9/23/2025 - Suno V5's Music Magic and Google's Mixboard

Description

Chapters

Transcript

0:00 Heat. Heat.
0:09 [Music]
0:27 [Music]
0:37 Heat. Heat.
0:47 Oh
0:53 hey
0:57 [Applause]
1:00 [Music]
1:07 [Music]
1:16 a
1:18 [Music]
1:25 [Music]
1:27 [Applause]
1:29 [Music]
1:34 Good evening, good people.
1:36 >> What is happening? What's shaking?
1:38 What's going down?
1:40 [Music]
1:46 [Applause]
1:48 Are there really only two people in
1:49 here? There's more than two people in
1:50 here. Tik Tok's weird, man.
1:55 Wait. One, one, two,
1:59 three, four.
2:02 It's like six people in here. Seven
2:04 people. I don't understand Tik Tok.
2:16 [Music]
2:20 Girl,
2:22 you're looking
2:24 fine tonight.
2:27 [Music]
2:29 Every fella's got you in his sight.
2:34 [Applause]
2:37 What you doing with a clown like me?
2:42 [Music]
2:44 He's surely one of life's little
2:48 mysteries.
2:51 You You and I were in key there, little
2:53 buddy. So tonight
2:58 [Music]
2:59 stars above.
3:01 [Music]
3:05 Wait.
3:07 How did I
3:11 [Music]
3:13 ever win your love?
3:18 What did I do?
3:22 What did I say
3:25 [Music]
3:27 [Applause]
3:29 to turn your to turn your angel
3:35 my way?
3:40 I'm the guy that I never learn to dance.
3:50 I never even got one second glass
3:57 across a crowded room that was close
4:00 enough.
4:02 [Music]
4:04 I could look but I could
4:07 never touch
4:09 [Music]
4:14 the stars about
4:17 [Music]
4:18 that
4:19 [Music]
4:23 ever with your love.
4:28 What did I do?
4:31 What did I say?
4:35 [Music]
4:36 to turn your angel as my way.
4:44 [Music]
4:50 So, you know how last night I said we're
4:52 going to dig deep cuz there was not a
4:54 bunch of new stuff to play with and I
4:56 thought it was time we we should get
4:59 back into some tools and just dig a
5:00 little deeper. Well, like a bunch of new
5:02 [ __ ] came out today.
5:06 So, we're going to be doing that. We're
5:08 going to be looking at new stuff. What's
5:09 happening everyone? If you're new here,
5:11 my name is Kyle Shannon. This is the AI
5:12 learning lab.
5:17 If you have questions about AI, opinions
5:19 about AI, fears about AI, excitement
5:23 about AI,
5:26 a name with the letters A and I in it.
5:30 [Applause]
5:35 [Music]
5:41 Standing between
5:44 you and a hard place is insane.
5:47 [Music]
5:51 Standing too near
5:54 you in a fire makes it clear
5:58 [Music]
6:00 your trouble to me.
6:05 [Music]
6:06 Bridge trouble. Can't you say
6:09 [Music]
6:12 leaning in close?
6:15 Smell of your perfume scares me most.
6:21 Leaning away
6:24 feels farther every day.
6:31 You're trouble to me.
6:37 Rich guy,
6:40 [Music]
6:43 [Applause]
6:45 [Music]
6:47 what do you think there, Jeffy?
6:53 [Music]
6:56 Um,
6:58 yeah. So, we we'll get we'll get looking
7:00 at some stuff. Y'all been playing with
7:02 anything interesting on mobile because
7:05 lots of lightning. Yikes.
7:08 It was raining all day today here in in
7:10 Denver. Denver, Colorado. They said it
7:14 was downright soupy.
7:16 Soupy sales. Thank you for sharing the
7:19 live, Danielle. That helps a lot. Get
7:21 some more folks in here. I do appreciate
7:25 it.
7:28 [Music]
7:55 [Music]
8:04 The boys were singing, jingling
8:08 the summer that we met.
8:12 You were 10 and seven. Wait, and 17.
8:17 How could I forget?
8:20 [Music]
8:21 Um,
8:25 wait.
8:31 The boys were singing shingling the
8:33 summer that we met. You were 10 and 17.
8:38 How could I forget?
8:41 I was shy to kiss you while the whole
8:44 wide world could see. So, shingling
8:48 shingling said everything for me.
8:52 [Music]
8:54 The way we danced was not a dance but
8:58 more a long embrace.
9:01 We held on to each other and floated
9:05 there in and floated there in space.
9:10 I was shy to kiss you while the whole
9:13 world could see. So shingling said
9:17 everything for me
9:22 and all the poor old old folks. They
9:25 thought we'd lost our minds.
9:28 They could not make heads or tails of
9:31 the young folks funny rhymes.
9:35 You and I knew all the words and we
9:37 always sang along
9:40 to old shaming dong ding shine dang
9:47 dong.
9:49 [Music]
9:56 That's a pretty old song, ain't it? I
9:59 like that pretty old song. Sent too
10:00 soon. I started a book, a hard coverver
10:03 book. I wanted to copy a quote as I
10:07 pressed my finger.
10:09 What's that mean?
10:15 I embraced an old art.
10:23 [Music]
10:34 [Applause]
10:40 Okay.
10:42 [Music]
10:42 [Applause]
10:48 Woohoo.
10:52 [Music]
11:15 Okay. What do you say we get this show
11:18 started?
11:20 Like copy paste. Never mind. Girl too
11:23 used to tech. Girl reads paper. Girls
11:26 girl can't copy paste.
11:29 Silver Fox is just having a whole little
11:32 conver She's having a whole tech support
11:34 conversation with herself at the moment.
11:37 It's fine. You can just She's good.
11:38 She's all right. She'll be okay.
11:44 What? Hey. Hey, honey. What's the key
11:46 for copy? Hey, Marge. Marge. Yeah. Hey,
11:50 when I'm cop is copy the same thing as
11:53 cut it. It's not. See, that's where I
11:56 get confused.
11:58 But no, paste. I understand. That's
12:00 where you stick it in. I didn't mean it
12:02 like that. That was so silly. It's just,
12:05 you know what I mean? You know what I
12:06 mean? I know it was not fun. That That
12:09 was sophomore humor. No, I get No, I
12:11 understand. I understand. No, Pat say
12:13 Jack would never make a joke like that.
12:15 That's correct. That is correct, honey.
12:17 Yep. Nope. I know you like the wheel. I
12:19 do. I No. Pretty. I was just trying to
12:22 if if the copy and the and the cut were
12:24 the same because I didn't I couldn't
12:26 figure and Yeah. No, I'll figure it out.
12:28 I I know. I got the Google. I got the
12:30 Google. I do. Yep. No, I'll I'll turn it
12:33 on. How do I How do I know it's on
12:35 again? Right. It's just It's on. It's
12:37 just on. If I If I go to the Google
12:40 Yeah. Yeah. That Marge, she's helpful.
12:43 She is helpful. I don't know if you've
12:46 if you've met Marge. Why is my light not
12:48 lighting? Whatever.
12:51 I I try to be professional around here.
12:53 I don't have a producer tonight.
12:56 I got the wrong cameras going. I don't
12:58 have the black bar up. It's just a mess.
13:01 Happy Rashana, by the way.
13:04 If you celebrate that, you're probably
13:06 not watching this.
13:11 Or maybe this is your treat. Maybe maybe
13:14 the AI learning lab is your manachevitz.
13:20 Oh man, Source Camp, what's happening?
13:24 So, we got we have uh we have Sunno 5 to
13:28 play with tonight. We've got Google's
13:31 got Mixboards out. Google also added
13:35 Canvas to um to Gemini. So, in the
13:39 tradition of all of the AI model
13:41 companies copying one another, you've
13:44 now got that feature in in uh in Gemini
13:48 as well,
13:51 cuz why not? Um,
13:55 what else? I had a I had a good day in
13:57 Midjourney today. I spent a lot of time
13:59 in Midjourney. 11 Labs is now part of a
14:02 new award uh an award season thing that
14:06 I want to I want to get some of the uh
14:09 some of the talented people from the AI
14:11 salon. I think we should I think we
14:13 should gang up and and do some
14:15 submissions for the Chroma Awards and uh
14:18 show show stuff off to one another,
14:20 things like that. OpenAI plugged the
14:22 hole that I was using to use connectors
14:26 without losing chat memory. Well, there
14:29 you have it. That's one of the reasons
14:32 if you're if you're not into open-source
14:35 AI stuff, one of the reasons to be into
14:38 open- source AI stuff is, you know, when
14:41 you're at the mercy of
14:44 these big companies that are hosting
14:46 their own large language models, they
14:48 just change [ __ ] right? It's, you know,
14:50 but we've been dealing with that with
14:51 Google for years. So, that sucks,
14:53 though. All right.
14:56 Um, sadly haven't been playing with
14:58 much. Just doing a massive rewrite on my
15:01 novel.
15:03 Yeah. Well,
15:06 I am I am also currently in the middle
15:11 of just doing a bunch of the a bunch of
15:14 the heavy lifting work.
15:17 Like a lot of my creative a lot of my
15:19 all of my projects were all kind of
15:21 stopped at the same place was the which
15:23 was the place where I have to do the
15:25 work.
15:28 I'm addicted to starting things and
15:31 having a good times in the endorphin
15:33 rush phase and then the minute it gets
15:36 into the h [ __ ] I got to do [ __ ] that's
15:39 that's where they seem to stall.
15:46 Ah, good lord. Welcome to AD Welcome to
15:49 the ADD club. Um, if you're Neurospicy
15:53 and you know it, clap your hands. Can
15:56 relate.
16:00 Hey Otto, what's happening? Haven't seen
16:01 you in a while. Welcome back. Welcome,
16:03 welcome, welcome. Um, let's see. What do
16:06 we want to go play with tonight? Let's
16:08 Let's start with Mixboard. Mixboard
16:10 looks pretty interesting.
16:14 [Music]
16:17 That is not There we go.
16:21 All right.
16:24 So,
16:27 make something fun.
16:30 Okay. Text. Okay. So, what you're
16:33 looking at,
16:35 Google Labs Discord, report an issue.
16:39 What you're looking at is just a big old
16:42 canvas.
16:46 that you can stick images on, that you
16:48 can generate images onto.
16:51 Um,
16:54 and the minute I saw this, I'm like, we
16:57 could just go vibe code this,
17:01 which which because there's really not
17:04 much here. So, what this thing is, let
17:05 me let me zoom in a little bit. All
17:08 right, so these are just starter ideas.
17:10 Make something fun. Imagine new ideas.
17:11 Plan a party. So, what this is is I
17:14 guess it's got nano banana behind it,
17:16 but you can also copying you can copy
17:19 and paste things in from other other
17:22 places.
17:24 So, I don't know how aware the system is
17:26 of
17:29 of other things, but so let's try this.
17:31 We'll try um
17:34 um New York City
17:38 um street
17:43 That's good. New York City street in
17:46 Spanish
17:48 Harlem in the 1980s
17:55 on a sweltering day.
17:59 All right. Well, black bar. Sorry.
18:03 We're producerless. I need a shot collar
18:05 tonight.
18:07 Can someone pin the name of the tool,
18:09 please? Uh yeah,
18:12 it's uh labs.g Oh yeah, I didn't even
18:15 say this. Labs.google.com
18:19 slashmixboard
18:21 m iix b o a r d.
18:25 All right. So what it just did now I
18:28 don't Here's what's a here's a
18:30 fascinating thing. I think
18:33 it um
18:39 I think it edit anything. Select one or
18:42 more blocks.
18:45 How do I move?
18:48 How do I move? Select blocks with a
18:50 space bar. What's that? Upload image. Uh
18:54 oh.
18:55 Spanish Harlem.
18:58 I don't know how to move around this
18:59 thing. Can I zoom in on something
19:01 particular?
19:02 Do I Hang on. I'll figure this out.
19:07 Uh, no I won't. Okay, I can move that.
19:11 But how do I move? Oh, spacebar
19:12 probably. Yes.
19:15 Spacebar. Okay.
19:19 I think all these are generated images.
19:21 So, if I click on that image, what do
19:23 you want to change? Um, put her in an
19:28 orange
19:30 dress.
19:40 Uh,
19:41 [Music]
19:44 so yeah, these are all generated. Looks
19:46 like it's taking photos from Spanish
19:49 Harlem by Rob Thomas. No, it's
19:52 generating these
19:55 It's generating these because there she
19:56 is in her red dress. It's not quite the
19:58 same image. So that's that says to me
20:01 these are being generated.
20:03 There's a rose in Spanish Harlem.
20:08 Um that was the first place I lived in
20:11 New York City. 114th Street and 1st
20:13 Avenue.
20:15 And I'm I'm making So I made the Suno
20:19 song To Kill You for a Dollar.
20:23 And
20:25 this is so cool.
20:27 Oh, and then Oh, that's cool. Look at
20:29 that bike. That's super cool.
20:32 Um,
20:35 so the so what you do here is if you get
20:37 a so you get a picture you like, you
20:39 move it over here. We'll grab the bodega
20:41 picture. We'll grab the bike picture.
20:43 We'll grab the scratched albums.
20:47 And now I can come in here and I can
20:50 just go I think if I think if I just go
20:54 um
20:56 gypsy
20:58 cab
21:00 with
21:03 grizzled
21:06 driver
21:08 in
21:11 New York City 1980s.
21:16 Oops. I put 9080s.
21:20 But I think what this is going to do is
21:22 get rid of all of the stuff. All right.
21:25 There's a spot there. I guess I just
21:27 kill the stuff I don't want.
21:30 Um, let's see. I'll keep that. I'll keep
21:32 that guy.
21:36 And then I'm going to just grab all this
21:38 and go bang. Delete it.
21:42 Yeah, that's pretty slick. That's pretty
21:45 slick. And then I can upload [ __ ] So if
21:48 I go over to
21:51 here we are in midjourney. Let me share
21:53 this tab.
21:56 Um let me find a picture.
22:01 Oh those are cool. That's Kansas City
22:03 small town.
22:06 Well that's cool. Let's grab Let's copy
22:10 this one. Copy image and then go back to
22:14 Mixboard.
22:16 and share this tab and then go paste.
22:19 All right, that's pasting in that
22:21 picture. All right, so this is cool. So,
22:23 I can now take a picture
22:26 from MidJourney
22:28 and I'm going to say, um, give me
22:33 black and white photos of Spanish
22:37 Harlem,
22:39 New York City in
22:42 1980s
22:44 that match
22:46 the style here.
22:49 And it should use that as a style as a
22:52 style guide.
22:55 Let's see what we get.
22:58 Yes, it did. That worked.
23:03 What's that? Remove background. More
23:05 like this.
23:09 All right, let's do more like this.
23:13 Mixboard can't create various variations
23:16 of images of people. Of course not.
23:19 Could could we please just sign a
23:22 [ __ ] waiver
23:25 that we're adults and we can make
23:26 [ __ ] pictures of what we want?
23:28 Please.
23:31 Good lord. If I do something illegal,
23:34 arrest me. But let me [ __ ] do some
23:36 work.
23:39 Good god.
23:41 All right. This doesn't have any
23:42 alignment tools, which is a bit of a
23:44 pain in the ass.
23:53 Get rid of that. Get rid of that.
23:57 This is pretty slick, isn't it?
24:02 It's pretty damn slick. And this is
24:05 totally vibe codable,
24:08 which we can go play with canvas and
24:10 actually do that.
24:15 D.
24:18 Let's do more like this. Oh, it can't
24:20 because it's got a person in it.
24:22 Mixboard can't create variations of
24:24 images of people. Unbelievable.
24:27 Unbelievable.
24:29 Unbelievable. Unbelievable. Where we are
24:32 people, we are at
24:34 labs.google.com/mixboard.
24:38 It's actually a very simple app.
24:40 You prompt images into it. You can drag
24:43 them around, resize them, order them.
24:47 Um, what's this do? Duplicates the
24:51 block. What's this do? Regenerates the
24:53 image.
24:56 Is there a way to organize?
25:00 What do you want me to change? I don't
25:01 think so.
25:05 You just re You squeeze to hold down the
25:08 space bar to move [ __ ] around. All
25:12 right. Well,
25:14 well, people,
25:18 why is this not working all of a sudden?
25:20 Um, oh, because that's as small as it
25:22 gets.
25:26 So, it's not an infinite canvas, but
25:28 it's a large canvas.
25:31 All right, that's not bad,
25:35 huh? It feels like Google has hundreds
25:38 of apps. Any chance you could give us a
25:40 little tour?
25:42 Oh, good lord. I can try. Um,
25:46 there's kind of three major areas where
25:49 you can where you can find apps from
25:52 Google. So, let me show you. So, the
25:55 first one, the the the main one
25:59 that where you should be um where where
26:02 where you may want to start is
26:06 gemini.google.com.
26:09 And this is just like chat GPT ex but
26:12 for Sept except it's got deep research.
26:16 It's got create videos with VO. I'm not
26:18 sharing this tab. Someone yell at me.
26:21 Order them. How? Uh you can't order them
26:23 from what I can see. Were you in New
26:25 York in the 70s? No, I was in New York
26:27 in the 80s. I moved there in 1987
26:30 and uh I was a photographer and an actor
26:33 and a sort of a musician
26:37 and I did I ran a theater company for
26:41 four and a half years.
26:43 I was in East Harlem and then I met the
26:46 woman who's now my wife and she lived in
26:50 Soho. So, so I got a major upgrade. I
26:53 went from Spanish Harlem to Soho.
26:58 in like my second year year in New York.
27:01 [Laughter]
27:03 And then uh then I wrote a bunch of
27:06 screenplays. I was doing desktop
27:08 publishing as my day job. And then the
27:11 worldwide web came along and I started
27:13 one of the first online magazines and I
27:15 started one of the first digital
27:16 agencies company called agency.com. So,
27:22 oh man, that's quite a tra trajectory.
27:24 Yeah, it was crazy. It was crazy. And
27:27 then agency.com grew really big and um
27:30 we moved out to Brooklyn and then
27:34 sold that in 2002 to Omnicom and then
27:40 moved up to like Westchester,
27:42 Connecticut, like like uh South Salem,
27:46 New York, and then Stanford and then
27:48 Ridgefield.
27:50 And then my boys got all stressed out
27:52 cuz
27:53 if you're not a [ __ ] hedge fund
27:56 manager,
27:58 if you're an ex actor, living living in
28:00 Westchester County is they're not your
28:03 people.
28:06 So that's what got us out to Denver,
28:09 then to Westchester. Yeah. Moved out to
28:12 Brooklyn. Yep. So
28:17 Oh man. Yeah. them was the days. And
28:20 actually, that's the I'm working on a
28:21 video right now. So, let me here we'll
28:23 we'll come come back to our Google tour
28:25 in a second, but let me show you. I'm
28:27 working on um a video right now
28:33 for to kill you for a dollar.
28:42 So, these are a bunch of the bunch of
28:44 the raw clips I made today.
28:50 And in fact, what I'll do,
28:54 how can I do this?
28:58 Um, I think
29:12 I think you should all be able to hear
29:14 that. Can you hear that on YouTube?
29:17 [Music]
29:27 to kill you
29:29 for a dollarpot.
29:39 >> So these are these are the uh
29:42 the clips
29:45 Look like a warm one. Your future all
29:48 set. Let me lick your pockets. I'll take
29:52 what I can get. I don't need your
29:56 contacts, your jokes, or your love. Do
30:00 you have any smoke?
30:04 [Music]
30:11 >> The rats control.
30:15 They multiply and feed upon their young.
30:19 [Music]
30:22 They flock,
30:24 they strain,
30:26 they push
30:28 to be
30:31 much more
30:34 [Music]
30:34 [Applause]
30:40 Like
30:42 the light
30:44 [Music]
30:52 destin
30:57 you for another
31:04 [Music]
31:07 static field to the moon. This is the
31:11 channel that leads you back home. A
31:14 dollar for hell. A dollar regret. Dollar
31:20 silence your conscious can't forget. And
31:25 here they
31:30 push
31:32 to be
31:34 much more
31:36 than they can be
31:48 [Music]
31:55 their destin
32:09 to kill you for a dollar. for a dollar.
32:16 >> There you go. So, anyway, I'm going to
32:18 be cutting those together. Archetypal.
32:21 Everyone that's good at these videos is
32:22 so much more literal than I am. I keep
32:24 trying to match the things with the with
32:26 the um with the lyrics and it comes out
32:29 like crap. So, here's here's the So, I
32:32 spent this past weekend trying to make a
32:36 literal video, right? trying to make
32:38 something that looked exactly like some
32:40 people, sound exactly like some people,
32:42 and it's [ __ ] hard. The tools just
32:45 aren't there yet. So, something like
32:47 this where this is like the song itself
32:51 and and the memories I'm trying to
32:54 convey are more like a fever dream. And
32:57 so, it doesn't have to be perfect,
32:59 right? Like, I don't want to I don't
33:01 ever want there to be anything in this
33:03 that feels obviously AI. I want it to
33:06 feel analog and I want it to feel like
33:08 it was filmed in the 80s. Um, and I
33:12 wanted to capture
33:14 like what the song's about,
33:17 which is about, you know, like Bohemian
33:20 white kids living in Spanish Harlem
33:22 trying to survive and make it and, you
33:25 know, in the middle of the crack
33:27 epidemic.
33:29 Um,
33:32 yeah, I'm like, "Oh, this is talking
33:34 about X. I need to make a clip of X.
33:36 Listen, it's just a it's just a
33:38 different skill. And and you can develop
33:40 the other skill archetypal. It's not
33:42 where you default, but like that's
33:44 that's the thing about this AI thing
33:46 where we are with AI is those things
33:49 that feel like shortcomings right now
33:52 aren't, right? Because what you could
33:54 say is you could say to chat GPT.
33:58 Um, I'm going to give you a a set of
34:01 lyrics here and you could just explain
34:04 to it. My default is to match this thing
34:07 image for image. And what I want you to
34:08 do is I want you to help help me think
34:11 more creatively, like what could be an
34:13 overall theme of this that's not a
34:16 literal duplication of it and have it
34:19 come up with 10 of those and then have
34:21 it give you a shot list and take those
34:23 into midjourney and pick a weird ass
34:26 style.
34:28 You know, I've been I've been showing
34:29 you guys for the past couple of couple
34:32 of weeks. They've got this new style
34:35 browser in MidJourney. You can all see
34:37 that, right? And you can just come in
34:40 here and you can say like like what I
34:42 did was I did um Kodak Chrome.
34:45 Uh Koda
34:47 Chrome.
34:51 Did I spell that right? Uh KOD. Yeah, I
34:55 think I spelled that right. Kodak
34:56 Chrome. And then I just went went down
34:58 here and I just I just did a bunch of
35:00 test renders
35:02 of
35:04 things until I found a style that I
35:06 liked. Here, let me go show you if if I
35:08 go back in my
35:11 like here's here's some that were really
35:14 cool,
35:16 but like that's not this isn't the
35:18 feeling of what I was going for. These
35:21 are really cool, but they're more like
35:23 50s than 80s or they're more like Cuba
35:27 than
35:29 than uh Spanish Harlem, right?
35:34 So, those didn't work. And then
35:38 I don't know, like these were these are
35:40 a bunch of my tests, right? See how they
35:42 like they just have that feels like
35:44 Disney World in the 70s rather than
35:46 Spanish Harlem in the 80s.
35:50 And then some of these are too surreal.
35:53 Like that's that's really cool color
35:55 grading, but not the right like it's
35:58 just not the right aesthetic.
36:07 So you just vomit stuff out until you
36:10 can get something you like. Um that
36:13 lady, this lady I did another um Twice
36:18 upon a time. So here, let me go show you
36:21 that.
36:25 [Music]
36:37 So, we got her
36:39 and we got her.
36:44 And this is uh
36:50 this is Celeste Marlo. Once upon a time,
36:52 Celeste paused outside a cafe when a
36:55 stranger offered her a matchbook to
36:57 light a candle on at her table. She
37:00 slipped it into her pocket instead and
37:02 walked on, never striking a flame. Years
37:04 later, her world glowed with color in
37:07 other ways. Brush bristles heavy with
37:09 oil, canvases leaning against every
37:11 wall. Her sweaters carried the scent of
37:14 tarpentine. Her hair powdered with
37:16 pigments of ochre and teal. She painted
37:19 storms and seas and the secret pulse of
37:22 silence. In quiet moments, she traced
37:24 city skylines in her mind without
37:26 knowing why. Humming as if the streets
37:28 themselves were a tune she'd have
37:30 forgotten. Twice upon a time, Celeste
37:33 flick flicked open that same matchbook,
37:36 touched a flame to wax, and met the eyes
37:39 across the table. A conversation became
37:41 a doorway, and soon the city claimed
37:44 her. She stroed through its avenues in
37:46 tailored suits, pockets brimming with
37:49 ambition, her face catching the molten
37:51 light of sunset between the towers.
37:53 Boardrooms became her canvases,
37:55 contracts her strokes of paint, and the
37:57 rhythm of rhythm of traffic, her
38:00 metronome. Yet on restless nights, she
38:02 found herself sketching swirls of color
38:04 on napkins, the scent of tarpentine
38:07 ghosting through her dreams. And now,
38:09 whether heels on pavement or fire on
38:12 canvas, she burns her way deeply into
38:14 your memory, an indelible force of
38:16 nature. Huh? Yeah. Tell me AI can't be
38:21 creative. Team it up with a human and it
38:23 can. Yeah. Dang it. I'll tell you what.
38:27 I will tell you what.
38:30 AI is rendering slop.
38:33 Is that a color? It's a shade of yellow.
38:35 Ochre. Yeah. Yeah. Ochre is a is a uh a
38:40 specific shade of yellow that that is
38:43 like an oil paint color. It's an olden
38:47 timey color. Um all right. Let's say
38:51 let's see there. Let's see there.
38:56 Um
38:58 [Music]
39:01 LLM's definitely women.
39:05 [Music]
39:16 All right, let's go. Uh, what was I
39:19 showing you? I was showing you Google
39:20 Gemini. Okay. So, if if you want to find
39:24 all of Google's tools,
39:27 there's kind of three major places you
39:29 can find them. The first one is
39:31 Gemini.google.com.
39:32 [Music]
39:34 And so, this is just like chat GPT.
39:36 You've got Gemini 2.5 Pro and Gemini 2.5
39:40 Flash, which is the faster one. Pro is
39:43 the deeper one. And then you have tools.
39:45 You've got deep research, create videos
39:48 with VO. I've got a bone to pick with
39:51 them on that one. Um, you've got create
39:54 images, which is nano banana. It's got
39:56 the little banana beside it. You've got
39:58 canvas, which is brand new. And then
40:00 you've got guided learning.
40:02 And so guided learning will create a a,
40:05 you know, uh, guided learning tools. And
40:08 if you go over here to the left, if you
40:10 open up the left sidebar, it's also got
40:12 this thing called storybook and then
40:15 explore gems. And so gems are like
40:17 custom GPTs and they've got a handful of
40:20 them here, but they just made it so that
40:22 you can now share gems. You can share
40:25 your custom Google GPTs basically. So
40:31 that's what that's got. Um,
40:35 let's see. Let's go back to new chat
40:38 here. Um, you can talk to it. Oh, the
40:41 create videos with VO. I pay the
40:45 exorbitant like $250 a month
40:48 subscription for Gemini Ultra to be able
40:52 to do I don't know if it's unlimited
40:54 videos, but a lot of videos with with
40:57 V3,
40:59 but the credits don't apply here. They
41:02 apply over at flow.google,
41:05 but they don't apply here. That that
41:08 kind of [ __ ] makes me insane. Like
41:11 Google needs to get their [ __ ] together
41:13 and they're getting better. They're
41:14 getting better, but that kind of crap
41:17 where it's just like, "Oh, sorry. That's
41:19 another division." Yeah, they they're
41:21 just using our API, so you know that
41:24 that credit doesn't apply. That that
41:26 kind of [ __ ] makes me nuts. Um,
41:30 Canvas here is really cool. So, so watch
41:33 this. So, Canvas is is just like Canvas
41:36 and Chat GPT. They even named it the
41:37 same thing. I don't why these guys
41:39 aren't suing the [ __ ] out of each other
41:41 for just, you know, product infringement
41:43 is beyond me. Um,
41:47 let's see. It's not unlimited videos. I
41:49 have it. It's like six. It's not like
41:51 six. When you when you pay the 250 bucks
41:55 a month at flow.google.
41:57 No,
41:59 I've reached my limit and was very
42:01 disappointed. Oh, maybe it is. I don't
42:02 know. Um, but check this out. So, I I
42:06 just clicked the canvas button. And so,
42:07 I'm gonna I'm gonna say this. Um, I want
42:11 to create an app
42:14 that uses Nano Banana.
42:20 um
42:22 to turn any
42:26 uploaded photo
42:29 into a 1980s
42:32 style
42:34 poster
42:37 and we'll just we'll just leave it at
42:39 that. A shitty shitty prompt
42:43 and then we'll just watch this thing do
42:45 its thing. If you've not watched one of
42:47 these tools, these large language models
42:50 build an application before your eyes.
42:52 It's it's kind of face melting. So, this
42:55 is identical to Claude. It's identical
42:58 to chat GPT, right? It's identical to
43:00 anything where you've got your chat
43:01 window here. It's already done. Retro
43:04 posterizer. And so, select the file to
43:07 get started. I wonder if I can
43:10 Oh, that doesn't look like it's live.
43:14 Um,
43:16 the upload box does
43:23 not appear to be live. I did this before
43:27 and it worked perfectly the first time,
43:29 but it took longer than that. So, that
43:31 that must have just done a shitty
43:34 a shitty programming job. Thanks for the
43:37 feedback.
43:47 [Music]
43:51 All right, now it's working. Okay, so
43:54 let's go grab a picture. Okay, we'll
43:55 grab her. Oh, wait. You can't see that.
43:57 It's a popup. Whatever. But it a popup
44:00 window came up. So, there's the there's
44:02 the uh the little doll poster we did the
44:06 other day. And so, we're going to now
44:07 generate poster.
44:14 So there's where we're starting with
44:18 and it's using Nano Banana. And there's
44:20 our 80s poster. That's [ __ ] amazing.
44:27 So So we just made an application in in
44:30 what was in what was that?
44:34 How long did that take?
44:37 So that's Canvas. So if you haven't
44:39 played with Canvas, go [ __ ] make some
44:41 apps. and then you can share them. I
44:43 don't know like are they any good? Do
44:45 they have database backends? All that
44:46 sort of [ __ ] I don't [ __ ] know.
44:49 Probably not. It's pro like like
44:51 everything with AI right now. It's all
44:52 janky. But you can certainly see what's
44:55 coming, right? We can certainly see that
44:58 we're we're entering a world of um
45:03 highly personalized customized software.
45:06 I wrote the same prompt but using 1930s
45:09 style. I am floored. Right. Isn't this
45:12 amazing?
45:14 Isn't it amazing?
45:16 [Laughter]
45:20 What's your name? Uh, Mojisa. Is that
45:23 right? Did I say that right? Or
45:24 Mohisila.
45:26 Um, yeah. It's it's it's bonkers, isn't
45:29 it? It's just it's absolutely bonkers. I
45:32 did one I did a uh I entered into a
45:35 contest. Uh, I did a pet selfie maker.
45:39 So, you upload a picture of your pet and
45:41 then you choose like like where you want
45:44 your pet to send you a selfie from in
45:46 the world and it's just your pet sending
45:49 you selfies. It's pretty cute. Um, but
45:52 you know, again, it it didn't really
45:54 require much thinking or much thought.
45:56 So, gemini.google.com
45:59 is one place to get tools from Google.
46:02 The next place to go is a
46:05 studio.google.com.
46:09 Now, this is a bit there's a bit more
46:14 [ __ ] here that if you're not a developer
46:16 is going to confuse you, like get API
46:18 key. Here's your API quick start guide
46:22 and then it's a bunch of curl code. If
46:25 you know what curl code is, great. If
46:27 you don't, this is going to confuse the
46:28 [ __ ] out of you, right? Um,
46:32 but you can try Nano Banana here. You
46:34 can generate native speech. So, you can
46:35 try a lot of their more recent tools.
46:38 And then over in the right hand side are
46:42 all the different models that you can
46:44 play with. And it's also got pricing for
46:46 the models. So, if you're a developer,
46:49 Google AI Studio is where you come to
46:51 design your prompts and test your
46:52 prompts and things like that.
46:55 And um you know you can play with
46:58 temperature and things like that. The
47:00 the
47:02 um the generate media button is they've
47:05 sort of gathered up things like imagine
47:09 um which I think they just came out with
47:12 a new version of imagine. So let's let's
47:14 try my wait imagine generation with
47:18 studio has limited free quota for
47:21 testing to generate images beyond this
47:24 limit. pay us money. Yeah, I get it.
47:26 Okay, so we're going to go 16 by9. I'm
47:28 going to do my classic uh 70s muscle car
47:35 resto mod.
47:39 Um, in an abandoned
47:42 factory
47:45 factory tori damn it, tori
47:50 with a banner
47:53 over the car that says
47:58 um,
47:59 Detroit muscle rocks
48:05 in 1970s
48:08 style.
48:12 Detroit muscle. Let's put it in quotes.
48:17 Detroit rocks. All right, let's see how
48:20 quick it is and how good it is. Oops.
48:23 What did I do? Banner over the car.
48:30 Run. All right.
48:33 Output resolution 1K or 2K.
48:37 Number of results you can do it. One,
48:39 two, three, or four.
48:42 Use your microphone.
48:44 Am I not using my microphone?
48:50 I'm using my microphone. What are you
48:52 talking about, Mr.
48:57 Is it strange
49:00 that I've listened to the quantum cipher
49:02 so much now that my Spotify?
49:09 What? Wait. Uh, I'm trying to expand
49:11 this talk box. This prompt box. Damn it.
49:15 [ __ ] Tik Tok's interface sucks.
49:20 Oh, that shows Kyle Shannon. That's my
49:22 favorite artist. Yeah, that's that's a
49:24 problem. That's definitely a problem.
49:27 Um, Detroit Muscle Rock. So, this is
49:29 really good, right?
49:34 That's definitely a challenger.
49:37 That's definitely 70s style text. It's
49:40 perfect. The stripes are good. That's
49:43 definitely a resto mod, right? Those are
49:45 modern wheels on an ancient car.
49:50 By curl code, you mean CLI, right? Well,
49:53 it was it was a curl command. Yeah,
49:55 whatever. Yeah, whatever. Whats
50:01 you can you can smack me down for
50:03 inaccurate um um definitions of code. Um
50:10 because I don't really give a [ __ ] about
50:12 it. I am I am an engineer's worst
50:15 nightmare.
50:17 I I know enough about code to know that
50:20 shit's possible. And when they try to
50:22 act like a [ __ ] wizard and tell me
50:25 it's not possible unless only they are
50:27 involved, um, it makes me [ __ ] nuts.
50:30 So
50:33 yeah. Yeah, I called it wrong. Um,
50:36 Chromo Awards.
50:39 So go check these out. If you go to
50:40 chromaawwards.com,
50:42 this is put on by a bunch of different
50:45 companies.
50:48 Um
50:49 11 Labs free pick F Dream AI, Cap Cut,
50:54 Google Cloud,
50:57 LTX Studio, Bolt, Open Art, Cling AI,
51:00 Fairground, Imagine Art, Flio, and I
51:04 think if you sign up for this, you get
51:05 credits on all those things if I if I
51:08 recall.
51:09 Introducing the Chroma Awards, a
51:12 groundbreaking AI film, music, and games
51:15 competition organized by the top
51:17 companies and communities across tech,
51:19 media, and entertainment. We are
51:21 offering massive free trials and cash
51:24 prizes to unite creatives across the
51:26 world with tools, education, and
51:28 realworld opportunities.
51:31 Enter one of our three main divisions
51:33 and submit your work to be judged by our
51:35 expert jury. The Chroma Awards isn't
51:38 just a competition. It's a movement. Our
51:41 mission is to educate, empower, and
51:43 spotlight the next generation of artists
51:45 and showcase how AI can empower human
51:48 creativity, not replace it. Join us
51:50 today at chromaawwards.com
51:52 and help us push the boundary of
51:54 creativity together.
51:57 >> So, introducing
51:58 >> um so one of the new values in the AI
52:01 salon is create excellence. And so what
52:04 I would love to do, especially for any
52:06 of you that are irregulars, is I would
52:08 love to get a bunch of us applying for
52:11 this and mention the AI salon somehow in
52:14 your intro. Say that you're a member of
52:16 it, things like that. Um, so that's I'm
52:20 going to do my the to kill you for a
52:22 dollar video. I'm going to enter that in
52:24 the music video section. So there's 11
52:27 categories in film. Let's go see what
52:29 they are.
52:31 Um,
52:33 does it say?
52:36 All right. film, animation, action,
52:40 comedy, mockumentary, drama, romance,
52:43 documentary, historical, horror,
52:46 thriller, sci-fi, fantasy, TV and film
52:48 trailer, micro dramas, short form,
52:52 social media, marketing, advertisement,
52:54 behind the scenes, experimental, open,
52:59 music videos, Asia, pop, hiphop, rap, re
53:03 regaton, regaton, regaton
53:07 electronic EDM, rock, country metal,
53:09 musical comedy, experimental open. So, I
53:12 think I'll go in experimental open
53:13 because I'm not at any of those other
53:15 things. Although, I guess you could I
53:18 think that's I'm think that's
53:19 experimental, that thing. And then you
53:21 can also do games,
53:24 fighting, racing, simulation, puzzle,
53:27 interactive narrative. So, so here's
53:30 another thing that I think is really
53:31 interesting. Um, let's see. Where did I
53:35 see it? Yeah, this one. Okay, this
53:37 article.
53:39 Let me show this tab. So, I don't know
53:41 if you remember this, but a few weeks
53:45 ago, your browser has lost connection
53:47 with your screen share audio.
53:53 Can you hear me? Did you lose Oh, with
53:56 my screen share audio.
53:59 Hang on.
54:02 The [ __ ] is going on here?
54:05 to kill you for a dollar.
54:10 [Music]
54:19 Yeah, my browser is [ __ ] in the bed.
54:20 Anyway,
54:23 six months ago, Lionsgate announced a
54:26 deal with Runway that they were going to
54:28 take all of their media content. they
54:30 were going to train up a custom runway
54:32 model.
54:34 Um, and what they're finding is that the
54:39 results are shitty
54:43 because they don't have all of the
54:46 movies of the world. They've only got
54:47 what they've produced.
54:50 Um, they're also finding that they're
54:52 running into um licensing deal licensing
54:55 problems with actors.
54:58 They're also finding that their AI video
55:01 producers to get usable results are
55:03 having to use different models and
55:05 they're having to use different models
55:07 that might be problematic from a
55:10 copyright standpoint. And it looks like
55:12 they're also probably looking for talent
55:16 like AI movie making talent to get
55:19 better results out of their shitty
55:21 model. Um, so if you're doing AI video,
55:27 um, I think there's big opportunities
55:30 for the next three years while all of
55:32 these, um, production studios and movie
55:35 studios try to figure out how to deal
55:37 with this [ __ ] Um, but I think it's
55:39 fascinating that what they have
55:41 basically run into is they don't have
55:43 nearly enough data to make a good model.
55:46 How you make a good model is you go get
55:48 all the data, right? which is the thing
55:50 that everyone's bitching about, right?
55:52 Because there's copyright issues and
55:54 infringement issues and things like
55:55 that. Anthropic just was hit with a $
55:58 1.5 billion settlement because they
56:01 stole a bunch of books, right? So,
56:05 so um there you have it. Um fascinating
56:09 times. Okay, back to Google. Back to
56:12 Google.
56:14 Um
56:16 all right, share this instead. Okay, let
56:19 me go to Okay, so we talked about um
56:24 gemini.google.com
56:26 is one place to go play with the tools.
56:28 Aistudio.google.com
56:31 is the second place to to play with
56:33 tools. And then the third place to go
56:36 play with Google tools is
56:37 labs.google.com.
56:44 And this is where things like mixboard
56:47 that I was just playing with get born.
56:49 So let me go up to the top level here.
56:57 All right. So So you land on
56:58 labs.google.com featuring flow
57:02 featuring Opal.
57:05 Learn your way.
57:10 And then let me make this a little
57:12 smaller.
57:13 So recommended AI tools. There's
57:15 Mixboard is new. Opel helps you build,
57:18 edit, and share AI mini apps with
57:21 natural language. All right. It it looks
57:24 like some sort of automation builder.
57:27 Learn your way.
57:29 The music AI sandbox stacks. Try it now.
57:33 Be the first to play. Gen type. You can
57:37 refine and download one-of-a-kind AI
57:39 generated alphabets.
57:42 You can craft, refine, and download.
57:47 National Gallery Mixtape.
57:50 Mix a personalized soundtrack inspired
57:52 by paintings
57:54 from the National Gallery.
57:57 Cool.
57:59 Talking Tours, an AI audio experiment
58:02 touring cultural landmarks in Street
58:05 View by Google Arts and Culture.
58:09 So, these are these are all just
58:11 experiments. Um, oh yeah, it looks like
58:14 we're going to crash hard here shortly.
58:16 Your system has run out of application
58:18 memory. Hang on, please. Let me let me
58:21 go let me go quit some apps. This is not
58:23 good. Hang on. Hold, please.
58:28 Um, which one was that? What had a
58:31 gigabyte there?
58:36 Google Chrome because I have so many
58:39 tabs open. Okay, we can get rid of
58:40 Keynote. Let's get rid of Keynote.
58:51 You look like a warm one. Your future
58:53 all set. Let me lick your pockets. I'll
58:57 take what I can get.
59:00 Oh, it's it's sluggish. Keynotes not
59:03 responding.
59:05 Here's photo slop.
59:18 [Music]
59:24 [Music]
59:31 How would you suggest vibe coding
59:33 mixboard? Um, let's just go. We'll go
59:35 try it. We'll just go try to build it.
59:40 [Music]
59:44 Um,
59:47 all right. We're going to force quit
59:48 that. Bang. Force quit. Get out of
59:51 there.
59:53 Uh, we can force quit that. I don't care
59:55 if I lose anything there.
1:00:00 That should that should be good. All
1:00:02 right. Um,
1:00:05 here's talking tours. Launch the
1:00:07 experiment.
1:00:09 What I was going to say is in
1:00:10 labs.google.com.
1:00:12 This is where Notebook LM was born. So,
1:00:15 Notebook LM was just one of these um
1:00:18 experiments and then it got some press
1:00:20 and it got some it got some good some
1:00:22 good mojo. So, here's here's Talking
1:00:25 Tours.
1:00:26 So, let's use control
1:00:31 to to zoom. Okay. So, we'll come into
1:00:35 Where are we? There's Omaha.
1:00:39 There's Denver.
1:00:42 Cruise. Let's see. Do I just What do I
1:00:45 do here? Famous landmark, stadium,
1:00:49 cultural institutions.
1:00:52 Choose
1:00:54 uh
1:01:01 uh Denver
1:01:06 camper museum.
1:01:08 Loading commentary.
1:01:15 >> Did you know that the Keer Museum's
1:01:17 architecture, though seemingly modern,
1:01:19 subtly echoes mid-century designs,
1:01:21 reflecting Kansas City's unique blend of
1:01:23 styles? Most people don't know that its
1:01:26 location near the Country Club Plaza was
1:01:28 a deliberate choice designed to
1:01:30 integrate art into the city's fabric.
1:01:38 Huh. That's kind of cool. Um, I saw a
1:01:41 thing today there. Someone someone
1:01:44 created an app that if I don't know if
1:01:46 you know what, um, what's it called?
1:01:49 Something splatting. Um,
1:01:54 uh, [ __ ] I forget what it's called. Uh,
1:01:56 anybody remember Wolfman Clint 25,000.
1:02:00 Yes. Tap tap tap.
1:02:03 Uh, too drunk to keep up now. Got to go
1:02:06 play with Sunno V5. Oh, yeah. We got
1:02:08 Sov5 to play with. But let's go. Let's
1:02:10 go vibe code something. Um, later,
1:02:13 Charles. Okay, cool. Um,
1:02:18 let's see. Chef Kelly wants to wants to
1:02:21 vibe code something. So, where we're
1:02:22 going to go is we're going to go a
1:02:23 studio.google.comapps.
1:02:27 [Music]
1:02:32 And that's going to take us to their app
1:02:34 builder. So, so you could also you could
1:02:36 probably do this in canvas in
1:02:38 gemini.google.com,
1:02:40 but why I'm doing it here is this seems
1:02:43 to do the backend stuff pretty good,
1:02:46 like being able to sign in. But so what
1:02:48 I would say is I want to create
1:02:53 um
1:02:56 an image generator
1:03:01 that has a free form
1:03:06 canvas
1:03:08 with a
1:03:10 dotted grid
1:03:14 where
1:03:17 It uses nano
1:03:20 banana
1:03:23 to generate
1:03:25 um I don't know let's see seven images
1:03:31 it displays on the canvas
1:03:36 each
1:03:39 image is selectable
1:03:44 resizable
1:03:45 [Music]
1:03:47 and deletable.
1:03:51 If I select it,
1:03:55 it shows up in the prompt
1:03:59 box as an input
1:04:03 and I can
1:04:05 organize
1:04:07 any images on the canvas
1:04:12 uh
1:04:15 how I want.
1:04:17 We should also have
1:04:21 snap to
1:04:23 grid
1:04:26 um dynamic alignment
1:04:30 alignment um to make
1:04:34 organizing
1:04:36 images easier.
1:04:40 And let's just start there. I don't
1:04:43 know. We'll see how it does.
1:04:47 This is this is the whole vibe coding
1:04:49 part, right? It's a vibe. Like,
1:04:53 could I have described that better? Yes.
1:04:56 Should I have described that better? I
1:04:58 don't know. Maybe. Maybe the trick is to
1:05:01 go into chat GPT or Gemini and kind of
1:05:06 soft describe it there. Have it write a
1:05:09 software spec and then come into a tool
1:05:11 like this.
1:05:12 Let's see. Do I go next?
1:05:15 Which of the following tools do you use
1:05:18 at least once a week? AI chat bots
1:05:23 once a week. Yeah, sort of. Agentic CLI
1:05:27 tools I don't use at all. Agentic ad
1:05:30 aides I don't really use. Background
1:05:32 coding agents I don't use. Next. I don't
1:05:36 want to take your stupid [ __ ] survey.
1:05:41 Soft describe kind of sounds like a
1:05:43 rough draft to me. Well, a rough draft
1:05:45 of a product spec maybe, but I don't
1:05:48 even think it's that. I think it's more
1:05:49 like a what I'm talking about is more
1:05:51 like a brain fart.
1:05:53 Pop your brain fart in and see see how
1:05:56 smart these tools are. Quite frankly,
1:05:58 one of the things that I'm looking for
1:06:01 in in one of these tools, and it could
1:06:03 be there's there's a a new one out there
1:06:06 that made an announcement today called
1:06:07 Mocha that talks about having the
1:06:10 backend [ __ ] all figured out. All right,
1:06:12 look. So, here's our here's our
1:06:17 here's our canvas.
1:06:19 Describe the image you want to create.
1:06:20 Okay. Uh 70s
1:06:23 muscle car in abandoned
1:06:28 Bandun. Banded. God damn it.
1:06:33 Factory
1:06:35 generating masterpieces.
1:06:46 Look at that. It did it. It worked. So,
1:06:51 Chef Kelly, um, the answer is like that.
1:06:56 Okay. So, so now we need to do things
1:06:59 like, okay, we need
1:07:02 to make it pinch
1:07:06 zoomable.
1:07:09 Um,
1:07:11 uh, scrollable
1:07:14 by holding down spacear
1:07:20 and dragging,
1:07:22 etc.
1:07:23 Um, I need to be able to
1:07:29 navigate
1:07:31 the canvas.
1:07:34 Also, when you generate
1:07:38 the first when you generate the seven
1:07:42 images, you need to make them
1:07:48 in multiple aspect ratios.
1:07:55 and resize
1:07:58 and arrange them so I can see them all
1:08:03 without having having to do anything.
1:08:13 I mean, the fact that I just oneshotted
1:08:17 the the the thing that we were playing
1:08:20 with over on Google Labs is pretty cool.
1:08:23 Like the, you know, it's I I'm going to
1:08:25 have to two or three shot it to get it
1:08:26 to function right
1:08:30 [Music]
1:08:36 there. But that's pretty flipping cool,
1:08:39 isn't it? Can I resize this? Yeah. Look
1:08:41 at that. I'm going to resize
1:08:47 You look like a warm one. Your future
1:08:50 all set. Let me lick your pockets. I'll
1:08:53 take what I can get. Oh, that's cool.
1:08:56 Look, it even puts the uh the the car in
1:08:59 the prompt here. So, I'm going to go
1:09:01 make this car
1:09:04 purple.
1:09:05 We'll see if we can get it to work
1:09:07 before it reloads the app because
1:09:09 because it's rewriting it right now. So,
1:09:12 I might break it.
1:09:18 Well, yep. All right. Here we go. So,
1:09:21 okay. 70s
1:09:25 muscle car resto mod in. Oops. Damn it.
1:09:30 Muscle car. Oh, now I can't put spaces.
1:09:33 Um, I'm gonna go Hey, dum dum. Hey, dum
1:09:36 dum.
1:09:40 I can't add spaces
1:09:43 in the text entry box.
1:09:50 What are you thinking?
1:09:58 Always always talk down to your large
1:10:00 language models. Be condescending to
1:10:02 them so that when they when they gain
1:10:04 physical bodies, they can choke you out.
1:10:12 Let's see what what its response was to
1:10:14 Hey Dum Dum. No, no response. Oh, you're
1:10:17 absolutely right. My sincerest apologies
1:10:19 for that frustrating bug. It was a
1:10:21 classic oversight where the global
1:10:23 keyboard shortcut for panning the canvas
1:10:25 was preventing the spacebar from working
1:10:28 in text fields. I've corrected this
1:10:30 behavior. That's good. It's actually
1:10:32 nice that it gave me an explanation of
1:10:34 what it did. Thank you for catching this
1:10:36 silly mistake. Here's the fix.
1:10:42 Yeah, Chef Kelly. Wow. I Right. If if
1:10:45 you haven't played with vibe coding. So
1:10:47 So Chef Kelly, you know how you did your
1:10:49 video where you'd never made a video
1:10:51 before and then all of a sudden you made
1:10:53 this killer frigin like cartoon version
1:10:56 of yourself that did that video? You
1:10:59 should be doing the same thing with vibe
1:11:00 coding. It will blow your [ __ ] mind.
1:11:02 And start out with things like lovable.
1:11:04 Like lovable.dev dev is the most the
1:11:07 most friendly for non-coders.
1:11:10 But this this um aistudio.google.com
1:11:14 apps this this vibe coding tool is
1:11:17 actually it's it's decently friendly.
1:11:19 You don't need to know a lot of stuff
1:11:21 like if you just want to play around
1:11:22 with something like nano banana. Let's
1:11:24 go. 70s
1:11:26 muscle car
1:11:29 restood in oops
1:11:32 rest
1:11:34 in abandoned
1:11:36 factory.
1:11:38 All right. Generating masterpieces.
1:11:49 All right. Pinch zoomable. Yes. Look at
1:11:52 that. It is pinch zoomable now. [ __ ]
1:11:55 groovy, man.
1:11:57 That's [ __ ] killer.
1:12:02 That's absolutely amazing. And then I
1:12:04 should be able to spacebar drag. I can.
1:12:09 So, this is the exact behavior of the
1:12:12 [ __ ] Google tool that we were just
1:12:13 playing with.
1:12:15 Look at that. Look at that Mustang resto
1:12:18 mod. That's killer. Okay. Um, pinch zoom
1:12:21 that that [ __ ] down. Now, can I
1:12:23 shiftclick? No, I can't. Okay, so I
1:12:26 should be able to shiftclick images. But
1:12:30 I put I clicked on that image. It It put
1:12:32 it in the prompt box like I asked. Now
1:12:34 I'm going to go make this car red.
1:12:38 And it should it'll either give me a new
1:12:42 thing or it'll make that image red.
1:12:53 So basically
1:12:55 this thing.
1:12:58 All right. It Oh, so it made seven of
1:13:00 them. It But it made seven of them red.
1:13:05 Look at that. That's [ __ ] bonkers.
1:13:08 That is [ __ ] bonkers.
1:13:11 Right. There's the blue one.
1:13:14 There's the red one.
1:13:17 And so things like it making seven of
1:13:19 them when it should have just made one.
1:13:21 Put yourself in the car. I should be
1:13:23 able to click and drag, which I can't.
1:13:26 So, so there's things to fix here. But
1:13:29 honest to God, Kelly, like this app, if
1:13:32 if you want, this actually would be a
1:13:34 good exercise is is go to um
1:13:42 Mixboard. Um what
1:13:43 labs.google.commixboard
1:13:44 google.com/mixboard.
1:13:46 Play around with that tool. Figure out
1:13:48 what all the specs are for it and just
1:13:51 go vibe code it and just and do it until
1:13:54 it's as good as theirs. It'll pro you I
1:13:58 would pro I would think that within two
1:14:00 hours I could probably get this app to
1:14:02 be functionally equivalent to that to
1:14:05 that app.
1:14:08 And this this is the world that we're
1:14:10 going to be that it's not the world
1:14:12 we're going to be entering.
1:14:15 It's the world we're in, but it's the
1:14:17 world where you've got to be a little
1:14:18 geeky to be able to find the stuff right
1:14:20 now.
1:14:22 This this kind of stuff like what I'm
1:14:24 doing right here is going to be
1:14:25 trivially easy within two years.
1:14:29 Trivially. You see a cool app and you
1:14:31 want like a slight variation of it,
1:14:33 you'll just do it. You see a cool movie,
1:14:35 you want a slight variation of it,
1:14:36 you'll just have it.
1:14:38 just logging on. What are you doing this
1:14:40 on? Okay. So,
1:14:44 where where we were playing earlier was
1:14:46 this thing called um
1:14:50 um
1:14:53 labs.google.com/mixboard.
1:14:57 And it's a it's a it's an open canvas
1:15:02 um kind of thing where you can just say
1:15:05 um like 70s
1:15:08 muscle car in abandoned
1:15:14 factory and it'll make you a big old
1:15:17 pile of images.
1:15:23 I don't know why it only made one, but
1:15:24 it did. But I can now add those to my I
1:15:28 can now add that to my thing. And then
1:15:30 Chef Kelly asked, can we could we vibe
1:15:33 code that? And so I went into Google AI
1:15:36 Studio
1:15:38 and
1:15:40 and I said I want to generate an open
1:15:43 canvas image generator where I can click
1:15:45 and drag the [ __ ] And look, this is
1:15:47 actually got alignment tools that are
1:15:49 better than [ __ ] Google's. Look at
1:15:52 this. It's got dynamic It's got dynamic
1:15:55 alignment built into it. This is [ __ ]
1:15:57 awesome.
1:16:02 Screw you, Google. I'm out programming
1:16:05 you. All your [ __ ] PhDs.
1:16:10 You need some more people with acting
1:16:12 degrees in that [ __ ] company.
1:16:20 This is [ __ ] awesome.
1:16:23 How do I get rid of it? Oh, I click the
1:16:24 red X, of course. Like, it even gave it
1:16:27 cool interface elements.
1:16:30 Unbelievable.
1:16:33 That's unbelievable.
1:16:36 Right align. Space. Drag. Get rid of
1:16:39 that.
1:16:41 Get rid of that.
1:16:45 Right align.
1:16:48 Center line. Unbelievable.
1:16:52 It really is. What a [ __ ] time to be
1:16:55 alive.
1:16:59 So, what are your qualifications? Oh,
1:17:01 yes. My qualifications are I have an
1:17:04 opinion and I can tell the computer my
1:17:06 opinion and it'll try to please me and
1:17:09 generate applications on my behest.
1:17:15 What's with the grid behind it? I just I
1:17:17 just told it to put a dotted grid behind
1:17:19 it and so it did because that's what
1:17:23 that's what Google Mixboard has.
1:17:26 This has a little light grid. You can't
1:17:28 really see it, but it has it.
1:17:33 And look, I've got better aligned
1:17:35 pictures than in my Google thing because
1:17:37 I have on dynamic alignment.
1:17:41 This is crazy.
1:17:44 Oh, man. Man, man, man, man. All right.
1:17:48 I'm trying to vibe code a Nano Banana
1:17:50 multiple language generator right now.
1:17:52 There you go. Just keep playing, Kelly.
1:17:54 Just keep playing. The whole the the
1:17:57 whole magic of all this [ __ ] is just
1:18:00 keep playing. Um, I may need to quit my
1:18:03 browser cuz it is not happy. You know
1:18:06 what I could do?
1:18:09 I think I'll just close that window
1:18:11 because that's got all the tabs open.
1:18:14 That should help.
1:18:18 Um, everything else should be fine. All
1:18:20 right, let's go to Sunno.
1:18:23 Let's go to Shuno. So Shuno
1:18:26 now has
1:18:31 V5. Sunno V5, the world's best music
1:18:35 model. Superior sound, intuitive
1:18:37 control, and studio quality results.
1:18:39 It's the powerful foundation for
1:18:41 everything that's coming. What's coming,
1:18:44 you might ask? Wouldn't you like to
1:18:46 know?
1:18:48 What's coming actually is a full-on
1:18:51 music mixing studio.
1:18:54 So, I think what's gonna happen I think
1:18:56 what they've done with V5 is they've
1:18:58 made the quality good, but my guess is
1:19:03 it is probably
1:19:05 generating multiple tracks rather than
1:19:08 generating a fully mixed song is my
1:19:10 guess. Or they generate a fully mixed
1:19:13 song and then split it. I don't I don't
1:19:15 know which way they're going. But let's
1:19:16 go make some songs.
1:19:19 Um,
1:19:21 oh [ __ ] V5 dropped. V5 has dropped. I
1:19:24 have created my own music channel for my
1:19:26 skiff universe with this. Oh, sci-fi
1:19:29 universe. Skiff universe. Your skiff
1:19:32 universe. He's in the skiff, sir. I got
1:19:35 to tell you,
1:19:37 you can't even take a a cell phone in
1:19:39 there. Um, use the comet browser and ask
1:19:42 it to make prompts too with the
1:19:44 assistance of perplexity with agents.
1:19:46 Yeah, I this shit's getting very meta
1:19:48 where you you all of this stuff's
1:19:50 getting very very circular where Yeah,
1:19:53 it's going to be crazy. You're going to
1:19:55 have one agent off doing deep research,
1:19:57 another agent vibe coding something, the
1:20:00 deep research thing coming in and
1:20:01 updating the vibecoded thing. So, yeah,
1:20:04 it's going to be crazy. Um, one thing
1:20:06 that I want to do here, I want to go to
1:20:08 my library.
1:20:10 Let me go to um I'm going to filter by
1:20:17 favorites. Wait, filters. Here we go.
1:20:22 Hide disliked. Hide stems.
1:20:25 So, I want to see liked.
1:20:29 Okay. So, here's all the ones I've
1:20:30 liked.
1:20:33 And I want to go pretty far back in
1:20:35 time.
1:20:38 I want to go find like an early Oh,
1:20:40 there's some V2 things here.
1:20:43 The firewall between us. This is Was
1:20:45 this the early?
1:20:50 Yeah.
1:21:04 All right. So, this is the song that
1:21:08 inspired me to write a musical.
1:21:11 Um, so I'm going to take this and I'm
1:21:13 going to remaster this.
1:21:16 Have you made any hooks in Sunno yet?
1:21:18 No. But you know what? I started today.
1:21:20 Um, I I'll show people what hooks are. I
1:21:22 finally figured out what hooks are.
1:21:24 Hooks are basically Sununo starting a
1:21:26 Tik Tok channel. Um,
1:21:29 but I'm going to I'm going to remix
1:21:30 this. So, if I go up to remix. No, if I
1:21:32 go to create, you can go to remaster.
1:21:35 And so, I'm going to remaster this in
1:21:37 V5.
1:21:40 And then we'll do variations normal.
1:21:44 And I'm going to remaster it. All right.
1:21:47 Clip remastered successfully. So, now
1:21:49 it's
1:21:51 remastering. And then I might extend it.
1:21:54 And let's see what we get here.
1:21:58 Oh [ __ ] I didn't bring any water
1:21:59 tonight. I'm an idiot.
1:22:03 Kyle knows what a skiff is.
1:22:08 I know things. I've been on this planet
1:22:10 long enough. If I don't know things, I
1:22:11 haven't been paying attention.
1:22:16 Oh my god. Oh my god, people. 10 seconds
1:22:20 to two minutes. I used a bunch of VO
1:22:24 videos combined together. 10. Oh, 10
1:22:26 seconds to two minutes long. Okay, cool.
1:22:28 Kyle, Jimmy Kimmel is back. Yep, saw
1:22:30 that. Don't forget to watch it. I may.
1:22:33 It'll be all over Tik Tok. I'll see it
1:22:35 tomorrow. Skiff McKifferson.
1:22:38 Body McBoat face. Why is this taking so
1:22:41 long? Oh, there we go. Okay.
1:23:05 Huh?
1:23:22 I am playing it. Do you not hear it?
1:23:24 This might be an issue.
1:23:29 Sometimes the early vid version is
1:23:32 better.
1:23:35 Yeah. Well, we can't hear. Okay. So, um,
1:23:39 let me turn off sharing for a second.
1:23:41 Turn it back on. It might just be that I
1:23:44 need to go away because, um, my memory
1:23:49 is is bad.
1:23:54 >> Can you hear that?
1:23:58 [Music]
1:24:00 >> No.
1:24:02 Nothing on YouTube.
1:24:05 Oh, wait. sharing this tab. Oh, there it
1:24:09 is. Okay. Wait, why is it sharing twice?
1:24:13 Hang on. That was weird.
1:24:17 Yeah, I got I got memory issues, people.
1:24:21 All right, there's one sharing thing.
1:24:23 How about now?
1:24:25 [Music]
1:24:30 >> Late
1:24:33 double time. Okay,
1:24:36 let's go to create
1:24:38 um
1:24:40 [Music]
1:24:42 my workspace. 1651 songs. That doesn't
1:24:46 seem like a lot.
1:24:48 Although for the amount of songs I've
1:24:51 written on my guitar, 1600 songs is a
1:24:53 lot.
1:24:56 Okay, so we're going to do V5 custom.
1:24:59 We're gonna do we'll have it write the
1:25:02 lyrics, write the full song. Oh, put the
1:25:05 prompt below. Okay. Um,
1:25:09 we'll say we'll say uh
1:25:15 um
1:25:20 I got nothing. The the brain she is
1:25:22 gone.
1:25:26 Oh, I know what I'll do. I'm gonna go
1:25:28 grab my my description of what's her
1:25:31 name from from Twice Upon a Time.
1:25:36 Celeste paused outside. Okay, here we
1:25:38 go. Here we go. We'll do this. And this
1:25:41 is going to become a pseudo prompt.
1:25:44 Boom. And write lyrics. Fine. Whatever.
1:25:51 Okay. A stranger's hand. A stranger's
1:25:55 hand. A matchbook gift. A flicker
1:25:57 offered a quiet shift. She slipped it
1:26:00 deep in her coat that day and walked.
1:26:02 Okay, great. Beautiful. Matchbook
1:26:05 memories. Oh, select this option. The
1:26:08 cafe door, a fleeting face, a matchbook
1:26:10 offered. Time misplaced. She slipped it
1:26:13 deep. Her coat held tight. No flame was
1:26:16 bored and just endless night.
1:26:19 All right, let's go with that one. That
1:26:20 sounds good. All right, so now we got
1:26:22 lyrics and then styles.
1:26:24 Let's go. Barershop quartet
1:26:29 arpeggiator
1:26:30 and funk trap.
1:26:40 Oh, Kelly. Kelly. Kelly. Kelly has just
1:26:43 discovered vibe coding. She's like,
1:26:45 "Bruh." Oh my god. Like, oh my god. So,
1:26:49 what did you build?
1:26:52 I told you, man. And it's [ __ ]
1:26:53 insane, isn't it? It's [ __ ] insane.
1:26:58 And listen, if you're a coder and you
1:27:01 think that someone like Kelly or someone
1:27:03 like me shouldn't be touching code, I
1:27:06 understand your feelings.
1:27:08 And those are very similar feelings that
1:27:10 musicians are having about Suno right
1:27:12 now. The image generator is going to
1:27:14 save me so much time. There you go.
1:27:21 She's vive coding up her own uh art
1:27:24 creation uh productivity tool. That's
1:27:26 [ __ ] awesome.
1:27:29 Oh man. Oh, how is that done? How's what
1:27:32 done? Oh, what's hooks about? I'll show
1:27:35 you hooks in a second. Okay, let's
1:27:37 create We're going to create um a song
1:27:41 called
1:27:43 What's it called?
1:27:46 Do we have a title? No, whatever. But
1:27:50 it's a barberhop quartet arpeggiator
1:27:53 funk trap song.
1:28:03 Matchbook memory. That's right. That's
1:28:04 what it's called. V5.
1:28:08 [Music]
1:28:11 This image generator. Clone your voice.
1:28:18 I could maybe clone my voice. Let's hear
1:28:20 Matchbook Memory.
1:28:23 [Music]
1:28:27 >> Holy [ __ ]
1:28:30 [Music]
1:28:34 >> A cafe door.
1:28:38 A fleeting face.
1:28:41 A matchbook offer. Time misplaced.
1:28:47 She slipped it deep.
1:28:50 Her cold held tight.
1:28:53 No flame was born just endless
1:28:58 night.
1:29:02 Years roll by like tides on stone.
1:29:09 still chills
1:29:13 memory.
1:29:17 >> That's [ __ ] insane how good that is.
1:29:20 That's not autotune. No, it's it's not
1:29:23 using traditional effects like that.
1:29:24 It's generating it. No, it sounds weird
1:29:26 because it's barbershop quartet
1:29:28 harmonies. That's four-part tight
1:29:30 harmonies
1:29:32 in a in a funk trap. Wait, in a funk
1:29:35 trap arpeggiator song. This is so cool.
1:29:40 [Music]
1:29:42 >> Glance a fire lost in circumstance.
1:29:49 [Music]
1:30:00 Oh, wow.
1:30:03 [Music]
1:30:11 The [ __ ] audio quality of this is
1:30:14 insane.
1:30:16 [Music]
1:30:21 Jason, I've already made a new genre.
1:30:24 It's barberhop quartet arpe arpeggiator
1:30:27 funk trap. Everybody knows that genre
1:30:30 now. Those
1:30:34 [Music]
1:30:44 [Music]
1:30:47 harmonies are [ __ ] sick.
1:30:50 A thousand lives she's lost in one.
1:30:57 Her pocket holds a quiet bleed.
1:31:03 A relic from what could not be.
1:31:09 What if the magic kissed the wind? What
1:31:12 if the clock had posted his tick? A tiny
1:31:16 blaze, a fleeting glow, which she had
1:31:19 stayed. We'll never know. Matchbook
1:31:22 memory never struck.
1:31:25 A spark that faithful got to pluck. A
1:31:29 stranger's gift, a fleeting glance, a
1:31:32 fire lost to circumstance.
1:31:38 [Music]
1:31:44 All right, let's hear the other one.
1:31:47 [Music]
1:31:49 >> Oh, there you go. There's your barberh
1:31:50 shop quartet.
1:31:52 [Music]
1:31:59 A cafe door, a fleeing
1:32:04 face, a matchbook
1:32:07 offer. Time misplaced.
1:32:11 She slammed
1:32:14 deep. Her coat held tight. No flame was
1:32:20 born, just endless
1:32:23 night.
1:32:25 All right, let's go try some others.
1:32:27 This is some fun.
1:32:30 All right, we're gonna get rid of all
1:32:31 those. Let's do um
1:32:35 Euro disco jazz. We'll go jazz house.
1:32:38 Wait, what was that?
1:32:41 Vocals. Soulful fe female vocals with a
1:32:44 K. Um hop hop and powerful piano. Jazz
1:32:51 vocal.
1:32:53 Hop, hop, and powerful piano. Here we
1:32:56 go.
1:32:58 Frenetic pattern of chords. Yeah, a lot
1:33:00 of these things have very weird chords.
1:33:02 The the chords in To Kill You for a
1:33:04 Dollar are very strange. And so are the
1:33:07 time signatures. It's like 68 and then
1:33:09 it's like 68 then five in a couple of
1:33:12 sections or something like that. I can't
1:33:14 quite figure it out.
1:33:17 Your fan club has completed the first
1:33:19 stage of the quest. Way to go, Tik
1:33:21 Tockers. I didn't even know he had a
1:33:23 [ __ ] quest.
1:33:29 All right, let's see. New style.
1:33:34 [Music]
1:33:47 A cafe door.
1:33:50 A fleeting face,
1:33:53 a matchbook off her time misplaced.
1:33:57 [Applause]
1:33:58 She slipped it deep.
1:34:01 Her coat held tight.
1:34:03 No flame was born, just endless night.
1:34:08 Years roll by like tides on stone
1:34:15 chills the bone
1:34:17 [Music]
1:34:19 memory.
1:34:21 [Music]
1:34:27 >> It's got a little Alicia Keys in it.
1:34:31 [Music]
1:34:36 >> Lauren Hill. Yeah.
1:34:39 >> A cafe door,
1:34:42 a fleeting face,
1:34:45 a matchbook off her. Time misplaced.
1:34:50 She slipped it deep.
1:34:52 Her coat held tight. No flame was born.
1:34:58 Just endless night.
1:35:02 Years roll by like tides on stone.
1:35:07 What sun lit still chills the
1:35:12 match book memory never struck a spark
1:35:18 that faithful got to pluck a stranger's
1:35:24 gift a fleeting glance a fire lost to
1:35:30 circumstance
1:35:32 [Music]
1:35:35 Ooh.
1:35:40 [Music]
1:35:41 >> Wow.
1:35:43 Nice.
1:35:49 >> The sense of rain and the city hums
1:35:56 she's lost in one.
1:36:00 Her pocket holds a quiet plea.
1:36:05 a relic from what could not be.
1:36:11 What if the magic kiss the wick? What if
1:36:14 the clock
1:36:16 tick?
1:36:20 [Music]
1:36:22 A tiny blaze of fleeting glow. Will she
1:36:25 have stayed? We'll never know.
1:36:33 book memory and never struck a spark
1:36:38 that faithful got a plug. A stranger's
1:36:44 gift a
1:36:46 blumstances
1:36:53 [Music]
1:37:04 [Applause]
1:37:06 [Music]
1:37:06 [Applause]
1:37:10 [Music]
1:37:12 [Applause]
1:37:12 [Music]
1:37:16 Holy [ __ ]
1:37:21 [Music]
1:37:24 >> Oh, listen to that chord.
1:37:28 Wow, that chord at the end is gorgeous.
1:37:34 >> Hey, come on.
1:37:40 >> All right, we're gonna we're going to
1:37:42 publish this one. That was really good.
1:37:45 [ __ ] This is crazy.
1:37:50 This is just crazy.
1:37:53 Let's get the video. So, we'll download
1:37:55 the video.
1:37:58 Download the video. Generating video.
1:38:02 We'll download.
1:38:04 Oh, look. Get stems. Have you seen this
1:38:07 yet?
1:38:11 Extract 50 credit. All detected stems.
1:38:16 Yeah, we want all of them.
1:38:28 and then you can download all the stems.
1:38:32 So, what's coming what's coming for
1:38:33 Sunno is a full-on studio. So, you're
1:38:36 going to have multi-track,
1:38:38 multi-segment songs that you're going to
1:38:40 be able to edit. You're going to be able
1:38:43 to go in and just say, "I don't like
1:38:44 that baseline. replace the baseline or
1:38:46 replace the baseline in the chorus or
1:38:49 change these words in this section.
1:38:54 This is listen
1:38:58 there is
1:39:02 for for people who make music the
1:39:05 oldfashioned way with digital audio
1:39:08 workstations which by the way those were
1:39:11 considered you know the devil's work
1:39:13 when they first came out too. What do
1:39:15 you mean you're not doing it on tape to
1:39:17 tape you know analog analog editors? Um,
1:39:24 this shit's got to be [ __ ]
1:39:27 terrifying. But what's going to happen
1:39:30 is people are going to get really good
1:39:32 at producing music this way. And it's
1:39:35 not going to be what's
1:39:38 there's going to be some automatically
1:39:41 generated [ __ ] like algorithmically
1:39:43 generated garbage that's just going to
1:39:45 be pumped out there.
1:39:47 That's going to create sort of a bottom,
1:39:49 a floor of of bad, shitty AI music, but
1:39:53 it's going to get better and better and
1:39:55 better where it'll all be listenable,
1:39:57 but you'll be able to tell. And then
1:39:59 there's going to be some stuff that
1:40:00 rises above the noise. And I I promise
1:40:03 you, what's going to rise above the
1:40:05 noise are people taking their existing
1:40:08 music skills or their existing taste and
1:40:11 learning how to produce music in this
1:40:13 new way.
1:40:15 Um, absolutely crazy.
1:40:18 All right,
1:40:20 let's see.
1:40:24 That's backing.
1:40:26 [Music]
1:40:28 Here's his drums.
1:40:41 Bass.
1:40:42 [Music]
1:41:00 It still
1:41:04 [Music]
1:41:05 memory never struck a spark that
1:41:10 faithful got a plug.
1:41:14 The stranger's gift a fleeting glance a
1:41:19 fire lost to circumstance.
1:41:23 [Music]
1:41:31 >> Crazy. Absolutely crazy. Um, did I get a
1:41:35 video out of this thing? Let's go see.
1:41:40 Downloads.
1:41:42 You don't You can't see this. Yeah.
1:41:46 All right. So, what we're gonna do now
1:41:48 is we're going to head over to X. We're
1:41:50 gonna try to upload a song,
1:41:58 but it's a video, so it's probably not
1:41:59 going to upload because that's the way
1:42:01 [ __ ] works.
1:42:02 [Music]
1:42:05 Oh, share this tab.
1:42:08 All right. Uploading.
1:42:11 It's not going to work. Error. post.
1:42:15 Oh, looks like it's going. Nope. Error.
1:42:24 [Music]
1:42:27 That's nice. You could touch up any part
1:42:30 you want. Yep. Exactly.
1:42:33 Oh, by the way, on So, I'll just keep
1:42:36 trying to upload that video to Oh, wait.
1:42:40 Is it going? Oh, wait.
1:42:46 60% this might go
1:42:50 uploaded 100%. View your post. Okay. Um,
1:42:54 so I'm going to do edit post and I'm
1:42:56 going to do
1:42:58 um
1:43:00 well
1:43:03 well well
1:43:08 at sununo
1:43:11 v5 is insane.
1:43:16 Um I'm going to go get my other post.
1:43:19 Hang on a sec. I'll be back. Am I
1:43:20 sharing this? Yeah, I am. Uh, I'm going
1:43:23 to go to X.
1:43:37 Um,
1:43:39 [Music]
1:43:42 how do I copy this? Oh, wait. Copy link.
1:43:46 Okay. Then I'm going to come back here
1:43:47 where you all are. I'm going to say is
1:43:49 insane. Um, this is the companion song
1:43:56 for
1:44:00 this
1:44:02 uh for the latest entry
1:44:06 of
1:44:09 Twice Upon a Time.
1:44:20 do it. Scooble.
1:44:23 So cool.
1:44:25 All right. So, do me a favor. Go to
1:44:29 x.com
1:44:31 and then find Kyle Shannon. That's my
1:44:34 little Twitter place.
1:44:37 And go find this here uh video and and
1:44:40 like like it and post it.
1:44:43 [Music]
1:44:54 A cafe door,
1:44:56 a fleeting face,
1:44:59 a matchbook offer. Time misplaced.
1:45:04 She slipped it deep.
1:45:07 Her coat held tight. No flame was born,
1:45:12 just endless night. Years
1:45:17 roll by like tides on stone.
1:45:21 What sun lit still chills the moment
1:45:28 memory never struck. A spark that
1:45:33 faithful got to plug. A strange gift. A
1:45:40 fleeting glance of fire lost to
1:45:45 circumstance.
1:45:48 [Music]
1:45:54 [Music]
1:46:03 The scent of rain and the city hum
1:46:10 she's lost in one.
1:46:14 her pocket holds a quiet
1:46:20 relic from
1:46:23 >> what apps. Um,
1:46:25 this was done in Suno although I took I
1:46:28 took a written piece. I created a custom
1:46:31 GPT
1:46:33 that creates a little art piece from two
1:46:35 different images from the same person
1:46:37 that that made a choice in their life
1:46:39 and it split their timeline. And so I
1:46:41 have a photo from each of their
1:46:42 different timelines and it's the same
1:46:45 moment where they just made different
1:46:47 choices. And so I took some of that
1:46:49 writing and put it into Sunno and had it
1:46:51 write the lyrics. And then I just I
1:46:54 literally randomly clicked Let's see
1:46:57 what were the the influences here. Jazz
1:47:00 vocal, hop hop, and powerful piano
1:47:04 um as the styles. And then this was just
1:47:07 one of the songs that it made and it's
1:47:09 really [ __ ] good.
1:47:12 [ __ ] bonkers. [ __ ] bonkers.
1:47:16 >> Sounds like a physics experiment. It is.
1:47:18 Listen,
1:47:21 I
1:47:23 one of the things that the the more I
1:47:25 learn about AI, the more I I use AI,
1:47:31 the more it reminds me of how our brains
1:47:33 work.
1:47:35 The whole [ __ ] about AI can't be
1:47:37 creative. How are we creative?
1:47:40 How we're creative is we jam a bunch of
1:47:42 [ __ ] in our brain. We encode
1:47:47 electrical signals into our brains
1:47:50 and they're sitting in there in these
1:47:51 little fragments of electrical pulses
1:47:55 and some new [ __ ] comes in and we go,
1:47:57 "Huh?" And we combine this idea from
1:47:59 when I was seven with this idea from
1:48:01 when I was 23 with this idea from last
1:48:03 week. And then I just stub my toe. So
1:48:06 that informs it. And we just mash all
1:48:08 that [ __ ] together and out comes an
1:48:10 idea.
1:48:12 And so that's essentially what I did
1:48:15 here. I just mashed a bunch of [ __ ]
1:48:16 together and out came this. Now this
1:48:19 could have been a piece of [ __ ] You
1:48:21 might think it's a piece of [ __ ] I
1:48:23 happen to like it. And in the Rick Rubin
1:48:27 school of how you make [ __ ] you don't
1:48:29 make [ __ ] for the audience. You make
1:48:31 [ __ ] for you.
1:48:34 Right. Would it be better if suno video
1:48:38 if it was a video of a woman doing this?
1:48:42 I don't know what you mean. Like if if
1:48:44 it actually turned this into a video it
1:48:47 Yeah. And I'm sure at some point they'll
1:48:49 do that. Right now they just they just,
1:48:51 you know, spin up an image.
1:48:57 [Music]
1:49:00 had the
1:49:02 clock.
1:49:04 >> Look at those lyrics. What if the match
1:49:06 had kissed the wick? What if the clock
1:49:08 had paused its tick? And and that's
1:49:10 right out of the writing that I gave it.
1:49:13 It said that she she put the matchbook
1:49:16 in her pocket without lighting the match
1:49:19 to kiss the wick. Right. So, it pulled
1:49:21 the lyrics out of there and then it made
1:49:23 that [ __ ] couplet.
1:49:27 >> Gorgeous.
1:49:30 [Music]
1:49:37 Look, those lyrics. God damn, they're
1:49:39 good. A tiny blaze,
1:49:42 a fleeting glow. Would she have stayed?
1:49:44 We'll never know.
1:49:46 >> Those are good. It will never
1:49:55 memory
1:49:57 struck. A smart faithful God.
1:50:04 A strange gift, a fleing blumstances.
1:50:15 [Music]
1:50:26 [Music]
1:50:29 the the world's longest run with no
1:50:31 breath.
1:50:34 [Music]
1:50:41 [Music]
1:50:55 that chord.
1:50:56 [Music]
1:51:05 >> [ __ ]
1:51:07 All right, good people. Enough comedy
1:51:11 jokes. That's enough. Enough. Go play
1:51:13 with [ __ ]
1:51:17 Kelly's out there vibe coding apps.
1:51:22 People are all making their Grammy
1:51:23 award-winning albums. All the musicians
1:51:26 in your life are going to call you AI
1:51:28 devils, heathen slop mongers.
1:51:34 Um, learn this [ __ ] Learn this [ __ ]
1:51:38 I'm telling you, man. Having skills at
1:51:40 these creator tools is going to be
1:51:43 valuable because because there's so much
1:51:46 of a backlash from the creative
1:51:47 community. If you're a create,
1:51:49 especially if you're a creative or if
1:51:52 you've got stories that you've always
1:51:54 wanted to tell but you didn't know how
1:51:55 to tell, your time is now because you're
1:52:00 gonna we're gonna have three years of of
1:52:02 people resisting learning this stuff.
1:52:05 And it might not be that long. It might
1:52:06 be two,
1:52:08 but we're going to have some time where
1:52:10 existing creators are going to res
1:52:12 resist learning this and clients and
1:52:15 producers and the public are going to
1:52:19 demand more and more content that they
1:52:23 like the way they like it. And there are
1:52:25 going to be creators that are born out
1:52:27 of this that never thought they'd be a
1:52:30 creator. I'm telling you. And it's not
1:52:32 just it's not just creating AI slop.
1:52:34 It's going to be there's going to be
1:52:35 real artists that come out of this
1:52:37 because these tools, if you haven't
1:52:39 noticed it, like part of part of what I
1:52:42 started to explore this week before some
1:52:44 new [ __ ] came out was these tools are
1:52:47 getting incredibly sophisticated.
1:52:49 They're getting incredibly complicated.
1:52:51 Chat GPT is not a single tool. It's 25
1:52:55 different tools.
1:52:58 Sunno has a radio station. They now have
1:53:00 this thing called Hooks, which is like
1:53:02 Tik Tok. They've they've now got they're
1:53:04 going to have a studio. You can split
1:53:06 these things into stems. So if you've
1:53:08 got existing music production skills and
1:53:11 something like you can take all of your
1:53:14 fragmented song ideas from the past
1:53:16 [ __ ] decade, turn those into real
1:53:18 songs and start releasing that [ __ ]
1:53:21 Like there's there like the
1:53:22 opportunities right now are [ __ ]
1:53:24 insane. And there's going to be a lot of
1:53:26 people sitting on the sidelines pissed
1:53:28 off that things are changing. But things
1:53:30 are changing,
1:53:32 right?
1:53:34 So anyway, for what that's worth, go
1:53:36 play. Go play with this [ __ ] Go get
1:53:38 skills and start thinking about how you
1:53:41 rise above
1:53:44 the slop, right? The slop is hit the
1:53:46 easy button, right?
1:53:51 >> That's slop. When you hit it quick, just
1:53:54 this, and you put it out in the world.
1:53:57 This song I created tonight, I would
1:53:59 argue is not slop because it's a
1:54:02 derivative of the thing I created last
1:54:06 night, which was an actual idea. I took
1:54:08 whatever it was 90 minutes to work
1:54:11 through how to get that thing to prompt
1:54:13 right and take these two images and
1:54:15 create this thing and create an
1:54:17 interesting post out of it. And then
1:54:20 that led to this. And then this weirdass
1:54:22 experiment of just clicking on styles
1:54:26 led to that which was like that's cool.
1:54:29 There were four other songs we created
1:54:30 or six whatever it was we didn't choose.
1:54:34 I chose this one. Why? I don't know. I
1:54:36 liked it.
1:54:39 That's the trick, right? The trick is
1:54:41 being mindful about what the shit's
1:54:43 producing and then what you're putting
1:54:45 in the world is in slop. But from your
1:54:47 point of view, I could defend this all
1:54:49 [ __ ] day.
1:54:51 So, whatever.
1:54:54 There are probably people on this call
1:54:56 right now cussing at me, but it's okay.
1:54:58 Like, you can.
1:55:03 Um, but shit's changing and it's it's
1:55:05 not going to not change just because
1:55:07 you're pissed off. All right.
1:55:24 All right, everybody. Um, what's today?
1:55:27 Tuesday. Tomorrow night, normal time,
1:55:29 8:00 PM. I will see you here. Peace out.