6/18/2025 - Exploring Midjourney's New AI Video Generation Feature

Description

Chapters

Transcript

0:05 [Music]
0:09 Champ, are you ready?
0:12 Are you ready, big boy? Are you ready
0:14 for the singing? Come on, boy. Get over
0:17 here.
0:19 [Music]
0:33 She came on
0:37 [Music]
0:40 his feel.
0:47 She sat on a stool and he said, "What do
0:51 you want?"
0:54 She said, "Give me a love that doesn't
0:57 freeze up."
0:59 [Music]
1:06 Said, "I have melted some in my time,
1:10 dear.
1:13 But to sit next to you, well, I shiver
1:16 and shake.
1:21 And if I knew love, well, I don't think
1:24 I'd be here.
1:28 Asking myself if I've got what it takes
1:35 to melt your icy blue heart.
1:39 [Music]
1:43 Should I start
1:45 turning
1:47 for years?
1:52 into a river.
1:54 [Music]
2:11 What do you think, champ? What do you
2:13 think, buddy? You good? Why you so
2:16 you're you're a little ramy tonight?
2:17 What's going on?
2:25 [Music]
2:38 sitting in this lonely town. Yeah.
2:42 Wonder when things are going to change.
2:48 [Music]
2:50 Dream my life away.
2:53 Seems these dreams have turned to a
2:56 bunch of glass.
2:58 [Music]
3:00 Get my love up, but my past is pulling
3:04 me down.
3:06 [Music]
3:08 Wondering how long
3:11 this black sheep will stick around.
3:19 Somebody told me once before, you can
3:23 never go home again.
3:26 Won't you leave?
3:28 Santa things to ste me away. Yeah. From
3:33 the truth of who I am and what I
3:36 believe. So I thanked him for his two
3:39 cents with a handshake
3:41 and some sympathy. Yeah.
3:46 packed on my blue jeans and headed for
3:50 that big prize
3:53 of my freedom.
3:57 Bye-bye,
3:58 black sheep. To the black sheep of the
4:02 family,
4:05 you're not singing anymore, champ.
4:07 Bye-bye.
4:11 Oh, seems so very much to me.
4:15 Bye-bye,
4:17 black sheep to my friends and my family.
4:24 Bye-bye.
4:26 How going to set my soul
4:31 set it free?
4:33 [Music]
4:40 Oh yes. How is everybody? What is going
4:45 down? What's going up? What's shaking?
4:47 What's sitting still?
4:49 What do you think is sitting still, but
4:51 it's actually shaking?
4:53 That happens.
5:04 Oh, good lord. Good lordy. Lordy lordy.
5:08 All right. So, I know what we're playing
5:10 with tonight, which is good.
5:14 You guys ready to play? We're gonna
5:15 play. We're gonna do a little playing.
5:19 I did another test run of feed your
5:22 prompt on the
5:25 AI readiness project,
5:28 but I was midway between a rewrite. And
5:30 here's what I learned. Don't share
5:32 something that's not written.
5:37 I didn't know what the [ __ ] I was
5:39 saying.
5:41 Have you checked out AI 2027? I have
5:43 not. Oh, wait. Was that out like a month
5:46 or two ago? I think I have checked that
5:48 out. Um, I thought that was something
5:50 that came out today.
5:52 AI 2027. Yeah, that's the sort of the
5:56 shit's coming, the robots are going to
5:57 eat us. Is it that one? The geopolitics
6:00 of super intelligence.
6:04 We predict pred predict the impact of
6:06 super human AI over the next decade will
6:08 be enormous exceeding that of the
6:09 industrial revolution. Yeah. Mid 2025
6:12 stumbling agents. That's right where we
6:14 are.
6:18 Early 2026 coding automation
6:23 sounds about right. We've got shitty
6:26 versions of that now. Mid 2026, China
6:29 wakes up. depends what they're talking
6:31 about with China because some of the
6:33 [ __ ] China's doing is already better
6:35 than what we're doing in the creative
6:37 spaces certainly
6:39 or at least as good. Um,
6:42 late 2026 AI takes some jobs. That's an
6:46 interesting one.
6:50 January 2027, agent 2 never finishes
6:54 learning. That we seem to be on track
6:55 for that.
6:58 China steals agent too.
7:03 Then we get into monthby-month
7:05 breakthroughs of once once we hit super
7:07 intelligence. Yeah, this is a cool
7:09 article. The infographic that that
7:11 dynamically uh does its thing is pretty
7:14 cool, too. Let's see if Brandon's here
7:16 yet. No, Brandon's going to be late.
7:18 He's got weather coming through Ohio. I
7:21 watched the first OpenAI podcast
7:22 YouTube. That was cool. Um, was that the
7:25 one Jeff? That is uh it was Sam Alman's
7:28 brother hosting it. Um yeah, I thought
7:31 it was quite good. I thought I thought
7:33 the best part of it was the opening
7:35 three minutes where they're just acting
7:36 like brothers. It was really good. It
7:39 was really good. Um
7:42 it was a little boring. So was that was
7:45 that an Open AI production? If it was,
7:47 that explains why it was boring because
7:50 his brother just asked him softball
7:52 questions.
7:53 Posted today. Posted today. Huh?
7:57 I don't know that one. I do not know
8:00 that one. YouTube.
8:02 Dear YouTube.
8:05 Was that the one with his brother? Oh,
8:08 no. Wait.
8:13 No.
8:22 All right. Yeah, we could we could
8:24 listen to that.
8:27 On brand.
8:29 Severe weather has passed. Now I'm
8:31 trying to get the kiddos settled.
8:33 Beautiful.
8:35 All right. So,
8:37 Well, it wasn't his brother. No, no, no.
8:39 Wait, it wasn't his brother. No, no, no.
8:42 There's a there's a an interview with
8:44 Sam Alman's brother from like three days
8:46 ago. Two days ago. I don't know. Maybe
8:48 yesterday. Is that two days ago? No,
8:51 that's yesterday. two days ago would be
8:53 a day before yesterday, right? Yeah. So,
8:58 sometime in some amount of days, there's
9:00 a video that looks like it's in a very
9:02 similar setting
9:05 with Sam Alman's brother and him. And
9:07 then this is different.
9:10 Uh, it's a 40-minute interview. Yeah,
9:13 the one with his brother's about 40
9:14 minutes, too. I saw the first and second
9:18 videos. Well, well, well. Ain't that
9:21 special? So, but is this Let's see. Open
9:25 AAI.
9:31 Yeah, that's not that one.
9:36 So, I'll go Sam alman brother.
9:42 [Music]
9:47 That's two years ago. Oh, he was
9:48 interviewed by his brother.
9:51 And then where is it? Come on, Sam.
9:55 What's his brother's name?
9:58 Jack Alman.
10:01 Jack. Whoops.
10:16 All right.
10:18 Share your screen, Kyle. I Yeah, I know.
10:20 I'm just getting [ __ ] together. I'm just
10:22 getting [ __ ] together. I don't know if I
10:25 want to share I don't know if I want to
10:26 look at this stuff yet because it's a
10:28 little boring.
10:30 Now I'm singing yesterday in my head.
10:32 Yesterday,
10:33 all my troubles seem so far away.
10:41 [Music]
10:44 Is Jack working today? I invited a nice
10:46 lady. She may be connecting soon.
10:48 Fantastic, Mr. Bring your friends. Bring
10:52 your neighbors. Come on. Come all to the
10:53 AI learning lab where learning
10:56 occasionally takes place, but not in the
10:59 ways intended.
11:07 Um,
11:09 so the thing we're going to play with
11:11 today is MidJourney dropped their video
11:13 model. It's out. You can go make videos
11:15 on MidJourney right now. And if you just
11:18 go there, you're not going to see it.
11:20 And you're going to be like, "Where's
11:22 the video model? How do I make one of
11:24 these here videos? This doesn't make any
11:26 sense to me." So, we're gonna I'm going
11:28 to show you how to do that.
11:31 Um, it looks to be pretty flipping good
11:35 from what I can see.
11:38 26,000 coming tonight. DQ Blizzard. Hi
11:42 Dave.
11:50 All right.
11:51 What are we gonna do? Who's got
11:53 questions? So, if you're new here,
11:56 by the way, I'm still getting weird.
11:58 Steo,
12:00 every time he likes something, it
12:02 displays. Ah, Mark viewer, prioritize
12:06 comments. Steo, you've been
12:08 deprioritized.
12:09 [Laughter]
12:12 All right.
12:14 Not that I don't love you. I do love
12:16 you, but I think it was a setting on my
12:20 side is why you were showing up
12:22 everywhere.
12:24 I might need to stop the live to get it
12:26 to change the setting. All right. So,
12:28 we'll just see a lot of Steo comments
12:29 likes tonight. You can keep liking,
12:31 Steve. It's fine. I see the other
12:33 comments in between them. Um Oh, the
12:36 videos are so good. All right. Let's
12:38 see. Let's see. I think my friend Dave
12:40 So, why don't we do this? We'll watch
12:42 the first We'll watch the first five
12:45 minutes or so of the one with his
12:47 brother
12:49 or until I get bored. Maybe we go longer
12:52 than five minutes. Maybe it's actually
12:53 compelling enough. Maybe I was watching
12:55 it in some sort of bad way and then
12:59 we'll go look at the new one that
13:00 dropped today. All right.
13:04 I knew it. I'm the last priority. Yeah,
13:06 Mr. We can't see anything you say. We're
13:10 just guessing.
13:14 Open AI drop transcription for meetings
13:17 for plus users.
13:20 Really?
13:22 Is it where is it?
13:27 New chat.
13:30 I'm a plus user.
13:33 Tools.
13:35 No.
13:38 Settings.
13:43 Where is it, Danielle? I've spent 410
13:47 hours in the learning lab. Wow.
13:52 Well, if it makes you feel any better,
13:54 Kak, I've got about 1,250 hours of
13:58 content up on YouTube. So, so I've got
14:01 you beat.
14:07 Oh my god. Lol. Low professional
14:09 background.
14:13 [Music]
14:25 [Music]
14:28 Rowdy Irwin. Rowdy's not liking the set.
14:30 What is your background? That's my
14:34 Listen, here's the deal. Those of you,
14:38 those of you trolling,
14:42 thinking like, "If this is supposed to
14:44 be some kind of learning lab, that sure
14:46 looks to me like a shitty home office
14:49 decorated by someone with ADD
14:54 who's got a hoarding complex." Okay,
14:57 that's what it looks like, right? What
14:59 you don't see on the other side of the
15:01 camera is the $20 million production
15:03 studio,
15:04 right? This was put together by a team
15:06 of 20 to make it look like a shitty home
15:09 office. So there, there you go.
15:12 Fantastic. Bob, tell him what he's won.
15:14 He had one crap. He thought it was a
15:17 shitty home office.
15:19 All right.
15:21 [Laughter]
15:23 No, man. I like it. It's Listen, I got
15:26 booze. I got I got 3D prints of of cars.
15:32 I got a a rabbit.
15:34 Rabbit R1. Not that kind of rabbit,
15:37 you weirdos.
15:39 I got a picture of the wife. I got look
15:44 a trophy from Super Bowl 12 when the
15:48 Orange Crush I I think they lost.
15:55 Got that at a yard sale. So, don't think
15:57 I don't have anything here. This not
15:59 This is not
16:02 amateur hour
16:05 and it's like you've got someone here
16:09 that if you ask for my qualifications
16:13 I could get out the certificates.
16:21 Is he glitching for anyone else? No, I
16:23 think that's just my personality.
16:24 Danielle.
16:31 Um, all right. So, we're going to watch
16:34 some movies. We're going to do movie
16:35 night. So, this is just like grade
16:36 school. Okay, kids. You know when the
16:38 when your teacher was hung over in grade
16:41 school? And she was like, "I've got a
16:43 surprise for you. It's movie day." And
16:46 then she put on the movie, turned out
16:47 the lights, and you'd hear her snoring
16:49 five minutes later. We're gonna We're
16:50 going to do that tonight.
16:53 I see a 69 Camaro. Exactly.
16:57 That was from That's from Vicki.
17:00 That judge is from Mr. It
17:03 Pikachu I got at the yard sale.
17:06 Uh the empty bottle of Bllandons. That's
17:09 from Serena and the Irregulars.
17:15 I printed Groot. I'm in that copy of
17:19 Wired magazine.
17:21 Bet you didn't know that. Want to see?
17:28 Want to see what I looked like when I
17:30 was younger and not fat? Well, less fat.
17:33 [Music]
17:42 So this the issue Richard Dawkins
17:53 I guess I probably should have had this
17:54 thing uh bookmarked.
18:00 What did you do the other night? I
18:01 watched the guy try to find a picture of
18:03 himself in a magazine.
18:11 Look at that. Huh?
18:14 Wired magazine, baby.
18:19 And there's my there's my little Urban
18:22 Desires magazine right there. There's a
18:25 little article. There's me looking all
18:27 [ __ ] hip, huh?
18:31 You see what you're dealing with here.
18:41 All right. That's my my one
18:43 qualification. I was in Myired magazine
18:45 once.
18:50 Best practice go into chat GBT and give
18:52 give you a list of all the work you have
18:56 done together that can be used as
18:57 professional accomplishments. That's
18:59 actually a great idea. An oldfashioned
19:01 magazine. My grandma told me about
19:04 those. I like it. I like our trolls.
19:11 Oh man. Impressive.
19:15 All right.
19:21 All right. Let's go. Let's go watch some
19:23 movies. I'll get drunk. We'll turn out
19:25 the lights. I'll get drunk.
19:35 Excuse me.
19:39 Anybody
19:42 have any questions before we get
19:43 started? Anyone want to run me down a
19:45 rabbit hole? If you're new here, my name
19:47 is Kyle Shannon. This the AI learning
19:49 lab. What we do here, it's the nickname
19:52 of the channel is Chat Add.
19:55 Chances are low that I'll ever finish a
19:57 sentence. Chances are low that I'll ever
19:59 do what I say I'm going to do. Um,
20:02 occasionally something interesting
20:04 happens here. But here's the deal.
20:06 Here's why you're here. Anyone
20:12 that tells you they're an expert in AI
20:14 is lying in generative AI. There's going
20:17 to be there's there's experts in like
20:19 building the models and in old school
20:21 AI, but generative AI, there are no
20:24 experts right now because the shit's
20:26 changing too fast. We don't know how to
20:28 use them. We don't know what AGI is. We
20:31 don't know what super intelligence is.
20:32 We can't handle the [ __ ] we have in our
20:34 hands now.
20:36 But
20:38 being in the conversation and being in a
20:40 community of curious people that are
20:42 trying to figure this stuff out, I think
20:45 is one of the most important things you
20:47 could do with your life probably for the
20:49 next three years, probably for the next
20:51 five years. PWC just came out with a
20:54 report today or no yesterday uh talking
20:57 about predicted job losses and all that
21:00 sort of [ __ ] One of the findings in
21:01 the study was that current employees
21:05 that had AI skills
21:09 had a 56% premium on their salaries
21:13 compared to nonAI literate employees. So
21:17 get get to a nunnery. Start hanging out
21:20 in groups like this. Start hanging out
21:22 in the AI salon. All right.
21:25 All right. [ __ ] All right.
21:29 Why is he lecturing at us? He seems it
21:31 seems very aggressive. I didn't I didn't
21:34 come here to be yelled at by some large
21:37 man that looks like Alec Baldwin on a
21:39 bad day. Jeez.
21:41 Jiminy Christmas. Okay.
21:48 Need to get that 56%. No [ __ ] I know,
21:51 right? I wish that applied to
21:54 carpenters. I think carpenters are going
21:55 to be in good shape. Anyone in the
21:57 handcrafts are going to be in good
21:59 shape. All right. A B2B business.
22:02 There's this whole Johnny IV thing which
22:04 I'm sure you know we can't really talk
22:05 about. Johnny. Johnny. H. We got to
22:07 start. I got to start over. I can't do
22:08 that. So,
22:10 leave that in, please. No, no, no. We're
22:12 going to All right. That's such a good
22:15 moment. It's such a brother moment. He
22:17 calls him Joanie Ives.
22:20 So far, we've got a consumer business, a
22:24 B2B business. There's this whole Johnny
22:26 IV thing which I'm sure you you know we
22:27 can't really talk about. Johnny, we got
22:29 to start. I got to start over. I can't
22:30 do that. So,
22:33 leave that in, please. No, no, no. We're
22:34 gonna cut. All right. Today, I'm here
22:37 with Sam. Sam, before we start, do you
22:39 have anything you need to say? You're my
22:40 literal podcast bro now. Wow. This is
22:43 great. How did you come to this? It's so
22:45 sad. You start a company, then you start
22:47 being a VC, and now I'm here. Are you
22:48 disappointed? Well, I went the other
22:50 way. What do you Well, I was like a VC
22:52 and then I did a podcast and Yeah, it's
22:54 been good for you. It's great. I'm
22:55 really proud of you. Okay, so but I
22:57 think this is great for you. Thank you.
22:58 Okay. I think you're an incredible
22:59 podcast. It's a very nice sweater, too.
23:01 Thank you. Thank you. Okay, so uh I want
23:04 to start by talking about the Stop. What
23:06 are you going to say? Go ahead. I'll say
23:07 it later when we're done recording. I
23:09 wanted to start by talking about the
23:10 future of AI and I'm I want to talk
23:12 about the medium-term because the short
23:14 term is not as interesting to me. The
23:16 long term, who knows? But like 5 10
23:18 years out is what I'm most interested in
23:20 talking about. And I kind of want to try
23:22 to pull out from you your best guess of
23:24 a bunch of specific things. One of the
23:26 places I wanted to start was in
23:28 software. It seems like the most
23:31 effective use cases. Yeah, Danielle and
23:33 Vicki both. I'm starting with this is
23:35 the video that was out three days ago or
23:37 two days ago with his brother. And then
23:40 I'll I'll go to the one that came out
23:41 today. So I'll just play a little bit of
23:43 this one until we get bored.
23:47 Uh oh, what did I do?
23:49 um coding and then chat and code. Yeah,
23:51 chat and code. I'm curious what's next
23:53 like on the next sort of what's the next
23:55 set of things right after that that will
23:58 come? Well, I think there will be
23:59 incredible like other products like
24:01 there will be crazy new social
24:02 experiences. There will be like Google
24:05 Docs style
24:08 AI workflows that are just way more
24:10 productive. You'll start to see like
24:11 you'll have these like virtual
24:12 employees. The thing that I think will
24:14 be the most impactful on that 5 to 10
24:16 year time frame is AI will actually
24:18 discover new science. And this is a okay
24:21 so in case you missed that he he he
24:23 mumbled through three things there. One
24:25 was a social network was the first thing
24:27 he said. So some sort sort of social
24:30 network. Mark I'm going to put money on
24:33 it now that that social network will
24:35 have human verification as part of it.
24:38 So like world ID. Um
24:42 then he said a Google Docs like uh
24:45 interface where it will just do your job
24:48 for you. So imagine I guess Google Docs
24:51 where it's got agents built into it and
24:53 it just does [ __ ] for you. And then now
24:55 he's talking about
24:57 you know boring stuff like you know
25:00 curing diseases and you know re
25:03 rediscovering or discovering new kinds
25:06 of physics.
25:07 a crazy claim to make, but I think it is
25:10 true. And if it is correct, then over
25:13 time, I think that will dwarf everything
25:15 else. Why do you think it'll discover
25:17 new science? Well, I think we've cracked
25:21 Tony Constant Costantino
25:24 or Constantino. Uh, I'm in the Matrix. I
25:27 literally just watched this, then I
25:28 watch your live and it's on again.
25:33 There are no coincidences, Tony. Tony, I
25:37 thought you were informed. You come to
25:39 the AI learning lab. We know everything
25:41 here. Oh, rather. Yes, it is rather
25:45 singularity, isn't it? Yes. Reasoning.
25:49 All right, we've got producer Brandon in
25:50 the house now. Models. We have a long
25:52 way to go. I think we know what to do.
25:53 And you know, 03 is already like pretty
25:55 smart. You hear people say like, "Wow,
25:56 this is like a good PhD." What does it
25:58 mean to crack reasoning? The models can
26:00 now do the kind of reasoning in a
26:02 particular domain you'd expect a PhD in
26:03 that field to be able to do. In some
26:05 sense, we're like, "Oh, okay. The AIs
26:07 are like a top competitive programmer in
26:09 the world now, or AIS can get like a top
26:12 score on the world's hardest math
26:13 competitions." Or I know. We'll we'll
26:15 I'll watch we'll watch this until he
26:16 says one particular thing which I think
26:18 is really interesting. AIS can like, you
26:20 know, do problems that I'd expect an
26:21 expert PhD on my PhD in my field to do.
26:25 And we're like not that impressed. It's
26:26 crazy. But it is sort of a crazy thing.
26:28 Yeah. You know, this reasoning ability
26:29 of the models over the last year. Are
26:31 you surprised? Yes. Yeah. You thought
26:34 that it was just going to be like the
26:36 next token type? No, I thought it was
26:37 going to take a little bit longer to get
26:38 where we are now. The last the last year
26:39 of progress has been faster than I
26:41 thought. Did the way reasoning happened
26:43 happen the way you thought it would
26:44 happen? Like often has happened in the
26:46 history of OpenAI sometime pretty often
26:50 the dumbest first approach turns out to
26:52 work. So I don't I like I like that. So
26:54 he's talking about 01 and 03 the
26:56 reasoning models. The dumbest first
26:58 approach was the one that worked. I
27:00 think I should which is why it happened
27:01 faster than they thought it would. be
27:02 surprised by that anymore. And yet it's
27:04 like a little surprising each time. So
27:06 reasoning will lead to science going
27:08 faster or just new stuff or both? Both.
27:11 I I mean you already hear scientists who
27:12 say they're faster with AI. Like we
27:14 don't have AI maybe autonomously doing
27:16 science, but if a human scientist is
27:19 three times as productive using 03,
27:21 that's still a pretty big deal. Yeah.
27:23 And then as that keeps going and the AI
27:25 can like autonomously do some science,
27:27 figure out novel physics. Is it all that
27:28 happening as a co-pilot right now? Yeah,
27:30 there's there's definitely not like you
27:32 definitely can't go say like, "Hey,
27:33 ChachiT, figure out new physics and
27:37 expect that to work." So, it is I think
27:39 it is currently co-pilot like, but I've
27:42 heard like anecdotal reports from
27:44 biologists where it's like, "Wow, it
27:45 really did figure out an idea. I had to
27:47 develop it a little bit more, but it
27:48 made like a fundamental leap." Yeah.
27:49 Will it be easier to get AI to Wait, I'm
27:55 toggling between Tik Tok and YouTube. If
27:57 you go over to YouTube, please welcome
27:59 my friend
28:00 Dave Jandelle. All right,
28:05 welcome Dave Jandelle. Welcome to the AI
28:08 learning lab. My name is Kyle Shannon
28:09 and you're gonna learn some stuff
28:11 tonight.
28:13 Is that Is that good, Kelly? Should I Is
28:17 he impressed? I have
28:23 I have impressive qualifications
28:28 and a set.
28:31 All right, back at it.
28:35 Welcome, Dave.
28:40 Show him your black bar skills. Build
28:42 you a whole business like build you a
28:43 whole e-commerce business or like do
28:46 like a hard piece of science or medium.
28:49 I wonder about this. If you could like
28:51 build AI, a hundred billion dollar
28:54 particle accelerator. Yeah. And say you
28:57 make the decisions. You look at the
28:58 data. You tell us like uh you know what
29:01 experiments to run and we'll go like
29:03 find the stuff and do it. So you spend
29:05 $und00 million doing that. Yeah. This
29:07 Yeah. The salaries Facebook offered
29:09 OpenAI people is what I wanted to check
29:12 out. A hundred million.
29:15 Facebook is offering Open AI OpenAI
29:18 employees a hund00 million signing Bose
29:20 bonus and as as much for a yearly salary
29:25 and Open AAI employees are turning it
29:27 down. It's amazing. Or you spend hundred
29:31 billion dollars like building it
29:33 infrastructure to like connect into the
29:35 economy. Which of those will it have an
29:37 easier time doing something remarkable
29:38 with? Yeah. And I think the physics is a
29:41 cleaner problem. M you know I think if
29:43 you could get like new high energy
29:44 physics data and then AI the ability to
29:46 like run experiments I think that's like
29:47 a a cleaner problem. Um, I've heard
29:50 people say that. I think that's a
29:51 fascinating comment that he says that
29:53 physics is a cleaner problem. Meaning I
29:57 think meaning
30:00 you will know you will know if it worked
30:02 more easily
30:04 than say
30:06 creating an intelligent swarm of agents
30:09 that can just do all of the jobs of a
30:12 100,000 person consulting firm, right?
30:16 that's not going to be as clean because
30:18 people are going to keep employees just
30:20 because people trust people and they're
30:22 not going to trust that the machine can
30:24 do all the stuff even if the machine can
30:26 do all the stuff. So I I think that's a
30:28 really fascinating
30:30 it feels like a very scientist
30:33 comment that the that the more
30:35 interesting problem to solve is the
30:37 cleanest problem.
30:39 I for me, you know, maybe it's a liberal
30:42 arts thing. I think the problem that's
30:44 most interesting to solve is the
30:45 messiest problem. They expect the first
30:48 area of science and this is like I don't
30:51 know if this is accurate or not but I've
30:52 heard people say that they expect the
30:53 first area of science where AI makes
30:55 autonomous new discoveries to be
30:56 astrophysics because there's just
30:58 mountains of data and we don't have
30:59 enough PhDs to look at it and maybe it's
31:01 not that hard to figure out new stuff
31:02 but I don't really know. Okay. So
31:04 science gets better the coding and chat
31:06 will just keep getting better. Does that
31:08 lead on the business side? Do you also
31:10 then have the ability to just can you
31:12 prompt a whole business like could you
31:14 just say build me this type of business
31:16 it looks like this will happen that were
31:19 there any people doing that for like
31:20 small stuff. Oh by the way not for
31:24 nothing.
31:27 I love that not for nothing corporate
31:29 double speak line. Not for nothing.
31:35 a single employee company
31:38 today, I think it was today they closed
31:41 just sold for $80 million.
31:45 And how long ago did that one person
31:47 start that company? Anyone?
31:51 Six months ago.
31:53 Six months ago, a dude started uh like a
31:57 like a lovable kind of company and it
32:00 sold today for $80 million. one
32:03 employee, $80 million.
32:08 And if you know anyone who's like,
32:10 "Yeah, AI is whatever." Okay.
32:14 You know, you hear these stories of
32:15 people who like
32:18 use AI to do market research and like
32:21 figure out new products and then like
32:23 email some manufacturer and get some
32:25 dumb thing made and sell it on Amazon
32:27 and run ads. Like there are people that
32:29 have actually figured out at small scale
32:31 in the most boring ways possible how to
32:32 like put a dollar into AI and get the AI
32:35 to like run a Mhm. toy business, but
32:39 it's actually working. Yeah. So that'll
32:42 climb the gradient. Yeah. What about in
32:43 the world of like physical stuff? Cuz
32:45 like I get that I mean it seems to me,
32:48 you know, very clear that like software
32:50 is just going this direction. Science I
32:52 know less. Take your word on it. What
32:53 about like moving physical things
32:55 around? behind, but I think we'll get
32:57 there. Uh, for example, I think we have
32:59 some new technology that could
33:03 just do self-driving for standard cars
33:06 way better than any current approach has
33:07 worked. And that might not be quite what
33:09 you meant by like humanoid robots, but
33:12 if our AI techniques can like really go
33:14 drive a car, that's still pretty cool.
33:16 Yeah. Um, humanoid robots are the dream.
33:20 Obviously, I really care about that. I
33:22 think we will get there eventually. It's
33:23 been like a hard mechanical engineering
33:26 challenge. That's more the issue. No,
33:27 both things are hard. But like even if
33:29 we had the perfect brain right now, I
33:31 don't think we have the body yet. Um we
33:34 we actually very early on open we used
33:36 to work on this robotic hand and it was
33:38 hard for all the wrong reasons. Like the
33:39 thing just broke all the time. The
33:41 simulator was like a little bit off.
33:43 Wow. But you know, we'll get there.
33:44 Yeah. I think 5 to 10 years we'll have
33:46 great humanoid robots. Yeah. Like
33:48 amazing. And they'll just like walk down
33:49 the street be doing 5 to 10 years great
33:51 humanoid robots. Um, how do I talk to
33:54 the $80 million man? Um,
33:57 hang on. You go.
34:00 Let me get get his name. He's a full
34:02 stack developer. All these guys making
34:04 this money are full stack developers.
34:06 Um, 80 million.
34:10 Um, one person
34:28 I forget his name.
34:31 It's somewhere in there.
34:36 All right, back at this. Where is my
34:40 There it is. Okay,
34:43 stuff. Yeah. I mean, you would think
34:44 that's where a huge amount of step
34:46 change unlocks, right? I think that will
34:49 be one of the moments that not only is
34:52 unlocks a bunch of stuff in the world. I
34:53 think that will feel the strangest. We
34:57 get used to a lot of things. We get used
34:58 to like Chachi BT doing these things
35:00 that would have sounded like a miracle 5
35:02 years ago. But if you walk down the
35:03 street and it's like half robots, are
35:04 you going to get used to that one right
35:05 away? I don't know. Probably you do, but
35:06 it feels like a big difference. That's
35:07 the one that will feel like there's like
35:08 a new species taking over us. Yeah, I
35:10 think that'll feel I feel like a new
35:11 species or that it's taken over, but I
35:13 think it will feel like the future in a
35:14 way that CHBT still does not. I think
35:16 also if we can figure out great new
35:17 computing devices to make that will feel
35:18 maybe like the future, but as amazing as
35:20 CHBT is or these new coding agents and
35:23 they are amazing. It's like still stuck
35:24 in the form factor of the past. Yeah,
35:26 it's also stuck in it's stuck in the
35:28 computer. Yeah, there's definitely
35:29 something about that it only can do
35:30 stuck out of the computer, but I don't
35:31 know like how much of the economic value
35:33 in all the world do you think is like
35:34 cognitive labor that can be done behind
35:35 the computer? Like half. I was going to
35:36 say a quarter maybe half. I don't know,
35:37 but some big number. Yeah. Does this
35:38 stuff get much riskier once we have like
35:40 super embodied intelligence because the
35:41 things are going to be way stronger than
35:42 us too. I don't know about way riskier.
35:43 I think like the ability to make a
35:44 boweapon or like take down a country's
35:46 whole grid you can do can do quite
35:48 damaging things without physical stuff.
35:50 It gets riskier in like sillier ways
35:51 like I would be afraid to have a
35:52 humanoid robot walking around my house
35:54 that might fall on my baby unless I like
35:55 really really trusted it. Yeah. What
35:56 what like um if you're thinking about,
35:57 you know, we're back here in 10 years
35:59 having another conversation and we're
36:00 like M A O R M
36:05 Mar Schlommo was the $80 million man.
36:10 M A O R S H L O M O. Did AI do what we
36:15 thought it would do? What metrics are
36:16 you expecting? Like is it that like the
36:18 GDP growth curve has a kink in it? Is it
36:20 that like life expectancy is up? Is it
36:22 like there's less poverty? Is it
36:23 something completely different? So every
36:24 year before the last like maybe up until
36:26 last year I would have said like hey I
36:27 think this is going to go really far but
36:28 it still seems like there's a lot that
36:30 we've got to figure out. I feel very
36:32 confident at this point the most
36:33 confident I've ever felt that we kind of
36:33 like know what to do to get the
36:34 incredible AI systems that are just
36:36 super super capable. If something goes
36:38 wrong wait I want you to hear that
36:47 is it like there's less poverty is it
36:49 something completely different? So every
36:51 year before the last like maybe up until
36:53 last year I would have said like hey I
36:55 think this is going to go really far but
36:56 it still seems like there's a lot that
36:59 we've got to figure out. I feel very
37:02 confident at this point the most
37:03 confident I've ever felt that we kind of
37:04 like know what to do to get to
37:06 incredible AI systems that are just
37:08 super super capable. So, we know what to
37:11 do to get to I basically what he says in
37:14 this interview is they know what to do
37:16 to get to super intelligence, which says
37:18 to me they I they're either at AGI or
37:22 they're on the path to get there and
37:24 they know how to get to super
37:26 intelligence.
37:27 If something goes wrong, I would say
37:29 like somehow it's that we build
37:31 legitimate super intelligence and it
37:34 doesn't make the world much better.
37:35 Doesn't change things as much as it
37:36 sounds like it should. How would that
37:38 happen? Seems like a crazy thing to say.
37:39 Yeah. But like I don't know if I told
37:41 you in 2020, maybe I did tell you like
37:44 we're going to make something like Chad
37:47 GPT and it's going to be as smart as a
37:51 PhD student in most areas and we're
37:53 going to deploy it and you know a
37:56 significant fraction of the world is
37:58 going to use it and kind of use it a
37:59 lot. Maybe you would have believed that,
38:01 maybe you wouldn't have, but conditioned
38:02 on that, I bet you would say, "Okay, if
38:04 that happens, the world looks more way
38:06 more different than it does right now."
38:07 Yeah. So, it's like, and he's made this
38:10 prediction before. I think it's really
38:12 fascinating that, you know, we blew
38:15 through the Turing test, right? And and
38:19 now we're going to blow through AGI and
38:21 then at some point we're going to blow
38:22 through artificial super intelligence.
38:24 And like one of his predictions here is
38:26 well what if not that much changes that
38:30 some people go off and use it and make
38:32 80 million dollar companies in six
38:34 months and sell them. But most of the
38:36 people just keep going and just keep
38:37 doing what they do. It's a really that's
38:40 a really fascinating thought. All right.
38:42 So that that that was that interview. I
38:44 think it's that that's an interesting
38:45 interview. This other one I haven't seen
38:47 at all. So this week we're hosting a
38:49 free trading and investing workshop
38:51 right here in Denver. Inside this
38:53 workshop, you'll learn you don't need to
38:55 be a after that. I work with companies
38:58 and individuals trying to figure out.
39:00 Welcome to the OpenAI podcast. My name
39:02 is Andrew Maine. For several years, I
39:04 worked at OpenAI first as an engineer on
39:06 the applied team and then as the science
39:08 communicator. After that, I work with
39:10 companies and individuals trying to
39:12 figure out how to incorporate artificial
39:13 intelligence. With this podcast, we have
39:16 the opportunity to talk to the people
39:18 working with and at OpenAI about what's
39:20 going on behind the scenes and maybe get
39:22 a glimpse of the future. My first guest,
39:25 someone said we didn't pass the touring
39:26 test with AI. I don't know. It is there.
39:30 They keep moving the goalpost on the
39:32 Turing test, but I swear to God, the
39:34 first time I use Chat GBT, I'm like,
39:36 well, there goes the Turing test
39:39 now. I I ain't no scientist. I ain't got
39:42 them there books smarts. Um, but
39:45 basically what they've got is um, how
39:48 they keep moving the B goalpost is the
39:50 amount of time that you would interview
39:53 a computer before you knew it was a
39:55 computer, right? So, it used to be, you
39:57 know, 10 seconds, then it was like a
39:59 minute, there was like five minutes.
40:01 Now, it's like now I think they've
40:02 pushed it out to if if I could interact
40:04 with a, you know, with a a machine for
40:08 eight hours and not be fooled, then we'd
40:10 pass the turning test. I don't know.
40:13 desktop computers were going to give us
40:15 were going to do all our work at one
40:16 point. Uh to our point of boredom didn't
40:18 happen. Here's the difference, Kelly,
40:21 that
40:23 here's why I don't think we can predict
40:25 with the generative A. It's the the G
40:28 and GPT is the difference, right? The
40:30 fact that
40:32 computers now
40:35 are generating original
40:38 work.
40:41 generating original work even though
40:43 it's prompted.
40:45 Um,
40:49 computers historically
40:51 have not been generators, they've been
40:53 processor, they've been computers. They
40:55 would compute the data we give them. So,
40:57 we would say, "Here's our spreadsheet.
40:59 Go do the math faster."
41:02 But we couldn't say, "Go invent a
41:05 spreadsheet.
41:07 Go invent a spreadsheet. Go find all the
41:10 data in the world, put it in that
41:12 spreadsheet, have that spreadsheet do
41:13 all this [ __ ] and right, you know. So,
41:16 we lost Tik Tok. Oh, wait. Did we
41:20 frozen? Yes, I lost him, too.
41:24 Uh,
41:27 I'm I'm restarting Tik Tok because it
41:30 was janky before.
41:37 All right. Am I back, Tik Tok?
41:41 Are we good now? It's back. Back. Back.
41:44 Back. Back. Back. Back. Back. Back.
41:47 Rejoin. Keeps dropping though. I think
41:49 it was dropping. I think it was [ __ ]
41:50 up before. I just I just force quit it.
41:52 No voice. Back. Back. We're And we're
41:55 back. Tik Tok is acting up. Agreed. But
41:59 I think humans will always find work to
42:01 do no matter what. No, I agree. Listen.
42:03 Um,
42:05 snap. Watch this one today, too. That's
42:07 pretty funny, Tony. Oh, don't tell me
42:10 Tony, what was the other thing you did
42:12 today? Play with midjourney video. I
42:14 guess I'm going to put on midjourney
42:15 video and start playing with that.
42:16 You're gonna be like, "Oh, yeah. Been
42:18 there, done that. Whatever, Kyle." Yeah,
42:20 I'm ahead of you by like 10 minutes. All
42:23 right, fine. Fine, fine, Tony.
42:27 Okay. Um, what was I talking about? Oh,
42:33 is it possible that these these
42:37 machines, these AI machines, the
42:39 generative AI stuff really does
42:44 um
42:47 make the need for work optional? I don't
42:50 know. I I think to someone's point, DQ
42:53 Blizzard's point,
42:55 um,
42:57 humans will always want humans need to
42:59 make meaning. Humans need to connect.
43:02 So,
43:04 jobs in the future might look a lot less
43:06 like they look today, but we'll I think
43:08 we'll still have jobs in the future. I
43:10 just think they'll look very different.
43:12 Okay. So, is is Tik Tok back? We're good
43:14 on Tik Tok, right? All right. Let's
43:17 watch Yan Yan movie.
43:20 Um, have you made AI videos of Champ as
43:23 a musical performer playing guitar? Not
43:25 yet, but we could probably do that
43:27 tonight. That's pretty easy.
43:29 Is Sam Alman, CEO and co-founder of
43:32 OpenAI. And we're going to find out a
43:34 bit more about Stargate, how he uses
43:36 ChatgBT as a parent, and maybe get an
43:39 idea of when GPT5 is coming. More and
43:42 more people will think we've gotten to
43:44 an AGI system every year. what you want
43:46 out of hardware and software is changing
43:49 quite rapidly. If people knew what we
43:50 could do with comput, they would want
43:52 way way more.
43:55 One of my friends is a new parent and is
43:56 using chat GBT a lot to ask questions.
43:59 It's become a very good resource and you
44:01 are a new parent and how much has chatb
44:04 been helping you with that? A lot. I I I
44:08 mean clearly people have been able to
44:10 take care of babies without chatbt for a
44:13 long time. I don't know how I would have
44:14 done that. Uh those first few weeks it
44:16 was like every qu I mean constantly. Now
44:18 I now I kind of ask it questions about
44:20 like developmental stages more because I
44:23 kind of I can I can do the basics but uh
44:25 is this normal? Yeah. But it was super
44:27 helpful for that. I I spend a lot of
44:29 time thinking about how my kid will use
44:33 AI in in the future. Um it it is sort of
44:36 like
44:38 by the way extremely kidilled. I think
44:40 every have a lot of kids.
44:42 Yeah, a lot of my friends at OpenAI,
44:44 former colleagues and current ones are
44:46 having kids and people go like, "Oh, you
44:47 know what is with the comically small
44:50 tables?"
44:51 [Laughter]
44:54 She would even helping you with that.
44:56 What What is with the comically small
44:58 tables? Come on. Come on. You can you
45:02 can pony up for the for the 18inch one,
45:05 you know.
45:08 [Laughter]
45:12 It's midcentury modern. Yeah. No, it's
45:15 comically No, it's not even up to their
45:18 knees.
45:20 Sam's not a tall guy.
45:29 Oh my god.
45:31 Okay, this is what happens when you send
45:34 interns to IKEA to buy furniture. This
45:37 is what happens people. Okay? It's why
45:39 you need a cranky Gen Xer in the corner.
45:42 All right, back to your regularly
45:43 scheduled programming
45:46 using AI and
45:50 obviously I think about that a lot. Uh
45:53 but I I think much more about the like
45:55 what they will have that we didn't than
45:57 what is going to be taken away. Um,
46:00 they're like I don't I don't think my
46:04 kids will ever be bothered by the fact
46:06 that they're not smarter than AI. I I
46:08 just like, you know, I there's this
46:11 video that one. Let's see. Midjourney. I
46:14 just saw a midjourney thing. Midjourney
46:16 update is insane. Have we talked about
46:18 this? No, Gareth, that's that's the next
46:19 thing. We're just these are sort of some
46:21 impromptu little movie watching, little
46:23 movie night. Uh, and but midjourney
46:26 video is gonna is going to be the big
46:27 thing we're going to do tonight or I'm
46:29 going to do tonight.
46:31 Um, budgets. Yeah, good thing. Good
46:34 thing they have pants on. Am I right?
46:40 The Finnish vocal fry.
46:45 Yeah. One of the things I think Yeah.
46:48 Sure. um a baby or like a little toddler
46:52 with a one of those old glassy magazines
46:55 um going like this on the screen. Is it
46:57 because it's an iPad? Thought it was a
46:58 broken iPad. Um and you know, kids born
47:01 now will just think the world always had
47:03 extremist AI and they will use it
47:06 incredibly naturally. Yeah. And they
47:09 will look back at this as like a very,
47:12 you know, prehistoric time period. I I
47:14 saw something I it's that's when I mean
47:18 it happened here live over a course of a
47:21 couple of weeks but when when 01 came
47:25 out when the first reasoning model came
47:27 out um
47:30 I realized that it it was the first time
47:33 that I would sit in front of something
47:36 you know one of these generative AI
47:38 tools and I literally didn't know what
47:40 to ask it like I
47:44 I did I didn't have the right life
47:46 experiences and the right training and
47:49 the right exposure to big complicated
47:52 problems
47:54 to need something like that. And
47:58 that was a trippy moment. And I think I
48:01 think most people
48:05 either by choice or by force
48:09 are gonna are going to be forced to
48:11 confront
48:13 um these machines are smarter than you.
48:18 They're they're more empathetic than
48:20 you. They're more compassionate than
48:21 you. They're smarter than you. They can
48:23 work harder than you. They can work
48:25 longer than you.
48:30 And if you don't think that's true, get
48:31 over it. Like, like, is it there yet?
48:35 Well, for me it is. Like, I these things
48:38 are already beyond my capacity. If you
48:40 watch Manis, go open up a hundred
48:43 websites, surf them all, read them all,
48:48 capture all of that data, analyze all of
48:50 that data, and write a report on what it
48:52 found
48:54 and write it into a spreadsheet in 15
48:57 minutes.
49:00 Yeah, it's smarter than me.
49:02 The table is the perfect height size for
49:05 two glasses of water without holding
49:08 without hiding the speakers. Yeah, but
49:09 they don't even have the water on it.
49:12 It's just a tiny table. Just just sit
49:15 there or put the table behind them
49:18 where, you know, so they could pick the
49:20 water up and not kick it. Tik Tok pin.
49:24 Some people don't have a $20 million
49:26 studio, Kyle. It's a good point. It's a
49:28 good point on social media. All right,
49:30 I'm bored by this interview already.
49:31 Well, let's hopefully we can get out of
49:33 the the parenting mode here and get to
49:36 some ASI [ __ ] where a guy talked about
49:38 he got tired of talking to his kid about
49:39 Thomas the Tank Engine. So, he put it
49:41 into chat GPT into voice mode. Kids love
49:44 voice mode. Chat and he was like an hour
49:46 later the kid still talking about Thomas
49:49 the train. Again, I suspect there this
49:51 is not all going to be good. There will
49:53 be problems. people will develop these
49:54 sort of somewhat problematic or maybe
49:56 very problematic parasocial
49:58 relationships and well society will have
49:59 to figure out new guardrails and uh but
50:03 the upsides will be tremendous and and
50:04 we society in general is good at
50:07 figuring out how to mitigate the
50:08 downsides. Yeah. So uh yeah I think
50:11 optimistic we're seeing some interesting
50:13 data where used along in classrooms with
50:17 a good teacher good curriculum chat
50:19 becomes very good used solely by itself
50:21 as sort of a homework crutch can lead to
50:24 kids sort of just doing the same thing
50:25 as trying to Google stuff. I was one of
50:27 those kids that everyone was worried I
50:29 was just going to Google everything when
50:30 it came out and stop learning and you
50:32 know it turns out like
50:34 relatively quickly kids in schools
50:36 adapt. So I think we'll figure this out.
50:38 Think of what you could have become if
50:39 you didn't Google everything, Sam. You
50:41 know, so we've seen this adoption
50:43 figures which are really insane. It's
50:45 open's most popular product.
50:48 Mr. Wonderful here. My friend Howard
50:50 Marx is the co-founder of Activision.
50:53 The shut up you
50:56 dipstick. I don't like him.
51:00 [Music]
51:05 All right, let's go do mid Journey. Mid
51:08 Journey, that's the name. Valerie Cox, I
51:10 think that this that the most amazing
51:12 thing about this tool, it can be used to
51:14 create shopping lists and research. Oh,
51:17 to research the most complicated topics.
51:20 Yeah. I
51:22 So, I'm I'm working on this book project
51:26 um feed the prompt. Feed your prompt.
51:29 And
51:32 I was talking to a branding person today
51:36 and they were asking me questions about
51:38 the book and she told me about her
51:40 friend who's got a kid who's going
51:43 through a phase where he won't eat
51:46 anything round
51:49 like a hamburger. Won't eat it. and and
51:53 she's going mad because she can't figure
51:56 out like there there's all these very
51:58 specific rules, right? And you know, for
52:00 the neurospicy people on this channel,
52:02 right? May seem familiar, right? You
52:04 know, don't don't don't pick on the the
52:07 neurospicy kids, but
52:10 that's a really fascinating use case
52:12 where you could just vomit all of that
52:15 uh context into chat GPT. Tell it, "I've
52:20 got this kid. It's driving me crazy.
52:22 Here's what he used to eat. Here's what
52:24 he likes. Here's the nutrition needs he
52:26 has. Here's the special diet. And he
52:28 can't even eat anything round. I need
52:30 recipes, creative ideas, and a shopping
52:32 list for the next week." Bang. Done. You
52:35 know,
52:37 so let's see. I tested the same prompt.
52:39 Oh, this is cool. Toddweller on AI LLM
52:42 and Manis and Manis one number one,
52:45 Gemini number two, Copilot 3, Deepseek
52:48 4, Chat GBT, Grock, Claude, Quen, Jen,
52:52 Spark, and Perplexity.
52:55 Oh, Chat GBT, Grock, Claude, Quen, Jen,
53:00 Spark, Perplexity all failed to complete
53:02 the task.
53:04 That's super cool, Todd.
53:07 Very
53:08 very cool. That those kind of tests are
53:11 important. Um, okay. My 88-year-old mom
53:15 wouldn't access chat GBT today. Oh,
53:18 couldn't.
53:19 And I thought she was going to lose her
53:21 mind. Yeah, these parasocial uh
53:24 relationships or something. Okay, let's
53:25 go to let's go to midjourney, shall we?
53:28 Let's get rid of Mr. Wonderful there.
53:30 Ah, so here
53:33 this video is a midjourney video. So,
53:36 this is a fictitious car I invented.
53:41 And then I just said make that a video.
53:43 So, now here's how you do it. Let me let
53:46 me flip back into create mode here.
53:53 All right. You can all see that, right?
53:56 So,
53:58 for
54:00 live will end in five minutes. Do your
54:02 physical dexterity challenge now.
54:04 Congratulations. You've passed. Mom is
54:07 completely addicted to chat GBT.
54:13 Oh, this was this one's beautiful.
54:18 This was just a cool image I did ages
54:21 ago.
54:23 And the the animation here is stunning.
54:42 Like the hair is all
54:45 like you look down here like the
54:47 individual strands of hair. There's no
54:48 jankiness in them. Wait, when did Mid
54:51 Midjourney start doing video? I'm so
54:53 behind. Cam, I can't believe you didn't
54:56 know MidJourney did video. It's been out
54:58 for like an hour and a half.
55:03 It dropped today. It dropped today. No
55:05 one's behind. No one's behind. No one's
55:08 judging anyone for not knowing that Mid
55:10 Journey did video. Um they announced it
55:12 like a week ago and we we went in I
55:15 think last week and did some um some
55:17 ranking. You could rank videos and then
55:20 they dropped it today.
55:26 That one's so cool. This one, this one
55:27 reminds me of like a 60s magazine ad and
55:30 now she's blinking. That news was so
55:33 three hours ago. I know. Yeah. Cam
55:36 really disappointed. Really falling
55:38 behind in the lab. That one was crappy.
55:42 I didn't like any of these. This one.
55:44 Oh, yeah. Come on, people.
55:54 My daughter is flying here this weekend.
55:56 I'm afraid it's for a chat GPT
55:58 intervention. Dr. Jay's gonna get she's
56:02 gonna get sat down by the family.
56:04 Listen, listen about this chat GBT
56:07 thing. We've uninstalled it from your
56:09 phone and your iPad. We've restricted
56:11 the domain on your computer. I'm going
56:13 to the library.
56:20 Oh, good lord. This is great for
56:23 consistent characters. Yeah, this is
56:25 pretty cool. Um,
56:28 why is that not going?
56:31 Oh, because I'm probably not on the
56:33 There we go.
56:50 [Laughter]
56:59 You see, you see what happens?
57:02 So, the history of this photo,
57:07 this little part of the photo right
57:09 here, I made a really low res version of
57:12 that in fall of 2022.
57:15 And this this was an image that when I
57:17 made it, it it I got emotional. It took
57:20 my breath away because
57:23 I'd probably created 1,500 or 2,000
57:28 self-portraits using stable diffusion
57:30 and dream booth
57:33 and they were all cool and I was writing
57:34 stories about them and I had this art
57:36 project and then this picture came up
57:38 and there was just something about it
57:40 that
57:42 it was the kind of
57:45 it was the kind of image that I would
57:48 have made without AI had I had the
57:50 skills.
57:53 And it was just one of these weird
57:55 moments where it's like that is
57:57 something I would have created. And it
57:59 felt very visceral.
58:02 And so now you can like upres these
58:04 things and upscale them and now you can
58:06 turn them into movies.
58:10 And he's holding a manila envelope with
58:11 some creepy [ __ ] on it with his creepy
58:13 hand. Crazy, right?
58:18 Oh, these this this I want to tonight we
58:21 might uh
58:23 we might make a a video tonight.
58:27 We'll go write like some sort of sci-fi
58:29 song.
58:30 We'll put some of these
58:33 blueprint animations together. I think
58:35 these things are just amazing.
58:40 Look at that one. The bacteria just sort
58:42 of growing all over the thing.
58:50 And I know I said I'm going to show you
58:52 how to do this, didn't I? And I haven't.
58:56 And you're probably like, "Kyle, hey
58:59 cow, cow cow cow cow cow cow cow cow,
59:06 shut up.
59:15 Look at that. That's so cool.
59:21 And the fact that it just made up these
59:23 music notes because there's some music
59:25 notes down here and it just made this as
59:28 sheet music.
59:30 Be cool to go play that. I love
59:32 MidJourney so much. Wow. Yeah, I know.
59:35 Crazy, right?
59:49 How cool is this?
1:00:13 Come on.
1:00:15 You kidding me?
1:00:20 Look at that.
1:00:22 I think this one where it opens its beak
1:00:24 is just crazy.
1:00:54 very steampunk. Yeah, these are cool,
1:00:57 aren't they?
1:01:00 Looks like some 1840s
1:01:02 science book just coming to life.
1:01:12 Look at that.
1:01:19 No, Vicki, you can extend the video.
1:01:22 You can extend the video. Okay, let me
1:01:24 show you how you make these things. Now,
1:01:25 we got to show Vidy. Vicki, I figured I
1:01:28 figured out extension.
1:01:31 Okay.
1:01:37 Yes, you can. Okay, so here's what you
1:01:40 do.
1:01:41 So, let me go to let me go to organize.
1:01:44 So, here we are in in in midJourney,
1:01:46 right? And you can go to organize, which
1:01:48 just basically lets you flick flick
1:01:50 through [ __ ] faster, right? Awesome 3D
1:01:54 prototyping. Yeah, I know. Crazy. Crazy.
1:01:57 Um, all right. So, let me just flick
1:02:00 down through these and find something
1:02:01 cool that I like. Actually, you know
1:02:04 what would be cool? I did this vibe
1:02:05 storytelling
1:02:08 image.
1:02:19 I think that's the one. Okay. So, we're
1:02:22 going to take this Vive storytelling
1:02:23 image. So, the way it works is right
1:02:25 down here in the lower right hand
1:02:26 corner. So, so you've always had these
1:02:29 other buttons, right? You can vary the
1:02:31 image subtle and strong. You can scale
1:02:34 the image. You can
1:02:38 Oh, you use different images.
1:02:42 And then um and then if I click on more
1:02:45 options, you can you can add other
1:02:47 things like zoom and remix and things
1:02:49 like that. So now you have this new area
1:02:51 called animate image. Let me see if I
1:02:53 can make this any bigger.
1:02:55 Yeah, there we go. So you've got this
1:02:58 thing called animate image and then you
1:02:59 can have autoan animate, low motion,
1:03:02 high motion. So it works the same way.
1:03:04 So you just click on it, bang, bang.
1:03:08 And then you can do manual. And if you
1:03:10 click manual low motion, you can say
1:03:15 um
1:03:17 so it started writing the prompt for me.
1:03:18 So the word storytelling
1:03:21 should shatter.
1:03:24 The word
1:03:28 vibe
1:03:30 should
1:03:33 float away.
1:03:40 The sun
1:03:42 glasses
1:03:45 should turn
1:03:48 dark purple
1:03:52 and go. And now I have not What's this
1:03:56 new personalized? What's is there
1:03:58 there's something new and personalized?
1:04:00 Anyway, okay. So that's how you make a
1:04:03 video. And then basically you just go
1:04:04 back to create and go to the top. And
1:04:07 then now it's making videos for us,
1:04:09 right? And so I can go in and just now
1:04:12 here we got Oh, these aren't done yet.
1:04:20 Come on.
1:04:24 So that was just the auto one. This was
1:04:26 this was low motion.
1:04:33 Then I think the other one I did was
1:04:34 high motion, which is this one.
1:04:43 That was kind of cool.
1:04:46 I mean, they're they're a lot less janky
1:04:49 than I thought they'd be.
1:04:55 Oh, this is the one where I asked for
1:04:57 the text to fly off. So, we'll see how
1:04:58 these do. Okay. So, Vicki, here's how
1:05:01 you here's how you extend the video. So,
1:05:04 let me go find something
1:05:07 that would be pretty. Okay, we'll do
1:05:08 we'll do this girl. We'll do the one
1:05:09 with the this one with the fire. So,
1:05:12 basically what you do is you just click
1:05:14 low motion again. And now it makes four
1:05:17 more videos
1:05:19 that the front half of it is what it
1:05:22 was. And then it's going to add now
1:05:24 another eight or five seconds, however
1:05:26 long these are.
1:05:28 Oh my god, I just made one of my
1:05:30 paintings come to life. Yeah, exactly.
1:05:32 Yep. I've animated a ton. Just don't
1:05:34 know how to extend.
1:05:37 So, basically, you just you just extend
1:05:40 it. You just add animation to an
1:05:43 existing video. I'm pretty sure that
1:05:46 does it.
1:05:47 We'll see here. All right, let's look at
1:05:50 our vibe storytell. Oh, cool. Right.
1:05:53 Right.
1:05:57 That's not bad.
1:06:00 Didn't really do what we wanted it to.
1:06:04 Vibe
1:06:25 [Music]
1:06:32 All right, 92% complete. Come on. Let's
1:06:35 see if we got 10 second videos here.
1:06:40 1 100 2 100 3 100 4 100 5 100 6 100 So
1:06:46 yeah, this extended. So there's a second
1:06:48 little bit here.
1:06:51 So there and then right here's the
1:06:54 extension.
1:06:58 Yeah. So that worked, Vicki. Now, can I
1:07:01 extend and extend? Let's try it. Let's
1:07:04 Let's pick the best one of these.
1:07:12 That one's pretty good. I like her
1:07:14 looking up like that. Okay.
1:07:18 So, here she is here. And now
1:07:21 kind of the same.
1:07:31 Yeah, they're all pretty much the same.
1:07:32 It's interesting.
1:07:36 Oh, you know what I bet I can do?
1:07:39 Yeah, I can bump up stylization. Okay,
1:07:41 here's what we're going to do. So, we're
1:07:43 go to create. We're going to take this
1:07:44 one.
1:07:47 Wait, what did that say when we were out
1:07:48 here?
1:07:50 Oh, look right here, Vicki. You can roll
1:07:52 over it. I had the but but yeah, the
1:07:56 button at the bottom of the video now
1:07:57 says extend auto. So, I'm going to say
1:08:00 extend manual.
1:08:03 I'm going to say
1:08:05 girl looks
1:08:08 into
1:08:10 the camera
1:08:12 and Her
1:08:15 irises
1:08:17 iris as
1:08:21 glow
1:08:23 fire orange
1:08:26 and then
1:08:28 I'm going to say
1:08:31 um oops oh crap video extended.
1:08:37 So we'll go auto extend.
1:08:41 Oh no, I didn't want auto extend. Damn
1:08:43 it.
1:08:44 Shut up.
1:08:46 Extend manual.
1:08:48 There's the prompt. I want to change our
1:08:51 stylization. Oh, we can't.
1:08:56 All right, we'll do high motion for this
1:08:57 set. All right,
1:08:59 I got it. Just bear with me here,
1:09:02 people. I'm trying to figure this out.
1:09:07 Imagine
1:09:10 all the people
1:09:12 [Music]
1:09:15 living.
1:09:22 [Music]
1:09:32 Are we done? No, we're not done.
1:09:35 you can extend the extended video. So
1:09:38 that's it. It's quite simple. There was
1:09:40 a really interesting statement put out
1:09:42 by by MidJourney today.
1:09:45 Let me go look at that. We'll go read
1:09:47 that because I think it's worth it. Uh
1:09:49 Midjourney, that's the game. If you've
1:09:52 made thousands and thousands and
1:09:54 thousands and thousands of midjourney
1:09:57 images, like honest to God, just [ __ ]
1:10:00 going back through your images and just
1:10:03 animating them like that those blueprint
1:10:06 looking things I made. So those
1:10:09 blueprint things, this is one of the the
1:10:11 ways that I'll sometimes if I'm just
1:10:13 [ __ ] around rather than having to use
1:10:16 my brain, I'll just go find an SREF that
1:10:18 I like and then I just I just have chat
1:10:21 GPT modify the SRF. I like I'm like keep
1:10:24 the style the same, but give me new
1:10:25 subjects
1:10:27 and then I just paste a bunch of the
1:10:29 chat GPT things into MidJourney. It just
1:10:31 generates a pile of images. Sometimes
1:10:32 they're really cool. Well, now I can
1:10:34 take those and turn those into those
1:10:35 cool videos and all of a sudden, good
1:10:39 lord, I've now paid 13 months of
1:10:41 MidJourney. Probably time to pay for an
1:10:42 annual and save some money. Probably
1:10:45 like MidJourney is one of the few, Todd,
1:10:47 that I would say, yeah, I would I would
1:10:49 support that. I would support that. My
1:10:52 it is my contention that in in no
1:10:56 uncertain terms, you should not be
1:10:57 buying annual subscriptions to any of
1:11:00 these AI tools just because they're
1:11:02 changing too fast. But Mid Journey's
1:11:03 pretty consistent. Although, you know,
1:11:06 they're being sued by the New York Times
1:11:07 and or no, by Disney and and Universal.
1:11:10 So, who knows how that goes. Okay. Is
1:11:13 this the one? As you might know, okay,
1:11:15 this is the one I wanted to read.
1:11:20 So, as you know, our focus over the past
1:11:23 few years has been images. What you
1:11:24 might not know is that we believe the
1:11:27 inevitable destination of this
1:11:29 technology are models capable of real
1:11:32 time openworld simulations.
1:11:36 So where midjourney
1:11:38 has been headed is to real time openw
1:11:42 world simulations. I've said it on this
1:11:45 channel before. We are not that many
1:11:47 years away from being able to speak a
1:11:50 movie or a AAA video game into
1:11:53 existence.
1:11:55 So, so, so there's that. What's that?
1:11:58 Basically, imagine an AI system that
1:12:00 generates imagery in real time. So
1:12:03 they've got right now they've got um
1:12:07 image video generating tools that are
1:12:11 generating
1:12:13 videos faster than 24 frames a second.
1:12:17 So 24 frames a second is is what film
1:12:20 film is. TV is generally 30 frames a
1:12:23 second or or 60 if you're doing sports
1:12:25 and want to slow-mo stuff. But 24 frames
1:12:28 a second is basically, you know, cinema.
1:12:31 Um, they've got video systems now that
1:12:34 are generating video faster than 24
1:12:36 frames a second. So that means you can
1:12:39 start to buffer. You could prompt a
1:12:42 scene into existence. It starts
1:12:43 rendering that scene
1:12:46 and it's it's now buffering
1:12:49 more more than it's you know than it's
1:12:53 it's it's building it faster than you're
1:12:55 watching it, right? And so what that
1:12:58 would allow you to do is just change the
1:12:59 course of a movie. I'm bored. Make it a
1:13:01 comedy.
1:13:03 Off it goes. Is everyone else ready for
1:13:06 the holiday yet? Yeah, exactly. Okay.
1:13:08 You can command it to move around 3D
1:13:10 space. The environments and characters
1:13:12 also move and you can interact with
1:13:14 everything. In order to do this, we need
1:13:17 building blocks. We need visuals, our
1:13:20 first image models. We need to make
1:13:22 those images move. Video models. We need
1:13:26 to be able to move ourselves through
1:13:28 through space. 3D models and we need to
1:13:31 be able to do this all fast real time
1:13:34 models.
1:13:35 The next year involves building these
1:13:38 pieces individually, releasing them, and
1:13:41 then slowly putting it all together into
1:13:43 a single unified system. It might be
1:13:46 expensive at first, but sooner than
1:13:47 you'd think, it's something everyone
1:13:49 will be able to use. So, what about
1:13:51 today? Today, we're taking the next step
1:13:53 forward. We're releasing version one of
1:13:55 our video model to the entire community.
1:13:58 From a technical standpoint, this model
1:14:00 is a stepping stone, but for now, we had
1:14:02 to figure out what to actually
1:14:05 concretely give you. Our goal is to give
1:14:08 you something fun, easy, beautiful, and
1:14:10 affordable so that everyone can explore.
1:14:13 It's 10 bucks a month, and you get the
1:14:14 video thing as part of it. Um, we think
1:14:17 we've struck a solid balance, though.
1:14:19 You may feel the need to upgrade at
1:14:21 least one tier for more fast minutes.
1:14:25 Today's video workflow will be called
1:14:27 image to video. This means you can turn
1:14:29 images to video. Um there's an automatic
1:14:32 animation setting. There's high motion,
1:14:34 low motion. Okay, that we've all played
1:14:36 with. Pick what seems most appropriate.
1:14:38 Once you have the video, you can extend
1:14:40 it roughly 4 seconds at a time, four
1:14:43 times total. So there's your thing,
1:14:45 Vicki. You can you can go four times
1:14:47 total for the extension. So, so
1:14:49 basically you can make 16-second videos.
1:14:54 We're also also letting you animate
1:14:56 images uploaded outside of midjourney.
1:14:58 Ah, drag an image to the prompt bar and
1:15:02 mark it as start frame. Hang on. Hang
1:15:05 on, people. Calm down everybody. Calm
1:15:08 down.
1:15:12 Just
1:15:13 We're gonna It's gonna be so good.
1:15:16 people. I just need people to calm down.
1:15:19 Wait, can we watch her eyes turn orange
1:15:21 here?
1:15:23 Let's see if it if it took our
1:15:25 instructions.
1:15:28 Come on, girly.
1:15:32 You can do it. I I am so impressed with
1:15:35 the the smoothness and the coherence of
1:15:39 the animation.
1:15:41 Like, nothing's really arai. There's not
1:15:42 a lot of jankiness.
1:15:44 Oh, she didn't turn orange.
1:15:47 It's a bummer.
1:15:55 All right, whatever. Okay, we can upload
1:15:57 an image. So,
1:16:01 um, let's get rid of that.
1:16:05 Oh, start frame. Look at that.
1:16:09 So, I gonna go grab something outside of
1:16:12 here.
1:16:14 That was made not in here.
1:16:19 I could do that.
1:16:31 [Music]
1:16:40 That's good. We'll do we'll do we'll do
1:16:42 disco Kyle. Okay. Drag it in. Oh, and
1:16:46 you can put multiple images in. Wait a
1:16:50 minute. Hang on.
1:16:54 Okay. Video start. Um, man dances
1:17:01 like the old
1:17:05 creepy guy
1:17:07 at the club.
1:17:12 women in the background.
1:17:19 Laugh at him.
1:17:21 [Laughter]
1:17:27 Settings. We'll go high motion. And And
1:17:31 off we go.
1:17:38 And yeah, 70s muscle car. Yeah, exactly.
1:17:42 I'll go find I'll go find some muscle
1:17:44 cars. You know what we can test is if
1:17:46 the wheels actually roll.
1:17:49 I'm off for Junth tomorrow. Going to use
1:17:51 midjourney all day. Oh, that's
1:17:52 fantastic. Yeah, perfect. Perfect. Uh
1:17:55 perfect use of a day off. Love it.
1:18:01 Oh, look at me dancing. All right, this
1:18:02 is going to be good. What's new and
1:18:04 personaliz?
1:18:06 Oh, V7. Yeah. Yeah. Yeah. B7 mood boards
1:18:10 now.
1:18:18 Select.
1:18:23 Oh my god, I'm going to be so good. Oh,
1:18:26 look. Look at the women laughing.
1:18:37 This is gonna be so good.
1:18:46 Yeah. Yeah. Ladies.
1:18:59 Oh yeah.
1:19:05 like the pied piper of
1:19:10 [Music]
1:19:13 Oh, if you're new here, one of the
1:19:15 things that that this channel is really
1:19:17 about is me just entertaining myself.
1:19:24 That's awesome. All right. Well,
1:19:27 there you go. He's a fun guy. He really
1:19:30 is.
1:19:33 All right. Let's go find
1:19:36 Oh, these these are cool cars. All
1:19:38 right, let's let's animate one of these.
1:19:41 So, we'll just do auto
1:19:46 auto.
1:19:48 We'll do low motion and high motion.
1:19:55 Why did that song I'm just a jigalo come
1:19:57 to mind? I'm just a jigalo.
1:20:00 [Music]
1:20:03 [Laughter]
1:20:08 I leave for two seconds and this is what
1:20:10 I return to. Kelly, we've talked about
1:20:13 this. You can't leave the channel. Just
1:20:17 like the Hotel California.
1:20:20 You can't be late. You can't leave
1:20:22 midstream.
1:20:25 You just never know. We've We've
1:20:27 discussed this. We've discussed this.
1:20:29 Everybody knows this. Okay,
1:20:32 let's see if we get Oh, this would be a
1:20:34 cool one to animate, wouldn't it? Uh,
1:20:36 some of these sci-fi things.
1:20:39 Let's see that. Let's see this one.
1:20:41 We'll do We'll do low motion here. And
1:20:43 then we'll do we're going to do manual
1:20:45 high motion. And we're going to do
1:20:49 um we're going to do um
1:20:54 camera
1:20:57 pushes forward
1:21:00 and people walk majestically
1:21:08 as Um
1:21:16 what are these called? Um
1:21:19 transport
1:21:21 craft
1:21:26 elevate
1:21:29 and
1:21:35 fly into the distance.
1:21:40 I I I assume this is going to suck, but
1:21:45 let's see. All right, that's that one.
1:21:48 Let's go find some more pitches. Pitches
1:21:51 worth playing with.
1:21:56 Oh, these were some cool ones.
1:22:07 Uh,
1:22:13 let's do low motion with him.
1:22:18 I think she's creepier. We'll do low
1:22:19 motion with her and him. All right.
1:22:24 Beautiful.
1:22:25 By the way,
1:22:31 this just a fun way to play.
1:22:35 Don't be precious with this [ __ ]
1:22:40 Like one of the ways you learn
1:22:42 what this stuff does well and not is
1:22:45 just go just [ __ ] vomit out work. Oh,
1:22:49 that one's cool. Let's do Let's do high
1:22:52 motion on that one.
1:23:01 That one's cool, too. We'll do low
1:23:03 motion on that one.
1:23:07 Okay.
1:23:10 Oh. Oh. Oh. People, we're gonna do a low
1:23:14 motion and a high motion. That's [ __ ]
1:23:17 trippy.
1:23:19 Why aren't you doing Sydney
1:23:21 s? I know. I need to do Sydney. I know.
1:23:23 Calm down.
1:23:28 Oh my god. These are so cool. What the
1:23:30 [ __ ] was this prompt?
1:23:36 This is This is like nothing.
1:23:40 The prompt is collaborative group.
1:23:44 Oh, I did a No, I just did a global
1:23:46 profile. I just had personalization
1:23:48 turned on. Stylization 150. These are
1:23:51 cool. Why are they so [ __ ] awesome?
1:23:57 Seems like the Oh, high motion on that
1:23:59 one. Low motion on that one.
1:24:03 Let's do low motion on that one. Those
1:24:05 were cool. All right,
1:24:10 I'm probably going to blow through my
1:24:11 credits here, which will suck, but
1:24:13 whatever.
1:24:14 There's me and Anne Murphy. Here's Pink
1:24:18 Car. Here's some muscle car action.
1:24:24 All right, let's do let's let's figure
1:24:26 out a muscle car here. We want to
1:24:27 animate
1:24:29 70 Chevy Impala. That was my first car,
1:24:32 people. That was it.
1:24:36 Olive green with keystone mags.
1:24:44 That's a good shot.
1:24:53 All right, let's go back to that one.
1:25:01 We're going to do low motion and then
1:25:03 we're going to do high motion with a
1:25:05 prompt. Then we'll say car
1:25:10 rolls forward
1:25:13 as camera tracks
1:25:18 and pans
1:25:21 around
1:25:24 the back.
1:25:26 All right, there's that. All right,
1:25:29 let's go back up and look at some of the
1:25:30 [ __ ] we've built.
1:25:37 I think it's important for us to watch
1:25:40 Disco Kyle one more time. Oh yeah,
1:25:44 ladies. I don't like to brag, but you
1:25:48 know, when you got it, you got it. Who's
1:25:51 the creepy guy in the club? Oh, that's
1:25:53 Kyle. He's unfortunate. He's harmless.
1:26:01 I had a 78 Chevy Impala. That was a good
1:26:03 solid car. It was the 70 was was a blast
1:26:07 to drive.
1:26:12 Okay. That's actually super cool. Like
1:26:17 that's super cool.
1:26:22 The wheel physics are not horrible.
1:26:25 The wheel physics are not horrible. This
1:26:28 is impressive. Like, I haven't had a
1:26:30 video model where the wheel physics
1:26:36 have been decent at all.
1:26:43 Wow. All right.
1:26:48 All right. That was bad. That's bad.
1:26:56 Oh, look at this.
1:27:07 Why does it have eyes? The car. Cuz I
1:27:09 was doing something like Cars. Like the
1:27:12 Cars movie.
1:27:14 This is really quite good. Wow.
1:27:18 Look at the Look at the the things in
1:27:21 the distance flying away.
1:27:26 How that's so cool. I know. This is so
1:27:28 cool.
1:27:38 You could totally, you know what would
1:27:40 be a really fun project to do?
1:27:46 Do you guys remember the movie [ __ ]
1:27:48 Scotsy by Philip Glass and Jeffrey Reio
1:27:54 where it was just like all these shots
1:27:55 of like the world falling apart and
1:27:58 industrial pollution and [ __ ] like that.
1:28:03 I think it's important to check your
1:28:05 regulars. Oh no,
1:28:10 it's never important to check. The
1:28:11 minute someone says it's important to
1:28:13 check your regulars, that means this is
1:28:15 at my expense.
1:28:19 Fourth of July video I extended a few
1:28:21 times. Me laughing at Kyle.
1:28:34 That's hilarious.
1:28:37 Oh, this is cool.
1:28:44 Oh, damn that thing.
1:28:51 That's from Vicki.
1:28:56 My nine-year-old will love this for
1:29:00 something for homeschool.
1:29:03 Yeah, this is
1:29:06 I mean, honest to God. So, so there's a
1:29:08 there's a couple of thing actually. Let
1:29:10 me show you a couple of things because
1:29:12 if you're doing some homeschool stuff,
1:29:14 there might be something fun to play
1:29:15 with here. I'll show you a couple of my
1:29:17 favorite things sort of in that genre.
1:29:20 Um, when you're in midjourney,
1:29:25 let's go look at what other videos do we
1:29:27 have here. Oh, we got the old crazy
1:29:29 people.
1:29:31 Oh my god, there's so Okay, there's so
1:29:32 much to look at. Oh Jesus, this is all
1:29:34 so cool. Okay. So, when you're in
1:29:37 midjourney, like one of the things you
1:29:39 can do is
1:29:41 um let me go to chat GPT
1:29:46 and I'm going to say um I'm
1:29:50 homeschooling
1:29:52 my son and we're at an 8th grade
1:29:59 level and he's really
1:30:04 um into enmology
1:30:09 and trucks.
1:30:12 I want you to give me
1:30:16 20 esoteric words
1:30:22 related to those fields. That would be a
1:30:27 good
1:30:30 uh vocabulary
1:30:36 uh starter. I don't I don't know
1:30:38 education, so forgive my stupidness. Um
1:30:43 a good vocabulary starter for him. Okay.
1:30:46 So now it's going to give me 20 words.
1:30:55 Enmology words. Elytra,
1:30:58 the hardened four wings of beetles that
1:31:01 protect delicate hard wings underneath.
1:31:03 Okay, so now we got cool words. I didn't
1:31:06 think Midjourney had it in them, but
1:31:08 this video model does some things that
1:31:10 others don't. This video model's really
1:31:12 quite good. It's It really is quite
1:31:14 good.
1:31:15 Okay. So, here's here's our words right
1:31:20 now. Let's go over to Twitter. We're
1:31:24 going to go to the Twitter and we're
1:31:26 going to search for sref sf. And we'll
1:31:30 put two dashes in front of it. D- sref.
1:31:33 Okay.
1:31:35 And then we should, if we do our job
1:31:38 right, see some cool style references
1:31:41 from midjourney.
1:31:44 And so you can just scroll until you
1:31:46 find one. That's cool.
1:31:49 Or if you collect them like I do, you
1:31:51 can just go grab one. Is that mine? Oh,
1:31:53 that's mine. That's cool. Maybe I need
1:31:56 this in lower case. Sref.
1:32:00 [Music]
1:32:09 It's a bunch of stuff I did. I don't
1:32:10 know why my stuff's showing up here.
1:32:13 All right, here's one. So, let's just
1:32:15 say that we got this cool this cool
1:32:17 style we like. So, I'm going to grab
1:32:18 this sref.
1:32:20 So, I'm going to take that over to
1:32:21 MidJourney.
1:32:23 I'm going to pop that in. And then I'm
1:32:26 going to come back over here to chat
1:32:27 GPT. And I'm going to go grab one of
1:32:29 these random words, elytra. In fact,
1:32:31 we'll do
1:32:34 we'll do
1:32:37 three of the four of these things. So,
1:32:39 we're going to go to midjourney
1:32:41 and then I'm going to just paste them
1:32:43 in.
1:32:46 Oh, I didn't copy. Dumb dumb. Copy.
1:32:51 I'm going paste them in. And I'm going
1:32:53 to get rid of the definitions.
1:33:16 Okay.
1:33:18 Get rid of the definition.
1:33:21 Get rid of that thing.
1:33:23 All right. So, we got four words. Let's
1:33:24 put commas between them so it knows that
1:33:27 they're separate words. It might know
1:33:29 this anyway.
1:33:31 And then we got that sref, right?
1:33:36 No fast hours left. Oh, you [ __ ]
1:33:39 [Laughter]
1:33:42 I burned through all my fast hours. See
1:33:44 what you did to me, people?
1:33:47 Fast turbo. Relax. Can I do turbo? Do I
1:33:51 have turbo hours
1:33:55 go?
1:33:58 No,
1:34:00 guess not.
1:34:12 Can I not do anything now? Am I totally
1:34:15 locked out?
1:34:18 Create.
1:34:21 Why can't I do draft hours, relaxed
1:34:23 hours?
1:34:25 Does anybody know?
1:34:28 This sucks. But anyway, then you could
1:34:31 turn those into midjourney is really
1:34:33 good at just taking random words. He ran
1:34:36 out as you predicted. I did predict. I
1:34:38 predict predicted it for myself.
1:34:43 All right, let's go look at more movies.
1:34:46 Oh, these are so cool.
1:34:48 This is so [ __ ] cool. PayPal, look at
1:34:51 this.
1:34:56 What money do you have? I'm on the $10 a
1:35:00 month plan.
1:35:08 I mean, you could do a whole [ __ ]
1:35:11 movie.
1:35:38 Wow.
1:35:49 Tell me that wouldn't stop you in your
1:35:51 tracks. In fact,
1:35:55 let's
1:36:02 was it this one?
1:36:09 Yeah. So, we're going to download that.
1:36:36 These
1:36:42 are trippy, man.
1:36:48 All right, we like that one.
1:37:01 It's weird. Mary's grandmother. It
1:37:03 really is.
1:37:07 This smoke looks very realistic, too.
1:37:11 Well, that one's a little janky.
1:37:16 God, I love these old characters, man.
1:37:29 All right,
1:37:31 we'll download him.
1:37:35 All right. So, what we're going to do,
1:37:37 how do we do this the cheapest, easiest
1:37:39 way? I know.
1:37:42 I'm going to just go into So, we did
1:37:44 downloads.
1:37:46 So, we're going to go bang.
1:37:48 Then, we're going to go add video to
1:37:50 end.
1:37:53 Add clip to end.
1:37:56 And we're going to go bang.
1:38:01 Did it do it?
1:38:08 No.
1:38:12 Oh, it crashed.
1:38:14 Like, why didn't that work? Because it
1:38:16 crashed. All right. So, we'll do her
1:38:18 first, then we'll go add clip to end
1:38:23 downloads.
1:38:27 Yeah, we'll do him next. All right. And
1:38:30 then we'll go add clip to end.
1:38:33 Add clip after selection. We'll do this.
1:38:37 Is that her?
1:38:51 Done. What's going on here?
1:38:57 All right, then we'll do
1:39:05 add clip to end.
1:39:07 Which one haven't we seen yet? Oh, the
1:39:10 one with the house.
1:39:13 Okay, done. Great. Now, let's go to
1:39:17 Sunno and make a song.
1:39:20 I got to change tabs or you won't be
1:39:22 able to see that dang thing. Oh, and by
1:39:24 the way, Brandon, I haven't been looking
1:39:26 at the post-it note all night, which
1:39:27 you're probably painfully aware of.
1:39:35 It's been quiet. Oh, that's good. Yeah,
1:39:36 because I'm I'm busy doing [ __ ] Um,
1:39:40 what am I gonna What am I Where am I
1:39:42 going? Oh, I got to make a puno tab. So,
1:39:44 we'll call this.
1:39:48 And then we're going to change that. And
1:39:50 then we're gonna share Sunna with you so
1:39:52 you can hear it. And then we're going to
1:39:54 go create.
1:39:57 Um, okay. I need a description from all
1:39:59 you creative people, all you creative
1:40:01 writers. When I what I saw in the
1:40:04 bathroom mirror this morning, those old
1:40:06 men. Um, I need a description for a
1:40:09 song. Like what would be the what would
1:40:11 be the kind of music that would go with
1:40:13 that? Like the the Weird Mary one could
1:40:15 could work, but like how do we describe
1:40:18 that? like dissonant
1:40:22 um
1:40:24 dissonant janky
1:40:28 um Appalachian
1:40:30 something
1:40:43 mel Oh this good side hustle maybe
1:40:45 melancholy melanch
1:40:47 Kie.
1:40:51 Oh, wait. Hang on. I got to turn on
1:40:55 dark reader.
1:40:59 Come on.
1:41:11 Is it on? Oh, here we go. Is it coming?
1:41:14 Yeah, there we go. Melancholy. Uhoh.
1:41:18 Did it turn off? Loading. Please wait.
1:41:20 On.
1:41:22 Okay.
1:41:24 Come on. Flip my flip my browser.
1:41:29 Melancholy
1:41:32 dissonant
1:41:36 appian
1:41:38 folk.
1:41:40 But I like dystopian unease with
1:41:44 dystopian
1:41:46 unease.
1:41:53 Melancholy dissonant appellacian. I just
1:41:56 spelled that wrong. Epilian.
1:42:00 Appalian.
1:42:07 Okay.
1:42:09 Oh, we want instrumental
1:42:12 create. Okay.
1:42:17 [Music]
1:42:18 [Applause]
1:42:22 [Music]
1:42:24 All right. Let's So, here's our video.
1:42:27 So, we're going to go save. We're going
1:42:29 to go save. We can't save it. Why can't
1:42:32 we save it?
1:42:36 QuickTime player. Close.
1:42:42 Save to downloads. Uh, creepy.
1:42:48 Creepy
1:42:52 into downloads.
1:42:55 I wonder if it's saved.
1:42:58 Creepy.
1:43:01 Yes.
1:43:17 All right, there's our creepy movie.
1:43:20 Now, can I just drop a longer audio file
1:43:23 on that and have it work?
1:43:26 Hey, Kyle. Yeah, two things. One, fun
1:43:30 fact from Ohio. It's actually
1:43:31 Appalachin.
1:43:34 Um, also we're still seeing Sunno. Yeah.
1:43:37 Yeah. Yeah. I was going to play it.
1:43:40 [Music]
1:44:08 [Applause]
1:44:14 I'm gonna add the word devastating at
1:44:16 the beginning of this.
1:44:18 Okay. Devastating.
1:44:25 Devastate
1:44:29 Appalachin
1:44:34 uh instrumental create
1:44:37 and I'm going to go create. We'll create
1:44:39 six of them. We'll get something good
1:44:41 out of this.
1:44:44 You tell him, Brandon. Looks like how
1:44:45 folks mis mispronounced Louisville.
1:44:48 Lewisville. Champy jamming out. It's not
1:44:52 creepy sounding. Maybe a minor note.
1:44:53 Yeah, that's why I put devastating. I
1:44:55 know.
1:44:56 [Music]
1:45:14 All right.
1:45:18 [Music]
1:45:46 Oh, there's some good ones here.
1:45:52 [Music]
1:46:06 Hey.
1:46:09 [Music]
1:46:21 This one,
1:46:22 right?
1:46:24 [Music]
1:46:40 All right. This This is the one we're
1:46:42 gonna We're gonna do this one because
1:46:44 it's weird. It's not right, but it's
1:46:46 weird. Okay. So, that's downloaded.
1:46:48 Okay. So, we're going to do this in
1:46:49 iMovie. We're going to do this with
1:46:51 transitions and [ __ ] because I'm going
1:46:52 to slow down. It's not weird. It's
1:46:54 irregular.
1:46:57 iMovie
1:47:06 tabs. I knew that. I knew that. I didn't
1:47:09 need I didn't need a producer to tell me
1:47:11 to do that. That's because I'm a
1:47:13 professional.
1:47:15 I know exactly what I'm doing. I don't
1:47:17 need I don't need fancy people telling
1:47:20 me what to do with my life with a with a
1:47:24 audio production degree.
1:47:29 Okay. New movie
1:47:34 that's giving Fiddler on the roof. And
1:47:37 Murphy, what's happening? We're doing
1:47:38 We're making a creepy movie, man.
1:47:42 Look at these things.
1:47:46 Okay, that one, that one, and that one.
1:47:49 And echoes from the hollow. We're going
1:47:51 to drop him on in.
1:47:54 All right, we're going to start with
1:47:55 her.
1:47:57 No, we're going to start with him. We're
1:48:00 going to slow him down. So, let's get
1:48:03 our timer thing there and go
1:48:06 slow
1:48:08 and we go play.
1:48:15 That's a little too slow.
1:48:19 Wait, let me do custom.
1:48:23 Let's do
1:48:25 70%.
1:48:30 Um,
1:48:32 champ, stop whining.
1:48:42 I think we can just lose it. We can go
1:48:47 right to there.
1:48:53 All right, there's that. Now we're going
1:48:55 to do her her because she's creepier
1:48:58 than [ __ ]
1:49:00 We're going to slow her down to custom
1:49:02 to 70. We'll slow her down to 85%.
1:49:18 I think we'll do a similar sort of cut.
1:49:20 I don't like the double look back. So,
1:49:22 she'll look into our eyes and pierce our
1:49:25 soul and melt our freaking minds. Okay.
1:49:31 And then we grab melty house. And I
1:49:35 think we lose the beginning here. No,
1:49:37 maybe not. Maybe we keep that. That's
1:49:38 creepy, right?
1:49:42 Yeah. Keep the whole thing here. But
1:49:44 we're going to do custom speed of 80%.
1:49:49 That's going to slow that down. Now
1:49:50 we're going to go get transitions and
1:49:53 we're put in cross. Do we want cross
1:49:55 blur or cross dissolve?
1:50:01 I think cross dissolve two.
1:50:06 Can someone explain why I can't make a
1:50:09 cross dissolve more than a second long?
1:50:13 Says there's not enough stuff, but there
1:50:16 is. Well, whatever. I don't care.
1:50:22 If anybody knows iMovie.
1:50:27 Wait, why is she so short?
1:50:37 What? What?
1:50:42 All right.
1:50:45 Yeah, something something weird
1:50:46 happened.
1:50:51 Okay.
1:50:53 And let's make her 75%.
1:50:56 Make it a little longer. All right. Now,
1:50:59 let's go back to our media. Let's drop
1:51:01 our little song in here
1:51:05 and we'll chop off the beginning of
1:51:07 this. We'll come right in, but we're
1:51:10 gonna
1:51:14 All
1:51:22 [Music]
1:51:26 right.
1:51:27 [Music]
1:51:40 Not the right song. You told me that. I
1:51:43 didn't listen. Hang on. You were right.
1:51:46 I was wrong.
1:52:15 I know here.
1:53:14 Yeah, I know you can't hear. You're
1:53:15 gonna have to just trust me on this one.
1:53:18 It's called The Hollow Hills.
1:53:21 Come on, download. Download already.
1:53:25 I know it's getting late, people. You
1:53:27 got to hang in there. It's going to be
1:53:28 We're going to creep out the Twitter.
1:53:31 Also, the video quality is only 480p and
1:53:33 it looks this good. Yeah, I know.
1:53:37 Bottleneck Banjo. I like
1:54:10 These are these are pseudo creations.
1:54:40 Okay, get rid of that.
1:54:45 Oh, get rid of that.
1:54:48 Delete media from project. Yes, I want
1:54:50 it out of there. And you don't got time
1:54:54 for that.
1:55:02 that.
1:55:07 No, this is it. Okay. So, wait. Where's
1:55:11 my iMovie? And then where's my finder?
1:55:14 And then where's my H? And I And then I
1:55:16 go H. And I go and and I put H. And then
1:55:21 I Wait.
1:55:25 Uh oh. We got to make this thing
1:55:27 smaller.
1:55:29 And then we're going to go to there. Oh,
1:55:32 I see what it did. What did it do? I
1:55:34 don't know what it did. Oh, we got to
1:55:36 fade to black at the end. God, this is
1:55:38 There's so much work to do here, people.
1:55:41 You don't understand how hard it is to
1:55:43 be me. I say it night after night after
1:55:47 night and you're all like, "It's so
1:55:48 easy. You just turn on the camera and
1:55:50 talk. Marge showed you how to do it
1:55:52 anyway."
1:55:55 All right.
1:55:57 And then we're going to do this to here.
1:56:00 And then we're going to do this to here.
1:56:03 And we're going to go.
1:56:08 [Music]
1:56:20 Is that too fast?
1:56:24 [Music]
1:56:35 not as hard as your producer trying to
1:56:37 watch you produce something.
1:56:39 [Music]
1:56:55 That's cool. All right. Good enough.
1:56:58 Good enough for internet work.
1:57:03 And now it won't it won't upload to
1:57:05 [ __ ] Twitter anyway because Elon Musk
1:57:07 can't fix his upload service. All right.
1:57:10 Uh, so this is what's this called? This
1:57:12 is called Creepilacia. Creep
1:57:16 Creepilacia. Yes, sir. Um, sorry. I
1:57:19 can't let this stand like this. You've
1:57:21 got to fade that audio out. Oh, fine.
1:57:26 Fine.
1:57:29 [Music]
1:57:40 I got to do Hang on. And I got to put a
1:57:43 point in there. And then another point
1:57:48 there.
1:57:50 And then drag that out and down. Drag
1:57:52 that over. And now drag it down.
1:57:57 Okay.
1:58:02 Fine. And you want to be the extra. You
1:58:05 could put a title card on that black
1:58:08 screen with like
1:58:11 produced in the AI learning lab by Kyle
1:58:13 Shannon using the journey video. Just
1:58:16 just you know, chef's kiss.
1:58:19 Yeah. All right. Uh
1:58:22 let's see.
1:58:26 All right. Um,
1:58:30 AI learning
1:58:32 lab productions
1:58:35 [Laughter]
1:58:38 by
1:58:40 Kyle. I know we'll say we'll say um I'll
1:58:44 do by Kyle Shannon down here by Oops. By
1:58:49 Kyle
1:58:51 Shannon. Uh, I'll say
1:58:56 mid
1:59:01 journey.
1:59:03 How do we want to do this? Uh
1:59:12 so do this
1:59:16 um
1:59:19 AI learning lab.
1:59:25 Thanks.
1:59:27 Midjourney
1:59:30 and Suno
1:59:35 and then let's see. We'll do this.
1:59:46 Oh no, it didn't save it. The hell.
1:59:49 God.
1:59:52 Thanks,
1:59:54 Mid Journey
1:59:58 and Suno
2:00:01 and then
2:00:05 AI learning. Look how good that was. We
2:00:07 all excited.
2:00:10 [Music]
2:00:15 I'm still here. Side hustle maybe still
2:00:17 here.
2:00:23 Transition will be the amount of time
2:00:24 you turn the clip.
2:00:29 Okay, whatever. Export.
2:00:31 Export. Share file.
2:00:34 Next.
2:00:37 Creepy movie. Creepy Lacia.
2:00:40 Replace. All right. Now, let's watch.
2:00:45 Oh, I got to change tabs. Oh, did you
2:00:48 not see any of that? I'm such a loser.
2:00:52 Oh, you did? Okay, good.
2:00:56 It was iMovie.
2:00:59 So.com
2:01:00 X. Can you see that? Yeah. Okay, you can
2:01:02 see what I see. All right, good.
2:01:05 G.
2:01:08 Um,
2:01:11 if you missed the AI learning lab
2:01:15 tonight, you missed a good one.
2:01:20 Exclamation point
2:01:27 uh atsuno.
2:01:31 Um, we'll do at scobilizer.
2:01:40 Don't
2:01:43 miss it
2:01:47 again
2:01:48 or
2:01:50 these people
2:01:53 will be in your dreams.
2:02:02 [Laughter]
2:02:06 I like being stupid.
2:02:08 Stupid is my jam.
2:02:13 Watch it. Try to upload. This should be
2:02:14 comical.
2:02:22 Error. There's a surprise.
2:02:24 Let's try to post it and see if it
2:02:26 uploads with a post.
2:02:30 16 17
2:02:32 18 19 25 20 18 19 17 Dear Elon Musk.
2:02:39 Hey, quickly just quickly curious, could
2:02:42 you uh what do they call it in uh
2:02:45 software development? [ __ ] fix the
2:02:48 upload to your Twitter.
2:02:51 Jesus good lord.
2:03:01 15 16. I think what we might have to do
2:03:03 is save this in a different format.
2:03:05 Yeah, we're going to have to. All right.
2:03:07 One of the things I've learned is that
2:03:09 you can you can hack Twitter by saving
2:03:13 it in a non-Apple format.
2:03:16 [Laughter]
2:03:18 Oh my god, this is so painful.
2:03:22 Uh, is this 480p? I think it's 720p.
2:03:27 creepily.
2:03:29 And then let's go back to the Twitter.
2:03:34 Did it upload? No. So, we're gonna get
2:03:37 rid of it here.
2:03:39 We're going to go grab it here. We're
2:03:41 going to go Why? Wait, where did it go?
2:03:44 Where did it go? There it is.
2:03:54 Oh, I got an error. There's a surprise.
2:03:56 Dear Elon,
2:04:00 hey, I know catching a rocket with
2:04:03 chopsticks is hard, but I didn't realize
2:04:06 that uploading a video to your stupid
2:04:08 [ __ ] social media site was harder
2:04:11 than catching a rocket with chopsticks.
2:04:14 Elon, dear Elon,
2:04:16 dear, dear Mr. Musk,
2:04:20 congratulations
2:04:22 on the Doge
2:04:25 savings and catching rockets. Quick
2:04:30 curious wondering, could we fix the
2:04:33 upload to X?
2:04:40 Losing federal funding for SpaceX. Maybe
2:04:43 they could take the federal funding for
2:04:44 SpaceX SpaceX and fix their upload
2:04:46 button. Anybody know how to do this?
2:04:53 This is unbelievable. It really like
2:04:55 it's just it's just stunning. It's
2:04:57 stunning how bad this is.
2:05:00 I
2:05:05 It's a 19-second video.
2:05:14 anyone
2:05:17 dear dear anyone dear Mr. Musk
2:05:21 post post.
2:05:32 Jeff Flanigan shakes head.
2:05:42 And the sad thing is there's still 33
2:05:46 people watching on on Tik Tok. 47 people
2:05:49 watching on other streaming platforms.
2:05:52 Oh, we're at 87%. It might just go. Come
2:05:56 on. You can do it.
2:05:59 You can do it. Please. Yes.
2:06:06 Okay. So, now
2:06:09 could you all please go like this?
2:06:17 Uh, let's let's edit this so we say
2:06:19 something about the new video feature.
2:06:21 You missed a good one. Um, we
2:06:27 played around with midjourney
2:06:33 V1 V1
2:06:37 video generation
2:06:40 and wow, it's impressive.
2:06:46 Thanks also to
2:06:53 Sununo Music for your musical swellness.
2:06:59 [Laughter]
2:07:02 Hey, Scoilizer.
2:07:07 Cool.
2:07:10 A question mark. Okay.
2:07:15 Don't miss the learning. Don't miss the
2:07:20 AI learning lab again or these people
2:07:25 will
2:07:26 haunt your dreams.
2:07:30 I know how to write. Okay. Update.
2:07:33 Beautiful. Fantastic.
2:07:36 Fantastic. Bob, why don't you tell them
2:07:38 what they want? Well, they haven't
2:07:40 really won anything. Seriously,
2:07:43 cuz you know.
2:07:44 [Music]
2:07:47 All right. Now, do me a favor, people.
2:07:49 Go find this post. Go to my Twitter,
2:07:51 Kyle Shannon. Find this post of the
2:07:54 creepy people. I got some hearty hearts.
2:07:57 Got a comment here.
2:08:01 I think,
2:08:05 huh?
2:08:08 Said there's a comment.
2:08:10 [Music]
2:08:12 Anyway, all right.
2:08:16 I'm out of here, good people.
2:08:21 [Music]
2:08:24 The fade on credits is great. Awesome.
2:08:26 All right. All right, everybody. Uh,
2:08:28 Wednesday night, it'll be normal time
2:08:30 tomorrow. Who knows what we're going to
2:08:32 do? Should be fun. But that was that was
2:08:34 cool. That midjourney video stuff's
2:08:36 pretty swell. Um,
2:08:40 I kind of feel like with V3 doing acting
2:08:43 and voice and sound effects, like I feel
2:08:46 like every other video model right now
2:08:48 feels very thin to me. Like I shouldn't
2:08:51 have to go make sound effects. I
2:08:53 shouldn't have to go make music. It
2:08:54 should just do it for me. Uh, but you
2:08:57 know, we'll get there.
2:08:59 All right, peace out everybody.
2:09:00 Midjourney rabbit hole time. Yeah,
2:09:02 everybody go off. Go play. Um, burn
2:09:04 through your credits like I did. They're
2:09:07 all gone. I lost them all.
2:09:12 All right, take care everyone. Bye.