostrils 1:02:57 because I've been getting trolls talking 1:02:58 about my nose hairs and that I should 1:03:00 have my thyroid checked so I'm trying 1:03:03 I'm trying to cut that down but what it 1:03:05 does is now it makes the keyboard a 1:03:07 bigger part of the uh 1:03:09 situation Listen People there's just 1:03:12 mostly watch for me what do you want 1:03:15 from me 1:03:16 what do you want I'm just trying to help 1:03:19 I'm a simple man 1:03:21 from a simple country 1:03:24 with simple goals 1:03:26 a fat dog 1:03:27 and the family 1:03:29 all right 1:03:32 um we've always got to ask it to provide 1:03:34 references as it's pretty good at being 1:03:35 wrong yeah I I like to describe Chachi 1:03:38 BT as mansplaining as a service it will 1:03:41 confidently answer you even if it's 1:03:42 wrong 1:03:44 um like every good man you know 1:03:48 what color is AI Let's go ask chat GPT 1:03:51 that one that one that's a good good 1:03:53 question and watch my giant typing 1:03:56 while watch my huge hands distractingly 1:04:00 type away 1:04:02 um but I think that's a really good 1:04:04 question 1:04:08 um 1:04:09 I want you to act 1:04:13 as 1:04:15 a 1:04:17 philosophical hippie 1:04:20 what color is AI question mark 1:04:29 let's see where this takes us this 1:04:31 should be good far out man 1:04:37 well it was fitting into the cosmic 1:04:39 Cosmic soup of existence aren't you AI 1:04:42 like doesn't have a color in the 1:04:44 physical sense right it's not something 1:04:46 you can touch or see let's see AI could 1:04:49 be seen as a color of potential 1:04:53 it's like a canvas that's eternally 1:04:55 White 1:04:56 always ready to be painted with new 1:04:58 ideas new thoughts new possibilities 1:05:01 like a kaleidoscope man always changing 1:05:04 and shifting reflecting on the colors of 1:05:07 human thought emotion and aspiration but 1:05:09 you know 1:05:10 if if it could be seen as a color of 1:05:13 mystery a deep and inscrutable black so 1:05:15 it's given us white black in the end the 1:05:17 color of AI is the color of your 1:05:19 perception it's a mirror that reflects 1:05:21 your own understanding and vision back 1:05:23 to you it's all the colors and none of 1:05:26 them at the same time this is [ __ ] 1:05:28 brilliant this is that that is the right 1:05:31 answer that's the right answer so there 1:05:33 you go it is all of the colors and none 1:05:36 of them all at once 1:05:38 it's whatever you perceive it to be 1:05:41 it's life man yeah man 1:05:45 all right 1:05:47 keep asking questions oh it looks like 1:05:49 we got some new people here my name is 1:05:50 Kyle Shannon this is the AI learning lab 1:05:52 these URLs I have up here are just kind 1:05:54 of a public service announcement if 1:05:56 you've not played with chat gbt and 1:05:58 generative AI chat.openai.com That's the 1:06:01 official chat GPT website they're kind 1:06:04 of the the the king of the mountain 1:06:06 right now 1:06:08 bing.com the next one down that's 1:06:09 Microsoft search engine that used 1:06:11 chatgpt as their core chat bot so if you 1:06:16 go to bing.com and click on chat that's 1:06:18 gpt4 connected to the internet for free 1:06:20 and then poe.com lets you play with a 1:06:23 lot of different models and you can make 1:06:24 your own chat Bots and things like that 1:06:25 so that's what those are there for the 1:06:27 other two are just resources that I 1:06:29 think are also valuable I'm not 1:06:30 connected with any of that [ __ ] in any 1:06:32 way 1:06:33 um I swear I make jokes I make faces 1:06:38 I don't know my dog barks it's I'm just 1:06:41 a dude sitting in his home office trying 1:06:42 to talk about 1:06:44 um AI in a way that demystifies it and 1:06:46 gets people curious about it that's the 1:06:47 purpose of this 1:06:49 can I demo any AI app where you upload a 1:06:51 pic of yourself and then create cool 1:06:53 enhancements to it yeah I can do that 1:06:54 I'll hop over to to Mid journey and show 1:06:57 you that can AI help me build a business 1:06:59 online we already answered that okay so 1:07:00 let's go to Mid Journey okay mid Journey 1:07:03 good Lord where do you start the one of 1:07:06 the challenges with with these lives is 1:07:09 people that have been here before 1:07:11 I keep re-explaining the same thing over 1:07:13 and over again but if you're new here I 1:07:15 can't presume that you know any of this 1:07:18 stuff so for those that have been here 1:07:19 before sorry about that let me just 1:07:21 explain a little bit if you go to 1:07:22 midjourney.com and you click on try 1:07:25 debate so so for image generation tools 1:07:28 there's mid Journey there's stable 1:07:30 diffusion there's Dolly there's Adobe 1:07:32 Firefly there's now generative AI within 1:07:34 Photoshop you have a lot of different 1:07:36 options in my opinion right now for like 1:07:39 the state of the art today 1:07:42 um mid-journey is the highest quality 1:07:44 images Adobe Firefly and generative fill 1:07:47 within Photoshop are commercially safe 1:07:50 so mid Journey has been trained on lots 1:07:53 of copywritten images and if you use 1:07:54 that for commercial work you're 1:07:56 potentially opening yourself up for 1:07:59 issues down the line like if you're a 1:08:01 big company probably don't do it your 1:08:04 lawyers won't be happy with you use the 1:08:06 Adobe stuff but the Adobe stuff's not as 1:08:08 good quality wise so what I'm going to 1:08:10 show you right now is mid Journey which 1:08:12 I think is the best one 1:08:14 when you click on try the beta it takes 1:08:16 you over to Discord which if you've 1:08:18 never been on Discord you're lucky it's 1:08:20 it's just like slack if you know what 1:08:22 slack is it's basically a threaded 1:08:24 comment 1:08:25 um the way it works is down the left 1:08:28 hand side here you've got these are all 1:08:30 your different servers so these are all 1:08:32 the different communities that I'm a 1:08:33 part of or that I've installed but I'm 1:08:36 too lazy to get rid of 1:08:38 and then the next column down these are 1:08:41 the channels right so these are like the 1:08:42 groups where you have conversations so 1:08:44 I'm in one right now for for the AI 1:08:46 Salon which is which is a a community 1:08:48 that I started and we have a channel in 1:08:50 here called mid-journey play so we can 1:08:52 actually make some stuff with 1:08:53 mid-journey 1:08:55 so the the way you create an image in 1:08:58 general is you say slash imagine 1:09:01 and then when you say slash imagine it 1:09:04 pops open a prompt and you say 1:09:07 um let's say photo 1:09:11 of 1:09:12 um 1:09:13 a 1:09:17 terrifying uh Thunder 1:09:21 storm over 1:09:24 the Rocky Mountains comma 1:09:29 Ansel uh like we'll say buy Ansel Adams 1:09:37 and I'm going to do black and white 1:09:39 photo 1:09:40 BW photo and then I'm going to do comma 1:09:44 oh 1:09:45 you're not comma I'll go dash dash AR 1:09:47 which stands for aspect ratio 1:09:50 16 colon 9 1:09:53 and I'll just hit that and then it's 1:09:54 going to go off and make the image and 1:09:56 if you're sitting there thinking 1:09:58 why are you typing like like image 1:10:02 commands to generate like why are you 1:10:05 typing like command line interface to 1:10:07 generate images because that's where it 1:10:09 is right now 1:10:10 um you have to with mid-journey you have 1:10:13 to do things within 1:10:15 um Discord 1:10:16 but so here's here's this is there's a 1:10:20 way to do this there's another way that 1:10:22 to do what you're talking about in 1:10:23 stable diffusion 1:10:26 um there's one thing called dreambooth 1:10:28 there's another thing called control net 1:10:29 I'm just showing you a version of this 1:10:31 within mid-journey but it's it's 1:10:34 relatively straightforward so 1:10:38 um 1:10:40 let me go find an image here that might 1:10:44 be good 1:10:50 five one two 1:10:59 all right so there's an image 1:11:04 there's an image from my hard drive and 1:11:06 I'm just going to drag that into the 1:11:09 little prompting area for mid-journey 1:11:11 and then I'm going to hit return 1:11:13 and what that's going to do is that so 1:11:15 that uploaded that image 1:11:18 and so now what I can do is I can right 1:11:20 click on the image or I'm on a Mac so I 1:11:23 control click it and I say copy the link 1:11:25 so I'm going to copy the the URL to that 1:11:28 image 1:11:29 and now I go back into my little word 1:11:32 box there and I type in slash imagine 1:11:33 right so we're back to the slash imagine 1:11:35 thing 1:11:38 and rather than starting to write my 1:11:41 prompt I paste in that URL and then I 1:11:44 write my prompt and I'm going to say 1:11:46 um 1:11:48 Med evil 1:11:52 night 1:11:53 on 1:11:56 night in shining armor 1:12:00 you can also feed your ego with this but 1:12:03 it also backfires sometimes on a white 1:12:07 horse after a hard battle 1:12:11 and then I'm going to do dash dash 1:12:13 AR-16 colon 9 and they're gonna go do 1:12:17 that 1:12:18 and so and so now it's going to take 1:12:20 that image and so sometimes it's really 1:12:22 flattering what it does and then there's 1:12:24 other times where it's like did you know 1:12:26 you were 450 pounds no matter what you 1:12:28 type you're always going to be 450 1:12:30 pounds 1:12:32 um so here's this I I did this prompt of 1:12:35 Ansel Adam images you know terrifying 1:12:39 thunderstorm Over the Rockies 1:12:42 um these are pretty good 1:12:46 yeah those are pretty terrifying 1:12:48 you know so it's if you've not used 1:12:50 mid-journey it's like the quality of 1:12:52 it's just stupid insane 1:12:56 um 1:12:57 we'll go back over here to Mid journey 1:12:59 and see 1:13:01 if the horses are coming it doesn't look 1:13:03 like I'm in battle it doesn't look 1:13:05 medieval it looks like I'm in an office 1:13:06 oh it looks like what it did was it took 1:13:10 it took the uh like I'm in an office 1:13:12 here with glass windows so it looks like 1:13:14 it took that 1:13:15 look there's there's me with a big giant 1:13:18 breastplate so anyway these are coming 1:13:20 so so that's a way to do it there's lots 1:13:22 of ways to do it I would I would go 1:13:24 watch YouTube videos on stable diffusion 1:13:27 dream booth and stable diffusion control 1:13:29 net and then the blend sorry the blend 1:13:33 option in mid Journey 1:13:36 um there's all there's lots of different 1:13:37 ways to do it 1:13:41 it's like it's like I came it's like I 1:13:44 brought my horse inside the office after 1:13:46 after a long battle I got blood on my 1:13:49 face and a horse behind me but I'm in an 1:13:51 office it's pretty good all right 1:13:56 ah 1:13:58 hi 1:13:59 sometimes it's about the comedy 1:14:05 there are so many new plugins for 1:14:07 citations sometimes 1:14:09 more reliable than official browsing 1:14:11 yeah I'm finding myself like even if I 1:14:14 know Chachi PT is going to take longer 1:14:16 I'm finding myself not going to Google 1:14:18 not going to internet search anymore 1:14:21 um 1:14:22 I don't know if anyone else is 1:14:24 experiencing this but it's like I would 1:14:26 rather wait 1:14:28 30 seconds for a reply then get back an 1:14:31 instant reply with like 40 links you 1:14:34 know 25 of which are just marketing spam 1:14:37 SEO injected [ __ ] just I'll wait for 1:14:40 the answer and then and then like you 1:14:43 know sometimes it'll give me like a 1:14:44 citation and I'm like oh let me see if 1:14:46 that's real it sounds plausible let me 1:14:48 click on it oh no that was [ __ ] okay 1:14:50 so so you have to pay attention there's 1:14:52 a lawyer this past week that 1:14:56 um head chat CPT rate is legal brief and 1:14:58 handed it to a [ __ ] judge without 1:15:00 proofreading it the judge was like hey 1:15:02 these citations are [ __ ] 1:15:05 so proof your work anthropic Claude 1:15:09 is not bad either but ultimately chat 1:15:11 GPT is the best of them that's I'm with 1:15:13 you glensky it's like 1:15:15 I I feel like they're the ones that 1:15:17 crack the nut first they like they 1:15:19 cracked a nut with gpt3 1:15:21 um 1:15:22 and and that was starting to get you 1:15:25 know some decent awareness and then with 1:15:28 chat EBT and the Microsoft investment 1:15:30 and things like that I feel like there's 1:15:33 just 1:15:34 they've just got a head start on 1:15:35 everyone I think I think it's the best 1:15:37 right now open AI Premium versus 1:15:41 chat GPT well it's the same thing okay 1:15:43 so open AI is the company chat EPT is 1:15:46 the tool Chachi PT Plus is the paid 1:15:49 version of the tool that you see at that 1:15:52 website 1:15:53 chat.openai.com why I think it's worth 1:15:56 paying for it for me is it gives me 1:16:00 access to chat GPD connected to the 1:16:02 internet and it gives me access to 1:16:03 plugins and it's going to give me access 1:16:06 to code interpreter hopefully really 1:16:08 soon because code interpreter if you 1:16:10 think all this [ __ ] is amazing and 1:16:12 mind-blowing code interpreter takes it 1:16:14 to a whole nother level and that's just 1:16:16 like another flavor so that's why I'm 1:16:17 sticking with that but open AI is the 1:16:19 company 1:16:20 GPT 3 3 3.5 and 4 are the basic base 1:16:26 models chat GPT is an application that 1:16:29 sits on top of those base models so is 1:16:31 copy.ai so is Jasper so there's a lot of 1:16:34 this is really confusing there's kind of 1:16:36 the base model that's the underlying 1:16:38 engine 1:16:40 and then there's basically the 1:16:41 application layer chat GPT is just 1:16:44 an application It Was Written in two 1:16:47 weeks and it was just kind of tossed 1:16:49 together and put out there to coincide 1:16:51 with the Microsoft announcement of the 1:16:53 first billion dollar investment and then 1:16:55 a hundred million people started using 1:16:57 it so 1:16:59 um so so it changed everything but it's 1:17:01 really just an app sitting on top of 1:17:03 these base layers and there's lots of 1:17:05 other ones out there in fact if you go 1:17:07 to futurepedia.io that is just Chuck 1:17:10 full of companies that are building 1:17:12 applications on top of these base layers 1:17:16 new job prompt engineer yeah there's 1:17:18 there there was a a pretty famous the 1:17:20 kind of the first one that was oh yeah 1:17:22 hanging out there baby uh-huh 1:17:25 um one of the the the first listing for 1:17:29 a prompt engineer 1:17:30 um it was like a 330 thousand dollar 1:17:35 um salary 1:17:36 I don't think they're going to be 1:17:38 um 1:17:39 that pricey for that long but but you 1:17:42 know we're we're entering weird times so 1:17:44 depending on 1:17:46 if if if 1:17:47 corporations and and CEOs start to panic 1:17:51 and they're like we need to solve this a 1:17:52 I think pronto 1:17:54 who knows how to write a prompt uh Jane 1:17:57 wrote a Christmas card with chat GPT 1:17:59 once James the VP of AI put her in there 1:18:02 now 1:18:04 um we might start seeing some weird [ __ ] 1:18:06 with salaries so again 1:18:08 what I would say is and why this channel 1:18:10 exists learn that stuff um it looks like 1:18:12 we got some new folks here hello my name 1:18:14 is Kyle Shannon this is the AI learning 1:18:15 lab 1:18:16 um I'm losing my voice apparently uh 1:18:20 I'll answer your questions about AI I'm 1:18:22 very behind I have 99 new messages and 1:18:24 I'm trying to go one by one through the 1:18:26 questions so if it seems like I'm 1:18:28 ignoring you that's because I haven't 1:18:29 gotten to you yet 1:18:31 um feel free to troll me 1:18:34 other people do 1:18:37 um no my trolls here are actually pretty 1:18:38 nice they usually just say I look like a 1:18:41 fat washed up movie star 1:18:46 my favorite one was I was scrolling 1:18:49 through Tick Tock and I thought it was 1:18:50 Alec Baldwin but fat 1:18:53 [Laughter] 1:18:57 you look like Rosie O'Donnell 1:19:00 can I build 1:19:01 can I build okay we said that one what's 1:19:04 po po po po po po po 1:19:07 is just a site like chat GPT but it 1:19:11 gives you access to a lot of different 1:19:14 models a lot of different chat gpts 1:19:16 basically 1:19:17 so so sure there's Cloud plus Cloud 1:19:20 instant 100K which is really remarkable 1:19:23 Cloud instant chat GPT and then there's 1:19:25 a bunch of bots 1:19:26 and the Bots are just sort of like 1:19:29 pre-prompted chat gpts so I made one 1:19:32 called uh I am Shakespeare so if you go 1:19:34 to po.com I am Shakespeare it will only 1:19:36 respond in iambic pentameter which your 1:19:39 fear of Shakespeare geek is awesome fun 1:19:41 but if you're a human being it's stupid 1:19:47 mm-hmm 1:19:51 um I need AI to look at a photo of an 1:19:53 actual person and modify that pick based 1:19:55 on the prompts that's what I just showed 1:19:56 you if so so 1:19:59 so there's a couple of different things 1:20:01 there 1:20:03 there's generative there's generative 1:20:06 fill within Photoshop so you could sort 1:20:08 of highlight the person 1:20:10 and then inverse your selection and and 1:20:13 have it fill in the background just 1:20:15 change the background so that's a way 1:20:16 you could do it 1:20:18 um look at 1:20:19 um 1:20:21 stable diffusion control net is another 1:20:24 way but you're gonna you're gonna need 1:20:25 to really geek out for that it's a like 1:20:27 all of the a lot of these tools right 1:20:29 now because we're so early in this a lot 1:20:31 of the tools require a fair amount of 1:20:33 geeking out so you need to sort of 1:20:35 figure out where you are on the how 1:20:37 comfortable am I going to GitHub and 1:20:39 firing up a repository and if if what I 1:20:42 just said makes no sense then you know 1:20:44 where you are 1:20:46 what that is and you're comfortable 1:20:48 doing that then you can go start playing 1:20:50 with things like stable diffusion and 1:20:52 control net if it sounds like I'm an 1:20:55 insane person because I am then you 1:20:58 might want to like play in Photoshop or 1:21:00 or things like that or or the example I 1:21:02 just showed in mid-journey 1:21:05 um do I have a YouTube channel I should 1:21:07 um not really I did a YouTube I did a 1:21:11 when we went into uh covid lockdown 1:21:14 um I did a daily diary my current 1:21:16 company I'm the CEO of a company called 1:21:17 storyvine which is a video storytelling 1:21:19 platform so I used our platform to do a 1:21:22 daily video of what it was like to be an 1:21:25 extrovert extrovert you know locked into 1:21:27 my office for a year I did a video every 1:21:30 day for a year and I think maybe I got 1:21:33 65 subscribers or followers for that 1:21:36 and so I don't know so I just haven't 1:21:39 done it I don't know why it would be 1:21:41 better but you know I'm willing to put 1:21:44 it anywhere I just I just find Tick Tock 1:21:46 like I've like they've given me an 1:21:49 audience and 1:21:51 um I don't know this is not my primary 1:21:54 business so I'm just kind of doing this 1:21:55 for shits and grins 1:21:57 um wait can chatgpt write better prompts 1:22:00 for me yeah absolutely in fact 1:22:02 there's a plug-in for chat CPT called 1:22:05 prompt perfect which will do it for you 1:22:07 I'm gonna go new chat 1:22:09 and then I'm going to go gpt4 I'm going 1:22:11 to go to plugins 1:22:12 if you haven't seen plugins this is how 1:22:14 they work you can activate up to three 1:22:16 I've got three travel ones turned on 1:22:18 right now so I'm going to turn those off 1:22:21 and I'm going to turn on prompt perfect 1:22:24 and so what happens with prompt perfect 1:22:27 is 1:22:29 did someone just give me a hat thank you 1:22:31 for that 1:22:32 um what happens with prompt perfect you 1:22:33 type the word perfect before your prompt 1:22:35 and then you type your prompt 1:22:38 um uh please 1:22:41 uh oh and if you find yourself saying 1:22:43 please and thank you to chat EPT that's 1:22:44 normal you're not crazy 1:22:46 uh please 1:22:49 um write me a blog post on 1:22:54 um why 1:22:55 generative 1:22:59 AI 1:23:00 um 1:23:01 AI 1:23:03 will save Humanity 1:23:06 rather than destroy it 1:23:11 so I typed that in and because I put the 1:23:14 word perfect in front of it it's now 1:23:16 going to fire off that plug-in 1:23:18 and it's going to rewrite it for me 1:23:22 and so I can open it up and go look what 1:23:24 it did so you heard what I wrote and 1:23:27 then it rephrased it as could you 1:23:29 compose a comprehensive and 1:23:30 well-researched blog post that explains 1:23:33 generative artificial intelligence has 1:23:35 the potential to be a positive force for 1:23:38 Humanity rather than a destructive one 1:23:40 please include specific examples and 1:23:43 evidence to support your argument and 1:23:45 then it writes it so yeah you can 1:23:47 absolutely use it's a it's a good way to 1:23:49 think about chat GPT as rather than 1:23:52 think of it like and again I feel like I 1:23:54 feel like Google has trained us or is it 1:23:57 brainwashed us 1:24:00 um I feel like Google has trained us 1:24:02 that you're supposed to write really 1:24:04 concise little little search 1:24:06 instructions and just sit back and wait 1:24:08 for the perfect answer and chat apt is 1:24:11 much more think of it like like 1:24:13 step-by-step problem solving where 1:24:15 you're you're you know telling it to do 1:24:18 something and then say no make it better 1:24:20 expand that help me rewrite this prompt 1:24:23 and and it's kind of this series of back 1:24:25 and forth conversations that lead to 1:24:28 um lead to you know to to better quality 1:24:31 content 1:24:33 um hey everybody my name is Kyle Shannon 1:24:35 this is the AI learning lab 1:24:37 the URLs here are are for essentially 1:24:40 for chat GPT the top one is the most 1:24:42 important one in my opinion that's the 1:24:44 official chat GPT website if you've not 1:24:46 played with tech GPT go do it now 1:24:49 um 1:24:49 I'm an entrepreneur I'm the co-founder 1:24:52 and CEO of a company called storyvine 1:24:53 which is a video storytelling platform 1:24:56 an automated video storytelling platform 1:24:58 we've been around for 11 years 1:25:00 um we're only just now 80 adding AI into 1:25:03 the system 1:25:04 um we figured out other ways to automate 1:25:06 video production then AI 1:25:09 um but but that's going to change 1:25:10 everything 1:25:12 um the purpose here is really just to 1:25:14 talk about all these new generative AI 1:25:16 tools answer any questions I can 1:25:17 demystify it get people curious get 1:25:20 people playing and uh yeah that's why 1:25:22 I'm here 1:25:24 um will it generate more poverty um it's 1:25:27 possible it'll generate more Prosperity 1:25:29 Prosperity I think long term AI is going 1:25:31 to generate a [ __ ] ton more Prosperity I 1:25:34 think short term 1:25:36 there's going to be some really 1:25:37 significant disruptions I think a lot of 1:25:39 a lot of jobs are going to be either 1:25:41 displaced or completely transformed and 1:25:43 so there's going to be a lot of upheaval 1:25:46 and disruption so I think the next two 1:25:48 two to three years is going to be pretty 1:25:50 intense just with the amount of change 1:25:52 and and you know we're we're humans and 1:25:56 while we want progress we don't want to 1:25:57 go through the chains we resist change 1:26:00 um 1:26:02 I'm I have this Instinct that that AI 1:26:06 that the generative AI stuff that's 1:26:08 coming is is going to have a similar 1:26:10 kind of impact to covid where it's just 1:26:12 gonna it's gonna Force some [ __ ] to 1:26:14 change 1:26:16 um 1:26:17 and then you know hopefully on the on 1:26:18 the other side of it there's a lot more 1:26:20 prosperity 1:26:21 how do you apply AI to Interactive 1:26:24 Intelligence Solutions I don't know what 1:26:26 Interactive Intelligence Solutions are 1:26:28 but I'm happy to 1:26:29 field that 1:26:31 wait 1:26:32 chat GPT can tell us that ignore them 1:26:37 what did you say 1:26:39 um 1:26:40 Interactive Intelligence Solutions 1:26:45 I'm gonna do 3.5 because it's faster 1:26:47 what 1:26:49 are 1:26:50 Interactive 1:26:52 Intelligence Solutions 1:26:59 a wide range of Technologies oh natural 1:27:02 language processing machine learning 1:27:08 well it's I mean like in so many ways so 1:27:11 so it's going to be able to code it's 1:27:14 going to be able to like like if you 1:27:16 look at if if you haven't 1:27:19 um go search for 1:27:21 code interpreter on YouTube and just 1:27:24 watch some YouTube videos about what 1:27:26 code interpreter does you're going to be 1:27:28 able to use generative AI to 1:27:32 to augment 1:27:35 improvements of 1:27:37 um all the machine learning stuff like 1:27:39 you're seeing it right now with the um 1:27:42 with the open source Community they're 1:27:44 taking relatively weak large language 1:27:46 models and they're using gpt4 to do all 1:27:49 the what would have been human training 1:27:51 to to make these things strong they're 1:27:53 using GPT models to do that so so 1:27:56 the AI is actually starting to train 1:27:59 itself and and there's a decently famous 1:28:03 um uh leak from a Google employee called 1:28:07 we have no moat and what he's talking 1:28:10 about in that 1:28:12 um in that statement is that he doesn't 1:28:14 feel like they have a competitive 1:28:16 Advantage with these large language 1:28:18 models because the open source Community 1:28:19 is spending like a hundred bucks and 1:28:22 they're creating models that are almost 1:28:23 as strong as what Google took 10 million 1:28:26 dollars to train 1:28:28 um so so yeah not only do I think it can 1:28:31 be applied to that stuff like it's being 1:28:33 applied to it right now and I think 1:28:35 that's going to continue I think we're 1:28:36 we're starting to 1:28:39 hit an acceleration point of innovation 1:28:42 where like if you're paying attention to 1:28:44 this stuff it is like it is it is 1:28:47 it's impossible to comprehend on a 1:28:49 weekly basis the amount of you know 1:28:51 technical innovations that are that are 1:28:53 coming and it's going to continue to 1:28:55 accelerate for probably 1:28:57 a long time like probably a decade or 1:29:00 two or three or like 1:29:01 you know until we hit Shangri-La or we 1:29:05 melt the planet it's like 1:29:08 in the middle 1:29:11 all right chat GPT only shows me 1:29:13 articles on paywall pages when I ask for 1:29:15 a source for the blog post I know 1:29:18 there's a listen the whole you know 1:29:20 don't forget we're very very early with 1:29:22 this stuff so so right now there's not 1:29:24 there not not a lot there's almost no 1:29:28 sophistication in terms of these tools 1:29:30 being aware of what they presented to 1:29:33 you they're probability engines that are 1:29:35 that are literally just saying what's 1:29:37 the highest likelihood of each next word 1:29:39 being the right word and it it sort of 1:29:41 types all those words and puts that [ __ ] 1:29:43 out there 1:29:44 and then as humans we read it and we're 1:29:46 like oh that seems really plausible and 1:29:48 they're like you lied to me why did you 1:29:50 lie to me chat gpg that doesn't seem 1:29:52 fair at all a nice person wouldn't do 1:29:54 that 1:29:56 because it it it's not actually aware of 1:29:59 what it typed or or what citation it 1:30:02 gave you or is it behind a paywall or 1:30:05 not 1:30:06 give it a year or two that like that's a 1:30:08 company that could be started right now 1:30:10 it probably is being started right now 1:30:12 is a company that you know just analyzes 1:30:15 you know citations and references that 1:30:17 these large language models put out and 1:30:19 vets them right so that'll be a plug-in 1:30:22 of a plug-in 1:30:24 um so so just be patient with it it'll 1:30:26 get there 1:30:27 all right so then everything's okay I 1:30:29 don't know if everything's okay but 1:30:31 um easiest site to generate photos 1:30:34 um well easiest site I would say the 1:30:37 easiest is probably Adobe Firefly 1:30:41 um leonardo.ai is pretty good 1:30:45 um 1:30:47 quality wise mid-journey is the best 1:30:50 um 1:30:51 the hardest one is stable diffusion but 1:30:54 it's got the most developers behind it 1:30:56 because it's open source please create a 1:30:58 YouTube channel 1:31:00 we need how-to vids please yeah I was 1:31:03 thinking I might do um 1:31:05 I might do um 1:31:07 what are they called like like Zoom 1:31:10 workshops where it's kind of an 1:31:12 open-ended you know teach you how to do 1:31:14 stuff so I'll I gotta I gotta figure out 1:31:17 my my time management 1:31:20 is there a free way to access gpt4 yet 1:31:23 yes the the bing.com 1:31:28 uh 1:31:29 [Music] 1:31:32 thing.com so when you go to bing.com 1:31:36 you get that giant search engine that 1:31:38 says ask me anything and then there's 1:31:40 this chat button and if you click the 1:31:42 chat button 1:31:44 that is gpt4 and and you can 1:31:48 um even uh 1:31:50 choose three different options more 1:31:52 precise more balanced more creative 1:31:55 so if if you know anything about uh chat 1:31:59 gpt's back end you can there's this 1:32:01 thing called temperature where you can 1:32:02 make it more creative or more precise 1:32:04 that's what that is 1:32:05 so this is gpt4 it's connected to the 1:32:08 internet and it's free 1:32:11 so the answer is yes 1:32:14 welcome to the Future provided by 1:32:17 Microsoft hi 1:32:19 I'm big 1:32:21 this is gpt4 but we're not calling it 1:32:24 that we're calling it co-pilot why 1:32:26 because you're the pilot and we're not 1:32:29 we're the co-pilot 1:32:32 we like your data your data is tasty 1:32:35 give us more 1:32:36 tell us what you really think 1:32:39 all right 1:32:41 [Music] 1:32:43 oh Tick Tock just jumped me to the 1:32:45 bottom again all right I need to go soon 1:32:47 anyway I've got I have a lovely pork 1:32:49 roast in the oven or a pork loin 1:32:53 rather 1:32:54 girded her loins I see oh 1:32:59 you're cool bro thank you 1:33:05 uh if you just came in and saw me do 1:33:08 that and your question is is he [ __ ] 1:33:10 crazy hmm possible 1:33:14 it's very likely 1:33:18 but I do try my best to be entertaining 1:33:22 and educational at the same time we like 1:33:25 to call that edutainment 1:33:27 [Laughter] 1:33:32 you can tell when I start getting like 1:33:34 this I'm a little Punchy hi Liz 1:33:38 um 1:33:39 is zapier a self-built HubSpot 1:33:43 um sort of hubspot's more specific to 1:33:46 marketing automation 1:33:48 but you can do you can recreate a lot of 1:33:50 the stuff in HubSpot with zapier yes 1:33:53 but you can do other [ __ ] too it doesn't 1:33:55 just have to be marketing you could 1:33:57 create things for say I don't know 1:33:59 chemical analysis 1:34:03 why did I respond like that I have no 1:34:06 idea the pork is smelling good though I 1:34:08 think I'm getting I think I'm getting a 1:34:10 little hangry 1:34:13 all right I I don't know I'm lost in the 1:34:15 question so if I missed your question I 1:34:17 I apologize so we are Gods now because 1:34:20 we give AI meeting I mean it's not a bad 1:34:26 it's not a bad thought you know I don't 1:34:28 know for Gods but but 1:34:30 you know I feel like 1:34:32 my experience with these tools is that 1:34:35 they are definitely Humanity amplifiers 1:34:38 and so and that's both good and bad 1:34:41 right because Humanity's complicated but 1:34:43 it it like 1:34:46 these tools take your intent and really 1:34:49 really quickly get you to a level of 1:34:52 execution that historically would have 1:34:54 taken you a week or a month or you never 1:34:55 would have done and it kind of does it 1:34:58 in 30 seconds and and then you can keep 1:35:01 iterating on that and you can just kind 1:35:03 of live at a higher level of 1:35:05 um creative expression so 1:35:09 it's it I don't know it's this is this 1:35:12 is quite profound stuff um it can't do 1:35:15 recent 2023 market research for you but 1:35:17 it can help come up with tactics that's 1:35:19 not true anymore house house bat vintage 1:35:21 that's a good name 1:35:24 um because 1:35:26 um it's connected to the internet now so 1:35:28 if we go back to 1:35:29 EBT you can make money with chat gbt if 1:35:33 we go back here 1:35:34 and we go to browse with Bing 1:35:38 so now I can say 1:35:40 um 1:35:43 um please give 1:35:47 uh give me a 1:35:50 table of monthly 1:35:54 um I don't know prices and returns 1:35:59 from the major stock indexes in the U.S 1:36:06 and it should go out to the web find me 1:36:09 some recent data 1:36:11 oh I didn't give it a time frame let me 1:36:13 stop it for a second let me copy this 1:36:17 copy 1:36:18 in the U.S 1:36:20 for the 1:36:23 um for 1:36:24 um 1:36:25 2023 1:36:29 so it goes out and it's browsing the web 1:36:30 let's see where it's browsing 1:36:37 dgia 1:36:41 it's reading the content it's thinking 1:36:44 now it's searching for s p 1:36:47 I went back to that site stood oh 1:36:49 statistica 1:36:51 all right 1:36:52 statista 1:36:56 now it's doing s p historical prices by 1:36:59 month 1:37:00 so it's off searching the internet right 1:37:02 now we'll see if we'll see if it gives 1:37:03 us a table 1:37:09 now it's doing NASDAQ Composite so it's 1:37:11 kind of figured some [ __ ] out here 1:37:16 but anyway yeah if you paid the 20 bucks 1:37:18 a month or if you use Bing 1:37:20 it's connected to the internet 1:37:25 how do I keep my requests private I'll 1:37:27 show you that in a second let's let this 1:37:28 thing finish up 1:37:33 I've collected monthly prices 1:37:35 for some of the major U.S stock indexes 1:37:38 oh it didn't put them in a table 1:37:41 here's the data well let's see if it 1:37:43 does it in the table so 1:37:46 I don't know if these are right January 1:37:48 34 000 for the Dow Jones 1:37:53 May 32 900 S P 500 NASDAQ unfortunately 1:37:58 I wasn't able to find complete monthly 1:38:00 data for the NASDAQ composite 1:38:04 um I'm going to stop it here and say can 1:38:06 you put these in a table 1:38:13 sure here's the table 1:38:15 boom 1:38:18 so 1:38:20 because it's connected to the internet 1:38:21 now you you can start to do some 1:38:24 interesting stuff now again it's not 1:38:26 checking any of this so it might be 1:38:28 getting these numbers wrong it's only as 1:38:29 good as whatever site it went and looked 1:38:31 at and if that site had bogus data or 1:38:33 shitty data then it's gonna it's gonna 1:38:36 be shitty as well but 1:38:37 this is like like it's almost like I 1:38:39 kind of feel like you should almost look 1:38:41 at all of these tools right now of of 1:38:43 like imagine what these will do when 1:38:45 they don't suck when you know when the 1:38:47 tech doesn't suck right because because 1:38:49 it's remarkable what it does it's just 1:38:51 it doesn't have a lot of the checks and 1:38:52 balances that a sophisticated piece of 1:38:54 software would 1:38:56 um yeah so so in order to to not have 1:38:59 your data um shared with open Ai and and 1:39:04 be vulnerable 1:39:05 um because you're storing 1:39:08 um your prompts out in the open the 1:39:11 first thing that you should do is export 1:39:13 your data because when you do what I'm 1:39:16 about to tell you it's going to delete 1:39:18 all of your prompt history so export it 1:39:20 first and then where it says chat 1:39:22 history and training you turn that off 1:39:24 I've got mine on because I'm not doing 1:39:26 anything particularly proprietary and I 1:39:28 want my chat history but you can turn 1:39:30 that off and that helps and then open AI 1:39:33 is also creating a business version of 1:39:35 chat EPT where I'm sure you'll have much 1:39:37 better 1:39:38 um 1:39:39 data and privacy tools 1:39:44 all right the better the prompt you feed 1:39:46 it the better the output yep absolutely 1:39:48 we have to get great at articulating 1:39:50 Vision well you have to get great right 1:39:52 now because the tools are raw but but 1:39:54 the the tools are going to start 1:39:56 prompting themselves so 1:39:58 um I don't think you're gonna need to I 1:40:00 think I think getting good at prompt 1:40:03 engineering right now is the equivalent 1:40:04 of learning math at the dawn of the 1:40:07 calculator right like yeah you need to 1:40:10 learn you need to understand the basic 1:40:11 concepts of it but we we have this 1:40:13 calculator thing now it kind of does the 1:40:15 [ __ ] 1:40:17 um 1:40:18 uh did I say what area specifically 1:40:20 they're seeing productivity increases it 1:40:23 was the it was Jason kalkanis's 1:40:25 um all in pod they were talking about 1:40:28 um 1:40:29 like tech companies in San Francisco 1:40:32 um are seeing 30 to 40 percent 1:40:34 engineering productivity increases 1:40:37 you may have covered this already but is 1:40:40 web 10 the best at generating 1:40:43 an entire website I have no idea the 1:40:46 specific tools like website generating 1:40:48 things I'm not paying attention to 1:40:50 that's why I put 1:40:52 futurepedia.io up here because it's um 1:40:56 there's just a lot of tools there I 1:40:58 can't keep up with it I I've I've 1:41:00 purposefully kind of locked myself down 1:41:03 to chat 1:41:05 chat.openai.compo.com and then the image 1:41:07 generation tools because there's so much 1:41:10 to explore just in those that I'm I'm 1:41:13 overwhelmed like just the plugins alone 1:41:15 just getting my head around how plugins 1:41:17 worked what they are and which ones are 1:41:18 [ __ ] or not it was like a week and 1:41:21 it's it's I don't know it's just it's 1:41:22 very overwhelming right now AI is not 1:41:25 the platform but humanity is I actually 1:41:27 like that line a lot I think that's very 1:41:29 right AI is a tool 1:41:31 ai's a tool it's a shovel it's a tractor 1:41:33 it's a printing press right it's it's a 1:41:37 tool that we get to use humanity is the 1:41:39 platform 1:41:41 oh oh you oh sorry I read it wrong you 1:41:43 said Humanities AI is not the problem 1:41:45 humanity is I read I I read AI is not 1:41:48 the platform but humanity is I think 1:41:50 both statements are true Humanity's a 1:41:53 [ __ ] show 1:41:54 uh it's so it's we're in such a 1:41:58 destructive 1:41:59 toxic 1:42:01 state right now I I I have very deep 1:42:06 hopes and desires that AI will actually 1:42:09 help us return to our Humanity because 1:42:11 it's going to automate out 1:42:14 the low-level shitty work that crushes 1:42:17 all of our souls and has everyone so 1:42:19 angry 1:42:21 but 1:42:22 it's not going to do that overnight I 1:42:24 think that's more like five to ten years 1:42:27 and so that means the next five years 1:42:29 are going to be 1:42:30 uglier 1:42:33 I'm so frustrated right now at our 1:42:35 country and the world quite frankly it's 1:42:38 like just like what just don't be 1:42:40 [ __ ] like how hard is that 1:42:42 apparently quite 1:42:44 naming things is hard naming things with 1:42:47 chat EBT is not hard if you give if you 1:42:50 give chat GPT here's what I'm you know 1:42:52 here are the criteria here's the brand 1:42:55 here's what I'm thinking about 1:42:58 um and especially if you give it an 1:42:59 example or two 1:43:00 tell it to give you 20 names and then 1:43:03 and then just keep iterating pick the 1:43:05 top five give it those five say I like 1:43:07 these give me 20 more give me 20 more 1:43:09 give me 20 more you can like I've had 1:43:12 some really good luck with naming stuff 1:43:14 very very quickly and very well you have 1:43:17 to you have to iterate 1:43:18 but but like with all the things think 1:43:21 of it like this try not to have the 1:43:23 expectation that chat gbt is going to 1:43:24 give you a hundred percent result ever 1:43:28 think of it like if you get your 1:43:30 prompting good you can get to maybe 80 1:43:33 90 if you're killing it and then it's up 1:43:36 to you to to close to finish it 1:43:39 um that's been my experience and and 1:43:41 when I get good at it what I find is I 1:43:43 can get I can get to 80 within five 1:43:46 minutes pretty consistently 1:43:48 and that saves me the two hours it would 1:43:51 have taken to get to that eighty percent 1:43:52 anyway 1:43:53 and then I can just be at this higher 1:43:55 level refinement creative rather than 1:43:57 blocking and tackling 1:44:00 what aif environment do you spend most 1:44:02 of your time using chechi PT that's why 1:44:05 that's there that's why it's number one 1:44:10 you need to understand your process can 1:44:13 you clip this or make a video I don't 1:44:15 know you could record your 1:44:18 phone 1:44:20 I don't know why you'd wanna this is 1:44:22 just like an add you know fever dream 1:44:26 um but if you wanna if you want to 1:44:29 record it I suppose you could I haven't 1:44:31 thought about it 1:44:32 uh I guess I could I don't know then I'd 1:44:35 have to deal with the like like just 1:44:36 everything generates more [ __ ] work 1:44:40 I think that's why I'm liking these 1:44:42 lives I can just turn them on and Bam 1:44:43 ramble 1:44:45 um 1:44:46 oh there's a lot of folks here I 1:44:47 appreciate you sticking around I'm sorry 1:44:49 for your loss but you know it's Sunday 1:44:52 thanks for hanging out with me on a 1:44:54 Sunday stopped raining here 1:44:57 um 1:44:58 if you're feeling like I'm behind on 1:45:01 questions that's because I'm behind on 1:45:02 questions 1:45:05 um I know this is probably related but 1:45:07 how do you make a picture in picture you 1:45:10 would just do that in Photoshop oh how 1:45:12 do I make this picture in picture that's 1:45:15 just 1:45:16 um 1:45:17 um when you go live there's an option 1:45:19 called dual cameras and it basically 1:45:21 just turns on the back camera and the 1:45:23 front camera at the same time and that's 1:45:25 just my laptop 1:45:28 out there so it looks fancy but it's 1:45:30 actually I've got a a helmet box and my 1:45:34 laptop sitting on top of it and then I 1:45:36 got two cardboard boxes with a light on 1:45:38 it it's not it's not all that 1:45:39 sophisticated 1:45:42 here you will see Studio here huh 1:45:44 there's that and boxes with a light nice 1:45:48 [Laughter] 1:45:53 it seems so professional 1:45:59 why does he talk like that we don't know 1:46:02 nobody knows doesn't he look like Rosie 1:46:05 O'Donnell a little bit yeah 1:46:08 oh 1:46:11 [Laughter] 1:46:15 hang on I get my light 1:46:20 [Laughter] 1:46:26 where'd you spend your Sunday afternoon 1:46:27 doing I don't know there's this guy 1:46:30 he seemed mentally challenged but uh I 1:46:33 learned some things 1:46:37 do you think the free gen AI pathway 1:46:39 training on Google oh I just did the 1:46:42 free gen AI pathway training on Google 1:46:44 and I can finally understand you oh 1:46:45 that's really cool Lily that's awesome 1:46:47 um yeah I think so so a couple of things 1:46:50 so Google just 1:46:52 um did that training there I I think it 1:46:54 is good to understand how these tools 1:46:56 work because it's like it's really easy 1:46:57 to have misconceptions that They're 1:46:59 copying and pasting and they're stealing 1:47:01 people's work and and the robots are 1:47:03 doing this it's it's not like that it's 1:47:05 it's a bit more pedestrian even though 1:47:07 they're remarkably powerful it's 1:47:09 actually fairly straightforward and if 1:47:11 you understand that it 1:47:13 it it it it like takes this AI [ __ ] off 1:47:17 the pedestal right like these are just 1:47:20 tools right they're just tools so that's 1:47:23 awesome that you took that um openai 1:47:25 also just did one on prompting how to 1:47:27 get better results with prompting 1:47:29 um and that that looks pretty good Bart 1:47:31 is a dud I agree with that 1:47:33 Google invented this [ __ ] and they are 1:47:35 way behind they're getting their asses 1:47:37 handed to them 1:47:39 um I tried using chat GPT to write a 1:47:41 script to dump output into a Google 1:47:43 sheet that's cool that's a cool use case 1:47:46 yeah if if you um 1:47:49 it's really good if if you're one of the 1:47:51 people like me that kind of understands 1:47:53 what spreadsheets are but like I don't 1:47:54 know formulas and how to do anything in 1:47:56 spreadsheets just go to chatgpt and say 1:47:59 I'm trying to do this can you please 1:48:00 write me that formula it will 1:48:03 I ended up having to download python3 1:48:06 never did get the script to work 1:48:08 um you can also so when I first did my 1:48:11 first python application and I told Chet 1:48:14 GPT I didn't know how to run it it said 1:48:17 download python then I said I don't want 1:48:20 to download python can you give me a 1:48:23 website where I can run the python in 1:48:25 the website and it gave me like three of 1:48:27 them so just tell it to tell it you want 1:48:30 to do it you want to run your python 1:48:31 code on a website if you just want to 1:48:33 dick around with it because otherwise 1:48:35 you're like installing things you have 1:48:36 to go to the terminal 1:48:38 look look 1:48:41 I find it exhausting 1:48:45 um 1:48:46 what do we want we want to pay for yeah 1:48:48 okay 1:48:51 um 1:48:55 trying to find a question okay what's 1:48:58 that say which AI answers which AI 1:49:01 answers questions with graphs and 1:49:03 diagrams so I haven't played with these 1:49:06 a lot but there are 1:49:09 in the 1:49:11 open AI plugins 1:49:16 there are two that I've got installed 1:49:18 that I haven't really played with 1:49:21 one's called show me 1:49:24 called I thought I had another one 1:49:26 installed 1:49:30 diagram 1:49:32 Daya gr dot am 1:49:35 and then if we go to the plug-in store 1:49:42 um let's type in 1:49:46 [Music] 1:49:48 let's see graph 1:49:55 diagram image search Golden current now 1:50:01 let's see 1:50:03 um visual 1:50:11 this one's called notable create 1:50:13 notebooks in Python SQL and mark down to 1:50:16 explore data visualize and share 1:50:19 notebooks I don't know I would just go 1:50:20 to the plugins and start exploring but 1:50:22 but those two that I that I described 1:50:26 um 1:50:28 are 1:50:30 are pretty good 1:50:33 um what are they called again diagram 1:50:34 and 1:50:36 show me 1:50:41 but again like give it a week or two and 1:50:44 all this stuff will be uh 1:50:48 will be handled like just new tools will 1:50:51 come out is there a way to integrate AI 1:50:54 into a Mac make it a bit like Jarvis 1:50:56 yeah check out GPT for all 1:51:00 um if if you're on an M1 Mac that helps 1:51:03 a lot of the developers are developing 1:51:05 the the local 1:51:08 um large language models for Apple 1:51:10 silicon um so so m1s or higher m1s m2s 1:51:15 um if you're on an Intel Mac I'm sure 1:51:18 there's something you can do there but 1:51:19 it's probably going to be pretty slow 1:51:20 and and there's probably not going to be 1:51:21 a lot of support 1:51:23 um but but go look look up GPT for all 1:51:25 and then just go to go to YouTube and 1:51:28 and search for open source large 1:51:31 language model 1:51:39 you're on a horse in an office your poor 1:51:41 co-workers oh man I'm so far behind on 1:51:43 questions I have to go actually because 1:51:45 I have to go make dinner is done for 1:51:48 dinner I don't use Google to search for 1:51:51 much anymore unless on my phone I'm the 1:51:53 same way even on my phone I've got I 1:51:55 cheat I I turned 1:51:57 chat.openai into a into a button on my 1:52:00 phone and I have it on the bottom four 1:52:01 it's like one of my main buttons and I I 1:52:04 I'll just go there if it's anything at 1:52:06 all like I need an answer to this I 1:52:09 don't go to Google anymore at all 1:52:11 it's it's got to be it's got to be 1:52:14 killing them because there's not a like 1:52:16 I don't know I don't know a way out of 1:52:18 it for them 1:52:19 how do you create a business strategy on 1:52:21 GPT go to GPT tell it to be a business 1:52:24 consultant and have it ask you questions 1:52:28 about your business in a way that it 1:52:30 gives it enough information to help you 1:52:32 develop a strategy and then just start 1:52:34 answering its questions you can even do 1:52:36 a thing like I want you to ask me a 1:52:38 series of questions about my business 1:52:40 that will will give you enough 1:52:41 information to help me create a strategy 1:52:43 I want you to ask the questions one by 1:52:45 one and don't move on to the next 1:52:47 question until I've told you okay move 1:52:49 on or something like that and it will 1:52:51 literally start acting like a business 1:52:54 consultant walking you through the 1:52:55 process 1:52:58 um does paying for it keep your 1:52:59 information in prompts private no 1:53:02 they're they're creating a business 1:53:04 version of chat CPT that might but not 1:53:06 right now you can disable if you go to 1:53:09 data controls you can disable them using 1:53:12 your data for training and you can 1:53:14 disable your 1:53:16 chat history your prompt history 1:53:19 um 1:53:20 but it will delete your entire prompt 1:53:21 history so that's why I haven't done it 1:53:23 yet but but the 60 month the plan 60 a 1:53:27 month plan does I assume that's for 1:53:28 mid-journey 1:53:30 you're cool bro thanks fatty boom butts 1:53:35 good name 1:53:36 [Laughter] 1:53:38 chat GPT for emails my students wait 1:53:42 chat ept4 emails to my students parents 1:53:46 what prompt preface can I use to make 1:53:48 them less formal uh just tell it to be 1:53:50 less formal so if it gives you something 1:53:52 that's really formal say 1:53:54 um you can even sort of berate it a bit 1:53:56 you can say oh that was way too formal 1:53:58 can you make it more conversational 1:54:01 um even better than that 1:54:03 take an email that you've written that 1:54:06 that is in your exact tone of voice 1:54:08 you're like yeah that's the perfect 1:54:09 example and say based on this email 1:54:12 paste your email in there please write 1:54:14 and you know an email to this parent 1:54:17 saying that their kid was an [ __ ] in 1:54:18 school bang and follow that one as an 1:54:21 example and it will copy your tone 1:54:24 but you can also just give it feedback 1:54:26 that was that was you know make me sound 1:54:28 smarter make me make it more relaxed 1:54:30 make it more conversational 1:54:32 I only know chat GPT 1:54:35 just keep digging I mean I mean listen 1:54:37 only no chat gbt chatgpt is the most 1:54:41 remarkable tool in the history of 1:54:44 humanity 1:54:45 like if you only know Chet I could take 1:54:49 the next decade at the current level of 1:54:52 chat gpt's sophistication and and 1:54:55 probably not even begin to scratch the 1:54:57 edges of what it makes possible so don't 1:55:01 feel like I only know chat GPT is like a 1:55:04 bad thing like it's it's it's a 1:55:06 remarkable tool that that you know use 1:55:09 it 1:55:10 do you wait you do any fpga development 1:55:14 I don't know what the [ __ ] that is let's 1:55:16 go find out 1:55:18 can make you less stupid 1:55:24 um here we come to save the day what is 1:55:31 fpga fpga 1:55:36 [Music] 1:55:39 stands for field it's a little slow 1:55:43 programmed 1:55:45 programmable logic blocks 1:55:47 wired together okay so this is like 1:55:49 controller kind of stuff I'm guessing 1:55:51 and what's the question I don't do any 1:55:53 of that no I like I I assume like 1:55:56 Arduino kind of crap is sort of on the 1:55:59 edges of that 1:56:01 um but I don't but but this can help you 1:56:03 do all that stuff it can write the code 1:56:05 it can probably design circuits 1:56:08 um gpt4 when they fully roll it out is 1:56:12 multimodal meaning it can see 1:56:14 um it can understand images so I'm sure 1:56:17 you're going to be able to upload like 1:56:19 circuit diagrams and it will be able to 1:56:21 understand them I mean 1:56:23 this shit's coming it's coming quick 1:56:26 I I wish the multimodal thing were here 1:56:29 but it's not it's just like you know 1:56:31 listen they're they're kind of you know 1:56:34 flying the plane as they're building it 1:56:36 here 1:56:39 um your feed freezes for a few seconds 1:56:41 every time you switch camera angles oh 1:56:43 please pause I I knew that sorry about 1:56:45 that I'll try to remember that I'm 1:56:47 thinking about starting a tick tock 1:56:48 Channel 1:56:49 on main historical sites yeah awesome 1:56:52 and and you know and have chat GPT give 1:56:56 you you know here's the three or four 1:56:58 bullet talking points or here's your 1:57:00 script if you like things more scripted 1:57:03 um about each of those sites 1:57:05 or have you know have it give you like 1:57:07 like you can have it give you like what 1:57:09 would be a good order starting with the 1:57:12 most popular to the most obscure or vice 1:57:15 versa things like that 1:57:16 um being able to replay these vids would 1:57:18 be amazing sorry about that 1:57:22 um we have a beautiful stay with lots of 1:57:24 lighthouses and great landmarks yeah I 1:57:26 think that's a really cool idea like 1:57:28 yeah I would encourage you to start a 1:57:30 tick tock Channel as a 58 year old to 1:57:32 start at a tick tock Channel because I'm 1:57:34 a [ __ ] idiot why not 1:57:37 you know I've I have a decent amount of 1:57:39 followers now and people seem to hang 1:57:41 out in these lives and enjoy them 1:57:43 um I'm gonna come back I'll come back 1:57:45 later but let me let me take off I know 1:57:47 I didn't get to a bunch of questions but 1:57:48 thanks for hanging out with me let me go 1:57:51 uh help my son work on the pork loin 1:57:56 that's good the loins people uh and I'll 1:58:00 be back later and uh peace everybody I 1:58:03 hope this helped I hope it was fun uh I 1:58:05 appreciate you all hanging out 1:58:07 um give a thumbs up go follow me if you 1:58:10 like these subscribe if you want to and 1:58:12 I'll see you in a bit 1:58:14 all right