6/17/2025 - Embracing the Jank: An Evening of AI Experimentation and Frustration

Description

Chapters

Transcript

0:12 [Music]
0:25 All praise is lovely. of grace. Yes. In
0:32 his own place. He discovered his own
0:35 place. Buried himself alive
0:39 inside his basement
0:42 of his face. Working away on
0:45 displacement.
0:47 What do we take to survive?
0:50 [Music]
0:52 When you're done with this world,
0:56 [Music]
0:57 you know all the next is up there. Yo,
1:01 [Music]
1:12 as he learned how to turn turn.
1:17 [Music]
1:50 [Music]
2:03 Oh, Champy, you're loud. You're loud and
2:06 grumpy tonight.
2:08 [Music]
2:12 Well, every time I say it now,
2:16 I get that look in mine.
2:20 Every time I see your mouth, I
2:24 hear that smile.
2:28 The early misty morning light. I heard
2:32 the engine.
2:35 The old outside.
2:40 [Music]
2:44 You a living man
2:47 again today.
2:51 You a convince me
2:54 again today.
2:58 Leaving the time looking for someone
3:02 else's golden ring.
3:04 [Music]
3:07 Should
3:09 so long Suzan
3:16 now don't cry.
3:22 [Music]
3:27 Oh, what's happening people? Happy
3:28 Tuesday. Tuesday. Tuesday. What's going
3:30 down? What's shaking? What's happening?
3:34 Everybody good? All right. What are we
3:36 doing? Why did my air conditioner go
3:37 off? Do I have it on auto? Hang on. I'm
3:40 I'm blasting hot here.
3:43 [Music]
3:46 Cool. Timer.
3:49 What's up?
4:01 [Music]
4:13 Please tell me this didn't just stop
4:14 working.
4:16 [Music]
4:22 I think the old air conditioner just bit
4:24 the dust. People, what are we going to
4:26 do? What are we going to do? Oh, now
4:29 it's coming back on. All right, fine.
4:31 Whatever.
4:34 How is everybody? What's going on? What
4:37 do you want to talk about tonight? We
4:38 can talk about
4:44 [Music]
5:22 Down
5:28 down down down.
5:31 [Music]
6:30 What is the superbase in lovable?
6:34 Superbase in lovable. So, Superbase is a
6:37 database and so it's a Supabase is like
6:42 a hosted database management service
6:47 that's partnered with Lovable. So, it's
6:49 incorporated into Lovable. So, when you
6:51 make an app,
6:53 like I made a an app that's a to-do
6:55 list, right?
6:57 And if I didn't have a database there,
7:00 then every time I would add to-dos to
7:03 the to-do list, the next time I launched
7:05 the app, all my to-dos would be gone.
7:08 Um, and I couldn't have any other people
7:10 log into the system. So, there you go.
7:14 YouTube requests. We don't do much with
7:17 Gemini, it seems to me. Can we play with
7:21 that a little? I'm completely unfamiliar
7:23 with its capabilities. Um, we can play
7:26 with it a little bit. I am not super um
7:34 I don't I haven't been following all
7:36 their models, so I don't know a lot of
7:37 their models, but we can go to um
7:43 Well, we'll start with with
7:45 gemini.google.com.
7:48 [Music]
8:18 pretty melancholy. Oh, the song. Yeah. I
8:21 don't know what I'm really playing here.
8:26 [Music]
8:32 Oh, this is
8:36 [Music]
9:19 I don't know. I don't know what it is.
9:22 I do not know what it is. Um, anyway,
9:27 Mr. Yeah, Superbase is the thing that
9:29 allows you to have a database in your in
9:31 your apps. Uh, Wolfman Clint, thank you
9:34 so much. I appreciate that. Appreciate
9:36 the gifts. It's quite generous of you.
9:41 Um,
9:44 Do
9:46 let's see. Oh, you know what we could
9:47 do? I have an idea.
9:52 This is going to be heresy.
9:56 All right, hang on. So, let's see. New
9:59 chat.
10:01 We go. Plus, I'm going to go
10:04 search
10:06 uh let's see
10:09 search the web and
10:13 tell me which models
10:17 are available
10:20 right now on
10:24 gemini.google.com.
10:31 and what each of them are best for
10:37 and
10:40 which ones have multi-
10:46 modality
10:48 or
10:50 special features.
10:54 All right, let's see.
11:00 Oh, okay. Cool.
11:03 All right. Let me go look there. Got a
11:05 DM from
11:08 Yeah, but that just didn't that just say
11:09 that's that's the new things they added.
11:11 I just kind of wanted a
11:13 a list.
11:20 Oh, yeah. They've got this chart.
11:23 Yeah, this is so exhausting.
11:32 Source camp in the house. What's
11:34 happening? You can make Gemini explain
11:36 itself to you. Yeah, I could.
11:41 [Music]
11:51 Yeah, silly me. Let the AI do the do the
11:53 stuff. Yeah, that's I went to chat GPT
11:55 to ask about Gemini.
11:58 All right, the Gemini 1.5 Pro series.
12:01 Midsize, high capacity, handles huge
12:03 inputs. Let me see if it's cool out. We
12:05 just had a rainstorm, so it might be
12:07 cool out.
12:13 [Music]
12:16 Oh, someone's putting on a fire where
12:18 something's burning.
12:20 Joe 1.5 Flash 2.0 series 2.0 2.0 Flash
12:24 2.0 Pro 2.0 You know 2.5 series 2.5
12:28 flash
12:30 2.5 Pro flagship reasoning model chain
12:33 of thought
12:38 [Music]
12:41 V3 Gemini Robotics multimodal live API
12:47 oh comparison
12:51 context window
12:53 about a million tokens for the 2.5 five
12:55 models.
12:58 That's kind of cool. All right. All
13:00 right. Yeah, we'll go we'll go play. In
13:02 fact, I can I can go over there now just
13:06 so I don't forget. gemini.google.com.
13:11 And then I think I've got to switch
13:16 better text and image quality with image
13:18 gen 4.
13:22 All right,
13:23 that's cool.
13:28 Let's see. Let me try this one.
13:31 Yeah, this is ultra. So, I should have
13:33 access to everything here. 2.5 flash and
13:36 pro both new
13:41 deep research canvas
13:44 video.
13:50 Oh, okay. All right. So, can you see can
13:53 you see can y'all say can y'all see what
13:56 I'm looking at here? I've been doing
13:59 some serious work in the past two days
14:01 on my biz. Couldn't have done it without
14:03 chat. Oh, that's interesting. 25,000
14:06 likes coming and brought to you by Mr.
14:08 It.
14:11 Yeah.
14:13 Didn't I say that? You should be looking
14:14 at Gemini right now in northeast
14:17 Northwest Ohio. It's 79° and about 300%
14:20 humidity. Yeah, it's a little moist
14:22 here, too. Although, it's a little
14:23 cooler out than than it is inside. So, I
14:26 think I'm going to be okay.
14:28 Um, but something is definitely up with
14:30 the old air conditioner. It's not happy.
14:32 Oh, you know what it probably is? I know
14:34 what it probably is. I probably haven't
14:36 cleaned it out and it's probably all
14:37 full up and so it's overheating.
14:51 [Applause]
14:55 vibe sweating.
14:59 Oh yeah.
15:05 Hey, I do have a question. Is anyone in
15:08 here, Danielle, you might be one of
15:10 these. Is anyone in here
15:13 using um Flux context? And if so, what
15:18 what tool are you using it at? Are you
15:20 using it in some other tool, some video
15:23 tool or some image gen tool? Where's
15:26 anyone using Flux Context?
15:35 [Music]
15:36 Yes, Flux Pro. You're using it at the
15:38 site. Mr. He's using a Hedra. Yeah, I
15:41 don't like My problem is I don't like
15:43 Hedra's implementation unless they fixed
15:46 the problem that I had the other day,
15:49 which was my problem was
15:52 I couldn't get it to do an aspect ratio
15:54 that wasn't the input as aspect ratio,
15:56 and that was pissing me off.
16:00 Hedra image
16:04 flux context
16:07 prot2.
16:11 All right. Well, let's put in a
16:15 reference image of Sydney.
16:20 We'll get if I know we're supposed to be
16:24 playing on Gemini, right?
16:28 Welcome to chat add.
16:32 Okay. So, we'll say um actress
16:38 in in
16:41 a flowing
16:46 red dress on a
16:50 Broadway stage with
16:57 See, I can't change my aspect ratio
17:00 here.
17:03 Can I? I can't.
17:08 God damn it.
17:14 On stage with
17:19 large curved
17:22 LED screen
17:26 behind
17:28 the actress.
17:32 on the screen
17:35 are the Swiss
17:38 Alps.
17:40 And then we're going to generate that
17:42 image and it's just going to generate a
17:44 tall image.
17:46 You can use it for free from from Black
17:48 Forest Labs. I just was on Black Forest
17:51 Labs and it said I had to pay. I think
17:57 Black
18:00 forest
18:07 collapse.
18:11 By the way, I'm watching a new TV show
18:13 called Stick with uh oh, what's his
18:16 name?
18:18 I don't know. Mark Marin's in it. And
18:20 then the one of the silly brothers.
18:25 It's about a golfer, little old has been
18:28 golfer taking on a talented kid. Flux
18:31 one context
18:34 log in
18:41 [Music]
18:50 available credits zero.
18:56 Oh, that's API. Why am I in the API? I
18:59 don't want the API
19:03 dashboard. Shouldn't this be app?
19:09 No,
19:13 [Music]
19:18 I'm there now. And you can try it for
19:20 free.
19:24 How can I add a video interregulars from
19:26 my computer without using YouTube?
19:32 Uh
19:44 flux.pro
19:46 Pro.
19:49 No.
19:53 God damn it. Flux Pro.
19:59 Flux Pro Web.
20:05 Oh, you know what?
20:16 Get started. Try demo.
20:24 Flux Pro AI.
20:31 Is this the one you use, Joy Perie?
20:35 Lucia.pro
20:39 fluxai prot
20:44 Jesus. How many URLs are there? This is
20:47 unbelievable.
20:49 fluxai.pro.
20:57 I just want like one where I can change
20:59 the aspect ratio.
21:03 Now I got to log in. Annual sale 30%
21:05 off. I the This has got to live inside
21:09 some tool I'm using. Hang on. There's
21:11 got to be Deco here. Is it in Decco here
21:13 yet?
21:15 Because I pay for other [ __ ]
21:19 Um features, use cases, get started,
21:23 log in.
21:31 I'll tell you what, the
21:34 if you're new to AI, the complexity
21:38 of this world right now, you know what's
21:40 funny? AI at this point is starting to
21:43 turn into it's not quite as bad as NFTTS
21:46 and NFT bros, but it's pretty [ __ ]
21:49 close because because we now have
21:52 powerful open-source models. The minute
21:54 an open source model drops
21:57 um or something like flux context that's
22:00 really good, everyone incorporates it
22:02 and then all these shitty little apps
22:04 spin up that have all similar URLs. It's
22:08 just it's just bad. Squirrel. Yeah, I
22:10 know.
22:12 I know.
22:14 Playground.bff.ai
22:18 AI
22:21 generate images.
22:27 Can I pick a model here? No.
22:30 No. Of course you can't. Of course you
22:33 can't.
22:36 I'm exhausted. I I just want to
22:44 Could something just work? I don't know
22:47 which site I have access to. See pin.
22:52 Okay. I don't see the pin.
22:57 Someone make him look at my pen.
23:00 What's your pin? There's nothing pinned.
23:03 Danielle,
23:05 I know. I saw that one before. I just
23:07 don't trust it. Playground.
23:12 Bff.ai.
23:18 No, that's not right.
23:22 Someone please make him look at my pen.
23:25 BL
23:27 Oh, Black Forest Labs. BFF.ai is
23:30 available if you'd like to buy it.
23:33 BFL.ai.
23:35 Welcome to the Flux Playground. Get
23:37 started. Okay.
23:40 Yes, I agree. I'm over 18. I I agree
23:43 with you. Yes, you can have all of my
23:45 data. I already gave it to Google. In
23:48 fact, rather than collecting my data,
23:51 why don't you just get it from Google?
23:53 They've already got it all. Okay. Okay.
24:02 So, this only has flux one and then
24:06 Hedra just had flux two.
24:10 Like there's our girl.
24:15 Oh, here's aspect ratios. But see, it
24:18 won't it it it does not honor the aspect
24:21 ratio.
24:30 Make this a wide image
24:35 of actress.
24:40 in three quarter view
24:45 with
24:49 LED screens.
24:52 Representing
24:54 a modern
24:57 news room.
25:02 Yeah, but I can't see. I can't select
25:08 aspect ratio. All right, so that
25:10 definitely doesn't work, which makes me
25:13 cranky. All right, so here we're going
25:15 to upload.
25:20 I like her. I need Kyle. Um,
25:24 do you have a bread maker that you would
25:26 recommend?
25:33 Oh, harsh weed. That is the best
25:35 question I've had all night.
25:44 Actually, you know what?
25:46 I I do harshed I do have a bread baker
25:49 I'd recommend. Watch this. We're go new
25:52 chat. We're going to go We'll do 03 mini
25:55 high. We're going to go deep research.
25:58 We're going to say, um, I want you to
26:03 create a,
26:06 um,
26:08 categorized
26:13 report
26:15 of three RECO
26:18 men
26:20 bread makers
26:25 in low, medium, and high price points.
26:33 You decide the criteria
26:38 and don't
26:40 mess it up.
26:45 There's
26:47 a stoned
26:49 dude who's munchies
26:54 depend on the answer.
26:57 [Laughter]
27:04 Which country or region? US.
27:08 Are there any preferred features? Um
27:15 uh flexible
27:18 loaf size
27:23 and should the
27:26 low do you uh you define ranges? Okay.
27:33 Tik Tok pin when you're done. Joy pretty
27:35 Kyle. I recommend you change your
27:38 actress character references to the
27:40 aspect ratio
27:43 you
27:45 wait you
27:48 want in the final. Will it put her on
27:51 different sides?
27:56 It's exhausting. Joy,
27:59 shouldn't these tools be magic?
28:03 I don't want to have to work.
28:06 Work is hard.
28:21 Don't forget about the critical question
28:23 of altitude. Oh [ __ ] Can I can I stop
28:27 it and put that in there? Oh, but I
28:30 don't know where where uh Harshweed is.
28:32 Oh, Harshweed. What? What? Uh
28:37 cuz I'm in Denver, so I'm a mile high.
28:41 Don't stop. We We'll just We'll let it
28:43 go.
28:45 Kyle, why are you on mini high? Cuz it's
28:48 good at doing thinky [ __ ] and it's
28:51 faster than 03.
28:53 Um I want to know the best AI
28:56 refrigerator to buy.
29:03 Oh my god. Okay. I just I you know
29:07 there's some nights I have patience for
29:10 AI jankiness and there's just some
29:12 nights this is one of them.
29:16 We are one one search away from becoming
29:18 a Tik Tok shop. Well actually so harsh
29:21 weed harsh high. What's the [ __ ] your
29:23 name? Harshw weed. I think one of the
29:25 reasons I wanted to do this in chat GPT
29:28 is chat GPT has incorporated
29:32 um a shopping uh a shopping API into
29:36 chat GPT. So when it recommends these
29:39 bread makers, it should give us links to
29:42 go buy them.
29:46 Elevation. Yeah, I know. All right.
29:50 [Music]
29:53 Context context
29:56 doesn't have aspect ratio choices in
29:58 fluxai.pro.
30:00 Okay. Kyle Shannon, the appliance king
30:02 of Denver.
30:11 Hi. You might know me from my famous Tik
30:14 Tok channel, the AI Learning Lab. But
30:16 what I bet you didn't know, I'm the
30:18 appliance king of Denver. Hi, I'm Kyle
30:21 Shannon.
30:23 [Laughter]
30:27 It was funny the other night I was on a
30:29 Tik Tok live and we were talking about
30:31 AI and someone asked me to do one thing
30:33 and then I was doing something else and
30:35 then someone else said, "Hey, can you
30:37 recommend a bread maker?" Funny enough,
30:39 I could.
30:47 I ache in my soul.
30:55 [Laughter]
31:00 It's only 1999.
31:03 Kyle, irregulars, please. Oh no. What
31:06 have I done? What have I done?
31:10 Or what have you done? What have you
31:11 people done?
31:13 You people
31:18 Is is this Flux Pro?
31:30 I I I feel I have been done a a ma major
31:34 disservice here. A major disservice.
31:38 This is not kind. This is really This is
31:42 seriously
31:44 paint modders in here. You You can't be
31:47 showing this this kind of
31:50 What is it? This is defamation.
31:52 [Laughter]
31:59 Oh my god. Bit of a time machine there.
32:03 Yeah, I know. It made me look so young,
32:06 those pictures.
32:12 Oh my god.
32:16 All right. So, Joy Party
32:20 being a buzzkill is like, "Kyle, stop
32:23 being lazy and just go change your
32:24 aspect ratio." And I'm like, "But Joy,
32:29 I don't mind.
32:32 It makes me cranky."
32:34 All right, fine.
32:37 Per Joy Party's strong recommendation.
32:43 [Music]
32:50 [Laughter]
32:52 [Music]
32:54 Does he always sing like this? This is
32:56 rather annoying. I thought this was the
32:59 AI learning lab. I thought we were going
33:00 to learn something here.
33:05 Dark.
33:06 dark green
33:10 background.
33:12 It's so funny. Last night I was all
33:14 talking philosophical and stuff. Tonight
33:16 I'm like, I'll do some work. And now I'm
33:18 starting to do work and I'm like I I got
33:20 no ideas in my head and I don't want to
33:22 work. I don't want to do this. I don't
33:24 want to have to deal with life.
33:29 It's just too much. It's too much. Too
33:32 much. All right, we're going to erase.
33:35 We're gonna We're gonna We're gonna
33:38 paint. We're gonna just Everybody calm
33:42 down. Everybody calm down. Everybody
33:45 calm down. This is not your fight to
33:48 fight. Dark green
33:50 gradient
33:53 background.
33:56 Styliz 300. Let's do styliz 50. We don't
33:58 want [ __ ] style. I don't want your
34:00 style. I don't want your stupid style.
34:02 Oh, it's probably because I had
34:03 personalization turned on. Listen, it's
34:06 not the point, people.
34:09 [Laughter]
34:13 I need that picture, Kyle. No, Mr. You
34:16 can't have Sydney. She's mine. She's
34:19 mine. You got to wait for her to be on
34:21 Broadway.
34:24 [Laughter]
34:28 Come to the AI learning lab. It's
34:30 educational and entertaining. We like to
34:32 call it edutainment.
34:35 TM.
34:39 So, listen. I will I will gladly gladly
34:43 Wait, did that not submit? It didn't. I
34:46 will gladly
34:49 admit
34:51 that there are certain nights on this
34:53 channel where no one should be subjected
34:56 to this. and and I I appreciate you all
34:59 more than you know.
35:05 All right, all of those are good. So,
35:07 let's let's upscale her to the gallery,
35:10 shall we?
35:12 And Joy Party's like, "See, Kyle, it
35:14 only took like 30 seconds." I look I
35:16 know
35:19 I know
35:21 but don't you understand how my brain
35:23 works once I get it in my head that I
35:26 should be able to start with a vertical
35:27 image and just tell it to make a wide
35:29 image and it should [ __ ] figure it
35:30 out because it's got all of the
35:32 knowledge of all of the history of
35:34 humanity.
35:36 That seems reasonable to me.
35:38 It's just it's just life is exhausting.
35:41 Life is exhausting. Have you guys tried
35:43 life?
35:45 It's a lot. It's a lot. It's a lot. All
35:47 right. There she is. She looks great.
35:49 No, Mr. It, you can't have her.
35:54 Is she working full-time?
36:04 I can't. Everyone is is is hitting on my
36:09 Sydney. I got I got to get her cast in
36:12 the role first.
36:14 Joy Pertie is trying to change her
36:16 aspect ratio. Mr. It is trying to date
36:19 her. I listen. You think this is easy
36:22 running a channel with all you weirdos?
36:25 We're not weird. We're regular. I know.
36:28 Well, what I' I've completely lost my
36:30 train of what I was doing. Where was I?
36:32 Oh, I was over in flux. Okay,
36:43 let's go back to Hedra. Let's go back to
36:45 Hedra, shall we? Okay, let's get rid of
36:49 this. We're going to get rid of her. Uh,
36:51 did I download her?
36:56 There you go. Screenshot it. Mr. It t
37:06 Life is exhilarating, but the
37:08 alternative is so final. I know, right?
37:12 Um, okay. So, there she is. We've got
37:14 this. Make this a wide image of the
37:17 actress. Three/4er view with Ellie's
37:20 modern newsroom.
37:23 Um,
37:25 okay.
37:29 Refresh irregulars, please. Fine, let's
37:32 go do that. Oh no, I have a feeling this
37:35 is not going to go well for me.
37:49 [Laughter]
38:07 It gave me one normal hand and one
38:09 Donald Trump hand.
38:18 Oh, good God. Who did this? Is this Who
38:21 did this? That's Brandon. That's
38:24 Brandon. What tool did you use? What
38:27 tool did you use? Embrace the jank.
38:31 GPT4. Okay.
38:36 What the heck is going on here tonight?
38:39 Leave you alone for 30 minutes. Oh my
38:41 god.
38:47 Oh, is that an AI refrigerator behind
38:50 him? Oh, let's go. Wait, where? Let's go
38:53 back to Hedra
38:56 and see if we got our Oh, there she is.
39:00 But wait, what was
39:03 Okay, so here's what I want to do now.
39:05 So So now let's let's put same image in
39:08 there.
39:11 Okay,
39:14 actress.
39:15 Let's see. Um wide
39:21 uh full
39:24 body
39:28 shot
39:30 of actress
39:35 on in the left
39:41 of the frame
39:43 in a
39:46 dark blue
39:48 LED lit dress
39:54 with
39:57 a
39:59 small home office
40:03 behind her.
40:07 And on the other side of the stage,
40:16 a huge
40:18 LED curved
40:21 screen
40:23 with a
40:26 chat conversation
40:28 that says,
40:33 "I love you.
40:35 I want you
40:38 to leave your wife.
40:42 Exclamation point. Okay.
40:47 Let's see if we can get uh what's it
40:50 called? Context. This thing uh flux
40:55 context. We'll see if we can get it to
40:57 do
40:58 what I'm hoping it can do.
41:04 Oh my goodness. My goodness. Glory. My
41:07 glory. Goodness.
41:09 Oh lordy. I thought it was Hedra. I'm
41:14 spelling it wrong. Yeah. Sorry. Uh yeah,
41:17 it's H E D R A. I've been pronouncing it
41:20 Hedra, which I think is right, but then
41:23 people that work there call it Hedra.
41:26 And so again, if these companies would
41:29 just hire
41:32 I love you. I want you to leave your
41:34 wife. All right. Well, that sort of
41:37 worked.
41:40 But wait, didn't I say
41:43 in a dark blue LED lit dress?
41:47 That is dark blue, but it's not LED lit.
41:50 This is Yuddio all over again. It really
41:53 is.
41:56 Um,
42:04 yeah, I gotta I gotta
42:07 You might need to say extreme wideshot.
42:09 I know. This will definitely be one to
42:12 rewatch on YouTube. I don't think there
42:15 Rick Olsen, I don't think there is
42:17 anything. So far, there has been nothing
42:20 on here other than me whining. If you
42:23 want a lesson in how to be an ingracious
42:27 um incompetent
42:30 host of a semi-educational channel,
42:33 tonight is the master class. If you
42:36 wanted to actually learn anything, I you
42:39 there's I'd go back a couple of days.
42:47 Oh, good lord. All right. Whatever. I
42:50 just
42:55 I've got no patience. I've got no
42:57 patience for the jank tonight. I tell
43:00 you all to embrace the jank and that's
43:02 that's on me. That's on me. I shouldn't
43:05 have done that. It's a lot of pressure
43:07 to put on yourself.
43:11 Flux context T2
43:15 T2i.
43:23 Um,
43:29 describe
43:31 the difference
43:33 between
43:36 flux
43:39 context,
43:44 T2, max,
43:47 and pro. See, it seems to me
43:52 if you're going to name something
43:55 with sort of a naming convention,
44:00 you shouldn't have the modifiers both be
44:02 like the best. Like max is the top and
44:05 pro is pro. Like what's above pro?
44:08 Nothing. Below it is amateur
44:11 or is normal.
44:14 I don't I I
44:22 Pro fast max max maximum fidelity.
44:28 All right. So, let's go back here.
44:32 Max maximum fidelity.
44:34 All right.
44:37 Let's see.
44:47 I know. I just don't care. I don't care.
44:49 It You know what? You know what's
44:50 happened tonight? You You caught me. I
44:53 finally I I I have been broken. I have
44:56 been shattered by the jank. The jank has
45:00 caught up.
45:03 It It has shut down any of my desire
45:07 to to figure this [ __ ] out.
45:12 It happens.
45:14 All
45:15 right, let's go to Gemini. Let's go have
45:17 some fun with Gemini, shall we? All
45:20 right, Gemini Pro.
45:23 All right, what are we going to do here
45:25 that would resemble
45:30 research with your own files? Uploaded
45:32 files for more tailored research
45:35 results. Okay,
45:42 let's see.
45:46 um
45:49 research.
45:51 This is something that always
45:53 freaked me not didn't freak me out. This
45:55 was always something I couldn't actually
45:57 understand and I don't understand how
46:00 people who are not in the industry
46:01 understand this. So, here's what we're
46:03 going to research. I'm going to research
46:06 house paints.
46:10 Um,
46:15 and I want to understand
46:20 why
46:22 some are
46:24 so cheap. Well, no, some are
46:28 so much more expensive. I want you to
46:34 find
46:36 the sweet spot
46:39 of cost
46:41 versus quality and explain why.
46:46 Um, let's
46:48 uh
46:50 let's do this for latex
46:55 interior paints.
47:00 I want you to label
47:06 which ones are the garbage brands which
47:12 should be avoided
47:16 and why. And which
47:20 ones
47:22 are just
47:26 for
47:28 stupid
47:29 rich people
47:32 who can't
47:35 figure out
47:37 how to spend
47:40 their money.
47:43 So they spend
47:47 $600
47:50 on a gallon
47:52 of paint.
48:00 I got to tell you, when you're salty,
48:04 when you're when when you're a little
48:05 moist, the air conditioner's not
48:07 working, your AI's janky, you're a
48:10 little just pissed off at the world. Uh,
48:13 having a smarmy attitude is
48:15 entertaining, at least for me. Pate pin.
48:19 Kyle, you've done so much with canvas um
48:23 for little web apps. Wonder how that
48:26 compares.
48:30 Oh, here. Yeah. Yeah, we should we
48:32 should do Oh, that's actually a good
48:34 idea. Can I do both? No,
48:38 let's do the research first. You know
48:40 what we'll do, Pate? We'll do the deep
48:42 research. So, so the little Gemini
48:44 little Gemini uh party here. We'll do
48:48 the deep research
48:51 and then So, so let's see. I want you to
48:57 write the report to be accurate and
49:03 informative,
49:05 but
49:07 as if it were written by a hilarious,
49:15 bitter, stand up comic like Bill Burr or
49:22 Tom Sig.
49:23 Gura.
49:25 Uh or uh what's his name? Uh the [ __ ]
49:29 his name?
49:31 Uh I can see him.
49:34 Uh
49:36 oh my god, it keeps going out of my
49:37 head. In in out of my head. I I hate
49:39 getting old. Getting old sucks. By the
49:41 way, if you're um not old, um don't.
49:46 Okay. Um Okay, go. So, this is going to
49:49 go off and do that. Now, let's go see if
49:51 we've got a bread maker recommendation.
49:53 All right. Hey, Kyle. Yes. Uh, so two
49:56 things. One, um, you it's going to
49:58 prepare a plan, and you have to activate
50:00 that plan on
50:04 Oh, I have to I have to tell it. It's
50:06 going to think. Yeah, it's going to it's
50:08 going to come up with a plan. You can
50:10 edit that plan and before it goes off
50:12 and uh does its thing. Also,
50:16 uh it would take a little uh squirrel
50:20 moment here. Yes. Um back to uh your
50:23 side hustle as uh the appliance king of
50:26 Denver. Yes. Yes. I don't I don't know.
50:29 Not many not that many people knew that
50:31 that I I did work for an appliance
50:33 company, but you know, I guess it's
50:35 going to come out at some point.
50:38 Oh no. What did you do?
50:41 From A to I.
50:44 Embrace the jank.
50:49 From A to I.
50:52 Embrace the jank. Oh, good lord.
50:58 Where'd you do that? Is that That's V.
51:01 That's V3. Since it's a Gemini night,
51:04 did you do image to image to uh image to
51:08 video? No, it couldn't do image to
51:10 video. So I had chat GBT describe your
51:14 picture in explicit detail and then I
51:16 gave prompt that's in a regulars for
51:19 anybody who wants it. Okay. All right.
51:21 So our paint research is happening.
51:24 Great. And then yeah paint. So what
51:26 we'll do is we'll take the the the
51:28 research of this and we'll turn this
51:30 into a hilarious
51:32 web application
51:34 like a a a paint selector dashboard.
51:38 It'll be it'll be fantastic. But more
51:40 importantly,
51:42 um, wellbaked. What's his name? Forever
51:45 baked. Forever harsh weed.
51:49 Forever baked. Thought for 12 seconds.
51:51 Here's a side by side. Oh, those are the
51:53 two flux context. Okay, here we go.
51:56 Bread machines.
52:01 Tik Tok pin. my high appliance guy.
52:10 Okay. Top bread makers by price tier,
52:13 budget, mid-range, and high-end.
52:16 All right. Budget tier for under hundred
52:18 bucks. The Oster Express Bake 2 lb
52:22 baker. It's got loafsize flexibility,
52:25 which is one of the things I requested.
52:28 But now, wait, why does this not have
52:30 shopping?
52:32 Maybe I've got to do it. Um, let me go
52:35 back to GPT40.
52:38 Um,
52:41 give me a a set of shopping
52:47 links
52:49 for
52:51 the recended
52:56 or mented
52:57 bread makers.
53:00 Why did it not do that? Maybe because I
53:02 was in four 04 mini high.
53:06 Yeah. See, look what it does.
53:13 Wait, the mid-range favorite is $123.
53:19 Was that right up here?
53:26 The Oster Express bake
53:30 was the budget pick and it's more than
53:31 the Quez and Art Compact.
53:34 And then the Zo Jurishi
53:37 Bakery Virtuoso,
53:40 high-end performance. More bread for
53:42 your buck. That Quezin Art looks like a
53:44 [ __ ] deal.
53:48 But I don't know if you knew this, but I
53:50 guess I guess you can't do it with 04
53:52 mini high, but chat GPT40.
53:55 You get shopping
53:57 links.
53:59 You go buy [ __ ]
54:03 or you can't.
54:05 Can you?
54:08 No.
54:14 I I hate the internet. I ate it. By the
54:18 way, if you're new here and you're
54:20 trying to figure out, hey, is this AI
54:24 stuff anything? You've got a channel
54:27 called the AI Learning Lab. Maybe you
54:29 could give me an opinion on the value of
54:31 AI. Um, don't waste your time. It It's a
54:34 fad. It's a fad. It's like NFTTS. It's
54:37 It's It's going to go away soon. It's
54:39 going to go away soon. I I I don't like
54:42 it.
54:45 Can we get an irregulars discount from
54:48 the appliance king? No. No. If you're an
54:50 irregular, I charge you double for your
54:52 appliances.
54:59 Oh man. All right. What's going on over
55:02 here? Slow thinking
55:05 sites browsed. Oh, look at all the sites
55:08 it looked at for all the paint paint
55:11 comparisons.
55:13 What will they think of next? I stopped
55:16 using my coffee thermos because it kept
55:19 it too hot. Wild. Wow, that's wild.
55:24 Can I tell you something? I was given a
55:27 uh a Yeti
55:30 coffee thermos thing, like a you know,
55:32 like a 32 oz or something like that. No,
55:35 it was it wasn't that big. It was like a
55:37 24 oz.
55:40 And I I had it with me every day for
55:42 like I don't know three months. And then
55:44 one day it just I couldn't find it at
55:47 work. I couldn't find it at home. It's
55:49 gone. I don't know what I did with it.
55:52 So I I've been without a thermal.
56:00 Paint works at Google.
56:04 Let's see. Deconstructing paint
56:06 ingredients.
56:08 Connecting ingredients to quality and
56:10 cost. Brand analysis and price tiers.
56:14 Researching websites.
56:18 God damn, that's a lot of websites. It
56:20 looked at deep dive into paint
56:22 chemistry. This is what we need. You see
56:26 what I'm saying?
56:28 Because when I go into like
56:31 Walmart and I look at the paints, I'm
56:34 like, well, these paints are probably
56:36 fine. But like if you buy anything else
56:38 at Walmart, it's not fine. It's crap,
56:41 right?
56:42 But then you go to like the Home Depot
56:44 and you're looking at paints and they're
56:46 like, "That'll be $167 a gallon." And
56:50 you're like, "Who do I hit in the head
56:51 with the paint can?" No. No. I would I
56:55 would rather my [ __ ] just rot and fall
56:57 apart than spend $167 on stupid paint.
57:02 And so you go back to Walmart and you
57:04 get the one for $27. You're like, I'm
57:06 the man. And then you try to paint with
57:08 it and it's like painting with Elmer's
57:11 glue with with with paperiermâché
57:13 shredded into it.
57:17 I don't know. I don't know.
57:20 I'm also saying this like I have a lot
57:22 of paint projects.
57:24 I think
57:26 in the past 10 years
57:29 I painted the front porch of this house
57:31 and the back porch of this house two
57:33 years ago and it's already flaked off
57:36 cuz I don't know paint.
57:40 And
57:42 I feel like I should still be given a
57:44 [ __ ] award for that.
57:47 But no one cares. They don't care about
57:51 my plight,
57:53 my toil.
57:57 Create something new based on this, like
57:59 an audio overview, web app, or timeline.
58:02 Hey, Pate, this is pretty cool.
58:05 It rolled under your car seat, Kyle. It
58:07 actually might have rolled under my car
58:09 seat, Dr. J. Now that I think about it,
58:12 I have not looked there.
58:14 Oh, if Dr. J's got it. She's going to
58:16 get a She's going to get a a cookie.
58:20 But look what this did, Pate. This is
58:22 cool. It said It said, "Okay, here's
58:25 your stupid uh thing."
58:28 So, you want to paint a wall, a guide
58:30 for the confused and annoyed,
58:35 but it said build on top of this. So,
58:36 we're going to create we're going to
58:38 create a web page.
58:41 Okay. Describe your own app.
58:46 Um,
58:48 I want this to be
58:52 an interactive
58:55 web app
58:57 that allows
59:00 user to
59:04 first get the
59:07 highest quality paint for the
59:13 most reasonable
59:16 price
59:18 and then select
59:23 the color of their
59:27 dreams.
59:29 Boom. Okay, so that's going to go off
59:31 and write [ __ ] But I bet I'm going to
59:33 lose. Wait, let me export this quick.
59:35 Export to docs. Come on, you can do it.
59:41 This is add multitasking where where I
59:44 fired off one web request and then fired
59:47 off another one
59:49 because I didn't want to lose my first
59:51 one. Um, wow. This is a 17page report on
59:56 paint quality. Oh, most of its
59:58 citations. Okay, let's go. Let's go look
1:00:01 at this bad boy.
1:00:03 [Music]
1:00:05 A cynic's guide to not being fleeced.
1:00:14 things started lagging out. Missed some
1:00:16 of that. Oh, um I it just I was over in
1:00:21 in Gemini doing stuff. Okay, so now it's
1:00:23 writing code.
1:00:26 Oh, the no BS painter. Step one, buy the
1:00:29 right can of goo.
1:00:32 Um Benjamin Moore Regal Select 78 to 90
1:00:35 bucks. Sherwin Williams duration a solid
1:00:38 number two. Okay, wait. Benjamin Moore
1:00:40 Regal Select. That's it. The holy grail.
1:00:43 It has phenomenal coverage. It's tough
1:00:46 as nails and it won't make you want to
1:00:48 throw your roller out the window. It's
1:00:50 95% of the performance of the crazy
1:00:53 expensive stuff for a price that won't
1:00:56 require a second mortgage. Just buy
1:00:58 this. I'm convinced. Let's pick a color.
1:01:02 Oh, come on.
1:01:06 Did Why did you not work?
1:01:14 [Laughter]
1:01:19 The paint is great. The room looks bad.
1:01:21 It's officially your fault. No pressure.
1:01:23 Click here to see the wall of shame.
1:01:25 Seriously, avoid this crap.
1:01:28 Cheap big box store brands. Valpar,
1:01:31 Glidden, and Bear. You're buying water.
1:01:34 They skimp on the good stuff. Titan
1:01:36 titanium dioxide, 100% acrylic binders,
1:01:40 and load it with cheap fillers. The
1:01:42 result, you'll be painting three or four
1:01:44 or five coats, hating every second of
1:01:47 it. It won't send you money. And don't
1:01:49 buy contractor grade paint. It's chalky,
1:01:52 has zero durability, and it scuffs.
1:01:56 All right,
1:01:57 I'm going to say, "Hey, Gemini, none of
1:02:01 the buttons
1:02:05 work." And the color
1:02:08 picker
1:02:12 doesn't seem to work
1:02:17 either.
1:02:19 Um,
1:02:22 either I have
1:02:25 a senior
1:02:27 Google
1:02:29 tensor
1:02:33 chip optimizer
1:02:37 optimizer
1:02:39 watching
1:02:42 you code this
1:02:47 and if you mess
1:02:50 it up
1:02:52 he will likely not have his job much
1:02:58 longer.
1:03:00 Save Pate's job.
1:03:07 I'm doing a little I'm doing a little
1:03:09 prompting. I'm doing a little prompt
1:03:11 engineering.
1:03:13 It's it I'm doing prompt comedy.
1:03:17 So, there's prompt engineering. I'm
1:03:19 teaching a new methodology called prompt
1:03:20 comedy, which is yell at your GPT, tell
1:03:24 it it's a piece of poo, tell it's a dumb
1:03:27 dumb, um, threaten it that someone's
1:03:30 going to lose their job if they don't
1:03:32 get the application right, and let's
1:03:34 save Pate's job. I think it's fair. I
1:03:36 mean, Pate gives a lot to this
1:03:38 community. You're welcome, Pate. I don't
1:03:41 know if technically I'm senior staff or
1:03:43 not. I might be, but I'm not sure. Yeah.
1:03:45 Don't you don't you get some sort of
1:03:47 level? Like, aren't aren't you like a
1:03:49 level 17 or a level 12? Doesn't Google
1:03:52 have some sort of militaristic level
1:03:54 thing?
1:03:59 Oh man, a buddy of mine just got booted
1:04:01 from Google. Sucks.
1:04:04 Oh, let's see. I got called in from the
1:04:07 bar once. Um, okay. The no BS painter.
1:04:12 Does this work? No.
1:04:16 I'm convinced. Let's pick a color. Click
1:04:18 here to see the wall of shame. All
1:04:20 right. How do we do the color picker?
1:04:21 Oh, yeah. But that No, and that's all
1:04:25 it's all it's doing is putting Look at
1:04:28 Look what it's doing here. That's
1:04:30 [ __ ] hilarious.
1:04:33 You pick a color and it just it just
1:04:35 colors the whole photo. It doesn't
1:04:37 actually change out the wall.
1:04:44 Oh yeah, janky is as janky does. But it
1:04:47 did make us a web app. Uh,
1:04:52 actually, you know what? Let's go. Let's
1:04:54 go try something. Let's go. Say,
1:04:57 um,
1:04:59 let's go. Listen, dum dum.
1:05:02 Listen, dum dum.
1:05:05 Um,
1:05:07 you only have
1:05:09 two recommended
1:05:13 paints.
1:05:15 Can't you
1:05:18 go research the actual
1:05:23 colors I can buy
1:05:28 and
1:05:30 then have me select
1:05:34 one or the or or or
1:05:39 the other paint and
1:05:42 and show me the colors
1:05:49 from the real paint store.
1:05:53 Also
1:05:56 just shifting
1:05:59 the
1:06:02 hue
1:06:04 on the fake wait the hue
1:06:08 on the fake living room is creative but
1:06:16 a bad solution.
1:06:22 Oh, you could have made an audio
1:06:25 overview for it. Deep research. Oh,
1:06:27 yeah. I wonder if I can go back. I
1:06:28 probably can. So, if I go So, that's
1:06:31 going to be doing its thing. Paint cost
1:06:34 versus quality.
1:06:38 Oh, yeah. It's up here. Is the canvas.
1:06:40 Okay. All right. So, now it's gonna
1:06:44 Okay. Let's see.
1:06:48 Hey, Paint. I found a bug.
1:06:51 When it's writing code, you can't scroll
1:06:54 up in your chat window. Could you tell
1:06:57 the people that make this Gemini thing
1:06:59 to fix that? Pate be nice if it were
1:07:01 done by like Thursday.
1:07:05 Kyle needs to build a date picker that
1:07:08 shows the person's true colors.
1:07:14 Oh, good lord.
1:07:17 All right. Step one, pick your paint.
1:07:28 The toy Toyota Camry of paint. It's
1:07:30 fine. It won't let you down, but it
1:07:31 won't excite you. The supercar of
1:07:34 paints. Emerald. Fantastic.
1:07:40 Yeah, that's not working. All right,
1:07:41 let's go back up here. Let's go back to
1:07:44 our canvas.
1:07:47 Is this the canvas?
1:07:50 Ready in a few minutes. Start research.
1:07:52 Open.
1:07:54 All right. Export. No.
1:07:58 Create audio overview.
1:08:01 Actually, this audio overview should be
1:08:03 really good. Super late. What did I
1:08:06 miss? Oh, Gareth. It has been an epic
1:08:09 night of
1:08:11 me
1:08:13 not only failing miserably, but failing
1:08:16 early and often and having no patience
1:08:21 for failing.
1:08:23 So, it's really just a it's like a
1:08:26 master class in whining.
1:08:32 Yeah.
1:08:37 is pretty much what it is.
1:08:41 I love you and I want Why Why does she
1:08:44 look so weird?
1:08:47 All right. I've been playing around with
1:08:49 Claude's
1:08:51 basically vibe feature making silly
1:08:53 things. Yeah, that's pretty cool. Yeah,
1:08:55 Claude's Claude's pretty good. Chat GBT
1:08:57 is pretty good. It looks like Gemini is
1:08:58 pretty good at this stuff.
1:09:01 I don't What model was I just using in
1:09:03 Gemini? I was using 2.5 Pro.
1:09:08 Math and code. Flash is fast for
1:09:10 allaround help.
1:09:12 All right.
1:09:16 This thing didn't work. Like none of
1:09:18 these hyperlinks actually hyperlink
1:09:20 anywhere.
1:09:22 The little buying links. Nothing there.
1:09:24 Can't go buy.
1:09:27 Ask about this.
1:09:30 Yeah, that's not good.
1:09:35 Danielle made a picture of me that then
1:09:38 look Danielle did this. This is what you
1:09:41 missed, Gareth.
1:09:45 [Laughter]
1:09:49 So, same as most nights. Exactly.
1:09:52 Exactly. It's just a normal night here
1:09:54 at the AI learning lab.
1:10:01 Oh, refresh your regulars, please. Oh
1:10:04 no. What have you done? What have What
1:10:06 are you people doing to me?
1:10:09 You're soiling my reputation.
1:10:14 Who is that? Wait. Made with Flux
1:10:16 Context Pro. I don't know this man.
1:10:24 There's Danielle out on her out on her
1:10:26 third date with a guy. She can't
1:10:29 remember his name, but she doesn't want
1:10:31 to admit it.
1:10:33 [Music]
1:10:37 Hey, Kyle. Yeah. While you're here, if
1:10:40 you go down one post Uhhuh. Um, I made
1:10:43 you a little something. Uh, wait. Scroll
1:10:46 up. Appliance king. No, no, two more.
1:10:49 Nope, not that one. The next one up.
1:10:51 That one. Yeah.
1:10:59 Don't Don't just play it. There's more
1:11:02 after suno had to get I had to get some
1:11:05 suno action in on this. We have
1:11:07 appliances from a to I
1:11:12 embrace the janky.
1:11:17 [Music]
1:11:27 Okay. Y'all, y'all really, y'all really
1:11:30 need to take some vitamins tomorrow.
1:11:33 This is Listen, everyone's blood sugar
1:11:35 is low. All right, everyone's blood
1:11:38 sugars low. Let's There will be no
1:11:40 judgment tonight. There'll be no
1:11:42 judgment tonight, but there might be
1:11:43 lawsuits tomorrow. Okay,
1:11:51 that flux is about to end some
1:11:53 relationships. Yeah, really right. Oh my
1:11:56 god.
1:11:58 Oh. Oh, good lord. Where where were we?
1:12:03 Wasn't I somewhere? Was Oh, I was at
1:12:05 Gemini. Oh, look. Okay, we have a six
1:12:08 minute. This should be good. Let me let
1:12:10 me change out my my tabs because I'm a
1:12:13 responsible tab sharing member of
1:12:15 society. Unlike Brandon who never tells
1:12:19 me to do this, I always have to remember
1:12:20 it myself. Okay, here we go. Okay. Have
1:12:24 you ever stood there in the paint aisle
1:12:26 just completely lost? I mean, rows and
1:12:29 rows, all these claims, prices from like
1:12:31 $25 up to what, $150 a can. You feel
1:12:35 like you're about to get totally
1:12:36 fleeced. You just want paint, but it
1:12:38 feels way too complicated. So, this deep
1:12:41 dive, we're going to rip the lid off the
1:12:43 paint can, so to speak. We'll break down
1:12:45 the uh the grift, maybe, and figure out
1:12:48 what really matters when you buy paint.
1:12:50 Okay, let's unpack this. Right. And the
1:12:52 first thing to get is that pretty much
1:12:54 all latex paint, the cheap stuff, the
1:12:56 expensive stuff. It's got four basic
1:12:57 things inside. You've got pigments,
1:12:59 binders, liquid, and additives. That's
1:13:02 it. The huge difference in price, and
1:13:04 yeah, how well it actually works? That
1:13:06 all comes down to the quality and the
1:13:07 amount of those ingredients. Okay. So,
1:13:09 what are those key things? What makes
1:13:11 one paint actually worth, you know, the
1:13:13 extra money? Where do we start? Let's
1:13:15 start with pigments. Pigments give paint
1:13:17 its color, obviously, but also its hide.
1:13:20 That's its power to cover up whatever
1:13:22 color was there before. And the absolute
1:13:24 king of hide is titanium dioxide. Uh
1:13:27 TO22. It's just brilliant at scattering
1:13:30 light. Gives you that bright white base
1:13:32 and fantastic coverage. Ah okay. So the
1:13:35 cheaper paints
1:13:37 they skimp on that. Is that the main
1:13:39 trick? That's a huge part of it. Yeah.
1:13:41 Cheaper paints will often swap out that
1:13:43 pricey TIO2 tutors for things called
1:13:46 extender pigments or fillers. think um
1:13:49 ground up limestone, things like that.
1:13:51 It bulks up the paint, but it doesn't
1:13:52 really cover it. Makes it more
1:13:54 transparent, kind of watery. Right.
1:13:56 That's why sometimes you buy cheap paint
1:13:57 and it feels like well, like you're
1:13:59 painting with slightly colored water.
1:14:01 End up doing three, four, sometimes even
1:14:03 five coats just to hide the old color.
1:14:05 Exactly. It's incredibly frustrating.
1:14:06 And then there's the binder. That's
1:14:08 okay. One of the things it didn't get
1:14:10 and I didn't actually, you know what?
1:14:12 Let's try something. Create
1:14:16 audio overview. Let's see. Describe your
1:14:18 own app.
1:14:21 Can I not describe? Let me let me see.
1:14:23 Um,
1:14:26 make this audio overview like
1:14:31 two bitter
1:14:33 comedians
1:14:36 on
1:14:38 a podcast.
1:14:41 Audio overview. I wonder if it took that
1:14:44 prompt.
1:14:49 You can leave this chat in the meantime.
1:14:51 All right.
1:15:06 No, I know. But normally with audio
1:15:09 overviews, like if you're in Notebook
1:15:11 LM,
1:15:13 um, and you create an audio overview,
1:15:15 you can give it you can give it
1:15:16 direction. So I assume they let me do
1:15:18 that here. But if not, probably not.
1:15:20 They probably didn't because tonight
1:15:24 like the the thing the way it should
1:15:26 work, it's not working. Why? Because
1:15:28 Google chose not to have some bitter Gen
1:15:32 Xer sitting in the corner office that
1:15:34 just looks at the apps and goes, "Uh,
1:15:36 no.
1:15:37 No, no, no, no. This No, this can't go
1:15:41 out. You You got to be able to fix click
1:15:44 the thing to put in the description of
1:15:47 the audio podcast like we do over in
1:15:49 Notebook LM. Have you seen the other
1:15:51 product? Oh, that's a different
1:15:52 division. Yeah. Not to your customers.
1:15:56 It's not. Okay. They're going to be
1:15:58 looking at these things and expecting
1:15:59 them to behave similarly. They don't
1:16:02 give a [ __ ] about your corporate
1:16:03 boundaries, your level 17 seniority
1:16:07 matrix flat hierarchy report
1:16:12 structure.
1:16:14 They just want to know, can I get my
1:16:16 podcast to sound like
1:16:19 comedy?
1:16:21 No. No. All right. Well, I'm going to go
1:16:24 to bed then.
1:16:31 I don't know what to tell you people.
1:16:33 Listen, listen. I know you came here
1:16:36 expecting insights and and brilliance
1:16:39 quite frankly.
1:16:43 I really wondered how silent they are. I
1:16:45 would lose my mind. Yeah, I know. I
1:16:48 know. Corporate corporate structures are
1:16:51 weird, man. They're weird. I do I will
1:16:55 say this. I know I don't usually say
1:16:57 good things about Google, but but you
1:17:00 know, because Pate was here, at least. I
1:17:01 don't know if he's still here. Um,
1:17:04 I will say that
1:17:07 Google in the past two months
1:17:10 are starting to tighten up some of their
1:17:12 their product chaos, right? They're
1:17:14 they're starting to migrate features
1:17:16 into Gemini, and they're starting to
1:17:18 actually not only announce really cool
1:17:21 things, but launch them that we can go
1:17:22 play with. That's new from them. The
1:17:25 past two years have been really really
1:17:27 shitty. Their their product strategy has
1:17:29 been abysmal, but it's getting it's
1:17:31 getting uh it does seem to be improving.
1:17:33 So, someone over there got a promotion
1:17:36 and said, "Uh, hey, let's get everyone
1:17:38 in the same room here and you know, just
1:17:41 go around the room. There's 674 people
1:17:44 in the room." And just everyone say what
1:17:46 project you're working on. I'm working
1:17:48 on Gemini 2.5 Flex Light Pro Max. I'm
1:17:53 working on Gemini 6.2 2.3.4.5
1:17:56 May of 2024 minor
1:18:01 and then you know someone was in there
1:18:03 and they're like okay you guys go work
1:18:04 together just turn that into one thing
1:18:07 you guys go put that in Gemini oh we
1:18:10 weren't told to do that well you're
1:18:12 being told now just go just
1:18:21 it was paint promote hate. Oh, I'm going
1:18:24 to experiment with putting onboarding
1:18:26 into Notebook LM podcast. Oh, that's
1:18:28 actually a great idea. That's a great
1:18:31 idea. Source Camp. Source Camp, one
1:18:34 piece of advice with Notebook LM. Um,
1:18:40 just generate 10 of them and download
1:18:42 them all because sometimes Notebook LM
1:18:46 will do like a five minute one,
1:18:47 sometimes it's like 15 minutes.
1:18:49 Sometimes it completely misses the
1:18:51 point. So just like generate a bunch of
1:18:54 them, but download them all. And what
1:18:56 you'll what you may end up doing what we
1:18:58 did with the Sydney podcast was I think
1:19:02 I generated nine of them and then I
1:19:05 ended up editing like the beginning of
1:19:07 one, the middle of the other, and the
1:19:09 end of the other one. So it's a it's a
1:19:11 Frankensteined thing that was put
1:19:13 together.
1:19:16 Oh man,
1:19:18 I can always count on a good laugh when
1:19:20 I'm here. And these days I need that.
1:19:22 Nice. So for me, tonight was a win. Oh,
1:19:24 that's good. Whether you intended it or
1:19:26 not. Yeah. Sometimes it's just it's just
1:19:29 uh everything's so ridiculous that that
1:19:32 There you go. All right. Here. Look.
1:19:34 Paint lies. Paint lies and dollar signs.
1:19:37 Let's see if this gave us a comedy
1:19:40 overview. Okay. Have you ever found
1:19:43 yourself just like standing frozen in
1:19:45 the paint aisle, totally overwhelmed?
1:19:47 Oh, yeah. You see hundreds of colors,
1:19:49 all these confusing terms, and uh prices
1:19:52 that go from maybe $25 all the way up to
1:19:56 $150 a can. It makes you feel like
1:19:58 you're about to get fleeced. Definitely.
1:20:00 You just want to paint a room. Exactly.
1:20:01 You just want to cover a wall, but it
1:20:03 feels like you need a PhD to figure out
1:20:05 what you're actually buying. Well, in
1:20:07 this deep dive, we're going to rip the
1:20:08 lid off the paint industry. We want to
1:20:10 deconstruct the um the grift, if you
1:20:14 will, and show you what really matters.
1:20:16 Okay, let's unpack this, right? And the
1:20:18 first thing to get your head around is
1:20:19 that pretty much all latex paint. Yeah,
1:20:22 the common stuff has four basic things
1:20:24 in it. Pigments, binders, liquid, and
1:20:27 additive. Four things, you know. All
1:20:30 right. So, listen. I'm going to go. I'm
1:20:32 going to get out of here. But here's
1:20:33 what I learned tonight. I learned a
1:20:35 couple of things. Um,
1:20:40 deep research is really cool. If you
1:20:43 haven't been playing with deep research,
1:20:44 just even a web search is pretty cool.
1:20:47 But like the ability to go out and yes,
1:20:50 you could Google how do they price
1:20:52 paints and how do you tell a good paint
1:20:54 for a bad paint, but you'd have to read
1:20:56 that [ __ ] And just the idea of like
1:20:58 just being able to not know anything
1:21:00 about something and now I know know
1:21:02 something about it is pretty [ __ ]
1:21:03 cool. So that was good.
1:21:05 um flux context does look like it's
1:21:09 pretty remarkable, but it requires a lot
1:21:12 of thought and prompting. And so my goal
1:21:15 for 2025 use not using my brain, I we're
1:21:18 going to have to do we're going to have
1:21:20 to do a a flux context
1:21:23 chat GPT prompting session to get some
1:21:27 prompt structures to take over to it
1:21:31 doesn't matter. I I know what I need to
1:21:33 do. All right, I learned some things
1:21:35 tonight. And then, you know, because I
1:21:38 was frustrated, maybe somebody left.
1:21:40 Okay, tomorrow
1:21:42 I got a podcast with Ann Murphy and
1:21:44 myself. We don't have a guest tomorrow,
1:21:46 but Ann and I both are working on new
1:21:50 frameworks. Mine is the feed your prompt
1:21:52 framework, and I think she's got one,
1:21:54 too. So, tomorrow's podcast is just
1:21:56 going to be Ann and I talking and kind
1:21:58 of reacting to each other's uh
1:22:00 frameworks, which is good because I will
1:22:03 absolutely heckle her, right? She's
1:22:06 going to be putting it out there like
1:22:08 here's my thing and here's my thought
1:22:09 behind it. And I I think what I'm just
1:22:12 going to do is like midway through it
1:22:13 just go.
1:22:16 Right. And then like just see if she
1:22:17 gets flustered.
1:22:20 And then at some point, I think what
1:22:22 I'll do is I'll explain her framework
1:22:25 back to her in a way like she couldn't
1:22:29 possibly understand what she just said,
1:22:32 right? Like don't you think that would
1:22:34 be kind of supportive of her to like
1:22:36 really explain it to her what she just
1:22:39 told me? I like that's just that's just
1:22:41 my instinct right now. So, I think
1:22:43 interrupt her with with negative
1:22:45 sentiment, you know, while she's sharing
1:22:47 it and then and then explain it back to
1:22:50 her as if she didn't understand it. So,
1:22:53 that that should be like 45 minutes and
1:22:57 then I'll I'll share mine. All right.
1:23:01 Okay. Fantastic. So, that's at 400 p.m.
1:23:03 Mountain time. That's 3 p.m. Oregon time
1:23:05 where Ann is, where Ann will be crying.
1:23:08 Um,
1:23:10 and yeah, it'll be good. And then what
1:23:12 else? Oh, AI Life Hacks is also
1:23:15 tomorrow. 7:30 Eastern time tomorrow. So
1:23:18 5:30 Mountain, 4:30 Pacific, AI Life
1:23:22 Hacks, go to the AI salon, go to events,
1:23:25 and then
1:23:29 I bailed on the appliance king song. And
1:23:31 it gets better 30 seconds in. I can't do
1:23:33 it. I don't My my my ego is too fragile
1:23:37 tonight.
1:23:39 Oh, it's only a 30 second song. Okay.
1:23:44 Oh, lordy.
1:23:48 Hang on. I gotta go. I gotta go. Uh,
1:23:52 whatever you call it. Change my tabs
1:23:54 again.
1:23:57 [Music]
1:24:04 Who's they? Who's who's sponsoring
1:24:07 tomorrow? Oh, the Denver appliance.
1:24:15 We have appliances from A to I.
1:24:19 Embrace the jank.
1:24:23 Need a fridge or a bread maker.
1:24:27 Kyle Shannon's got the deals for you.
1:24:30 From a to washers that singer'll
1:24:36 [Music]
1:24:50 adore.
1:24:51 [Music]
1:24:55 We should have a Kyle roast. I'm pretty
1:24:57 sure we're in the middle of it.
1:25:08 Okay. All right, people. Well, can't say
1:25:12 can't say we didn't all have fun
1:25:13 tonight, can we? Well, no, we can we can
1:25:16 we can absolutely say we didn't have fun
1:25:18 tonight. We can we can definitely say
1:25:21 that. Oh my god. Paint pin. Uh the paint
1:25:26 pin's gone.
1:25:28 Yeah, we have a Kyle roast five nights a
1:25:30 week. Yeah, that's called the Tik Tok
1:25:31 channel. AI was a mistake. Exactly,
1:25:33 Pate. That's I'm I'm with you there
1:25:35 tonight. I think I think what what we
1:25:38 can all conclude tonight is that AI is a
1:25:42 fad. It's not going to change anything.
1:25:45 Um it will not it will not stop
1:25:47 hallucinating. Gary Marcus was right.
1:25:50 And just this [ __ ] doesn't work. All
1:25:52 right. Anyway, good luck with whatever
1:25:55 you figure out what to do next. Uh, you
1:25:58 know, maybe it'll be I I don't know. Uh,
1:26:01 uh, hey, you know what could come back?
1:26:02 Augmented reality, you know, uh, 3D
1:26:05 worlds in glasses, but mixed with the
1:26:07 real world, right? Not the metaverse,
1:26:10 the the augmented verse, right? I think
1:26:13 that could be big.
1:26:15 That could be big. I think I think uh
1:26:18 going back to um uh data entry, manual
1:26:23 data entry, um faxes are coming back.
1:26:27 So pretty excited about the future now.
1:26:30 And by the way, your job's fine. Your
1:26:32 jobs are going to be good because this
1:26:34 is nobody's using AI.
1:26:39 Just not really usable. All right,
1:26:41 embrace the jank. That was a lie. That
1:26:44 was a lie. That was a lie. The the
1:26:45 jank's not getting better. All right.
1:26:48 Hey, uh boy, really been fun. All right.
1:26:53 Good night.
1:26:58 Second life. Yeah, exactly. AI is second
1:27:01 life. Second life is second. Second life
1:27:04 is the future.
1:27:06 I can't wait for hand cramps from using
1:27:08 pens. Pens and papers again.
1:27:13 All right. Uh peace out. YouTube and
1:27:15 Twitter and LinkedIn. Uh, I hope you
1:27:18 enjoyed this. There's only 22 of you.
1:27:19 Like, no one came tonight. No one cares.
1:27:21 I mean, you came, right? You came and
1:27:24 you stayed. And that should indicate a
1:27:27 lot,
1:27:29 right? You might want to you might want
1:27:31 to do some self-reflection there. Like,
1:27:33 what is it about you why you're here?
1:27:37 Right.
1:27:38 After tonight, I think that's a good
1:27:40 question to ask. All right. Bye.