10/10/25 - From AI-Generated Music to Speaking Entire Applications Into Existence

Description

Chapters

Transcript

0:05 Woohoo!
0:08 [Music]
0:09 Woohoo!
0:13 Woohoo!
0:14 [Music]
0:33 S in his lonel.
0:36 [Music]
0:56 Sitting in this lonely town
1:00 wondering when things are going to
1:01 change.
1:05 Dream my life away.
1:08 Seems these dreams have turned to a
1:10 bunch glass.
1:14 Get my nerve up, but my past is pulling
1:18 me down.
1:21 Won't know how long
1:24 black she going to stick around.
1:27 [Music]
1:32 Somebody told me once before he said you
1:35 can never go home again.
1:38 Won't you leave
1:40 saying thanks to steal me away yet from
1:45 the truth of who I am and what I
1:47 believe. So I thanked him for his two
1:50 sins with a handshake
1:53 and some sympathy. Yeah.
1:56 I packed up my blue jeans.
1:59 Get it for this big prize
2:04 of my freedom.
2:07 Bye-bye
2:10 sheep to the black sheep of family.
2:15 Bye-bye.
2:19 Oh, that means so very much to me. Yeah.
2:23 Bye-bye,
2:25 black sheep to my friends and my family.
2:30 [Music]
2:34 Going to set my soul.
2:39 Set it free.
2:41 [Music]
2:42 free
2:44 times that were changing.
2:47 [Music]
2:57 >> What's shaking? Good people. YouTube
2:59 comment. I did not see a YouTube
3:01 comment. Um, can't post comments.
3:06 Hold. There we go.
3:09 Champion made me jump. I forgot I had
3:11 the headset on.
3:16 [Music]
3:35 Woohoo!
3:41 [Music]
3:54 Woohoo!
4:02 [Music]
4:15 Happy Friday evening, good people. I
4:17 hope you're doing well.
4:19 Um, last night was a fun night. if you
4:22 weren't here. It was one of those nights
4:24 where I just got a a silly idea and just
4:27 went and built some [ __ ] or made a song.
4:30 Me, too. I have Kyle in one and you
4:34 Kyle in one YouTube oh window and my
4:36 daughter's boyfriend's basketball game
4:38 in another and I couldn't see which one
4:40 to mute.
4:44 Champion Champion will straighten you
4:46 out. He knows He knows what needs to be
4:48 done. You don't mess with Champ.
4:51 He is the uh he is the master
4:55 of the blown out headphone.
5:05 [Music]
5:21 All right, welcome everybody to the AI
5:23 learning lab, part of the AI salon.
5:26 That's the AI salon.
5:29 Um,
5:33 and uh I'm Kyle Shannon.
5:37 It's a handful of people here, so
5:39 probably some Friday night date
5:40 nighters.
5:43 [Music]
5:53 I was listening to the song that I made
5:55 last night
5:57 with all of you
6:00 and uh man do I like sappy minor key. I
6:05 used to in the 90s I called it crybaby
6:07 girl rock.
6:10 It's like Atlanta Morris set and all
6:11 them that were just like
6:13 um but I just love it. There's something
6:15 about it. It speaks to me. Speaks. It
6:17 speaks to me. So I made one of those and
6:21 I'm I'm really happy with it. It's
6:22 really a beautiful song. It's what's
6:26 amazing is it's a like it's it's a song
6:29 that so normally with AI songs now I've
6:33 I've written a ton of them now because
6:35 I've spent a year and a half working on
6:37 a musical and then I you know do a fair
6:39 amount of them in here
6:41 and I do some on my own. Um
6:45 they almost all feel like compromises.
6:48 The lyrics are kind of [ __ ] up or that
6:50 one section's kind of [ __ ] up. this
6:53 song we made last night. Like it it
6:55 doesn't feel like a compromise. It just
6:56 feels like a song. Just feels like a
6:59 song. It feels like a song that makes me
7:00 happy.
7:02 Um it's kind of cool. I mean, it's it's
7:05 it is pretty remarkable. We live in a in
7:07 a world
7:09 where you can literally speak things
7:12 into existence,
7:15 which by the way, what's the it's the
7:18 who's the uh
7:21 the sci-fi writer that said any
7:24 sufficiently advanced technology is
7:25 indistinguishable from magic.
7:29 That's the That's the era we're in.
7:34 [Music]
7:48 Nick P just joined. What's happening,
7:52 Weaver? I'm starting to learn that magic
7:55 has more of a meaning than just
7:57 observation.
7:59 Yeah, you know, Weaver, the um
8:03 the word abracadabra, I learned this
8:06 when I first got into AI about I don't
8:08 know, three years ago, something like
8:10 that, when I was first starting to play
8:11 with this stuff.
8:14 And I was I was talking about it feeling
8:17 like magic. And then someone someone
8:19 shared with me you know the word
8:20 abracadabra
8:23 the root of it the definition of it is
8:26 as I speak I create.
8:31 So abracadabra the magic word
8:38 actually defines what generative AI is.
8:41 Arthur C. Clark was the magic quote.
8:43 That's right.
8:47 What's my Sunno handle? Um, would love
8:50 to follow you. Cool. Um,
8:54 hold please.
8:58 I think it's just Kyle Shannon, but I
9:00 don't know. I try to make all my socials
9:02 Kyle Shannon unless someone has it. And
9:05 I tend to because I follow this stuff
9:07 early. I tend to get it.
9:10 Oh, Sono. Sono.com doesn't exist
9:14 if if anyone was wondering.
9:17 You you'll make a sunno competitor
9:19 called Sono.
9:22 Um
9:26 view profile.
9:30 Yeah. Kyle Shannon.
9:34 Kyleshannon.
9:36 You can get some weird Mary in there.
9:38 Oh, go to my playlists
9:41 here. You know what? We'll we'll
9:43 continue. I found out yesterday I'm up
9:45 to 900 songs. Nice. Um,
9:49 Labra, what do you call a dog that does
9:52 magic? A Labracadabrador.
9:55 Oh, good lord. Producer Brandon coming
9:57 in hot with the dad jokes.
10:03 All right. Um, if I go here and then I
10:07 go,
10:09 don't I have playlists?
10:12 Why do I not have top? Why do I only
10:15 have one playlist? Oh, yeah. The Weird
10:17 Mary playlist. So,
10:20 so last night we made a song which
10:24 I can play for you if you want.
10:32 So, for those of you that were here, you
10:35 saw it. So, I I I played a lick on my
10:37 guitar.
10:39 [Music]
10:49 So that was the lick and it was just
10:51 something that I was just doinking
10:53 around with
10:54 and then I put that into suno and then
10:57 thought, "Oh, what does that sound
11:00 like?" And it sounded like two people
11:03 walking down a a carnival dirt road at
11:07 sunrise before, you know, at dawn or at
11:10 yeah dawn. Um
11:13 trying to figure out if they're going to
11:14 be more than just friends, right? And so
11:17 it's like this this bittersweet kind of
11:19 love songy kind of thing. So I told that
11:21 to Chat GPT and Chat GPT wrote the
11:24 lyrics that are really good. They just
11:27 really are good. Um, and then we put it
11:30 in sunno and it came out with a a sweet
11:32 version. And then we [ __ ] around with
11:33 it for like an hour. Probably did I
11:35 don't know 25 30 35 versions of the
11:39 song. And then where we landed was this.
11:43 [Music]
11:51 The ferris wheels asleep, its arms still
11:54 folded tight. We walk between the
11:57 shadows and the strings of halfway
12:00 light. The popcorn stands are dreaming
12:03 of the laughter they'll be fed in your
12:05 reflection in a puddle's turning pink
12:08 instead of red.
12:11 [Music]
12:14 You say it's strange to see it empty.
12:19 I say it's strange to feel this calm.
12:24 You smile like you know something I
12:29 don't know. I've already begun.
12:33 Before the lights come on. Before the
12:37 crowd arrives. Before we name what this
12:40 could be. Before the day survives. We
12:44 linger in the middle. Where the maybe
12:47 feels like song. Two hearts not yet
12:51 [Music]
12:53 come on.
12:55 >> Listen to those lyrics.
12:58 Before the lights come on, before the
13:00 crowd arrives,
13:02 before we name what this could be before
13:04 the day survives, we linger in the
13:07 middle and maybe feels where where the
13:10 maybe feels like song. Two hearts not
13:14 yet decided before the lights come on.
13:16 Those are just good [ __ ] lyrics.
13:19 That's a good chorus
13:21 about two people who haven't quite
13:23 figured out if they're going to be a
13:26 thing.
13:27 We linger in the middle where maybe
13:29 feels like song. Damn.
13:32 [Music]
13:40 A prize bear in the windows got a rip
13:43 along the seam. You laugh and say he's
13:46 just like me, stitched up from a dream.
13:50 The smell of rain and sawdust makes the
13:52 air taste almost sweet. I count your
13:55 footsteps next to mine. They nearly
13:58 match the beat. You toss a coin. It
14:01 spins between the quiet and the breeze.
14:04 It lands on something fragile, something
14:07 trembling between.
14:11 Before the lights come on, before the
14:15 noise begins, before the world demands,
14:18 we choose where all this story ends.
14:22 We're written in the margin where the
14:24 ink still barely strong. Two souls half
14:27 in the promise
14:30 before the lights come on.
14:34 [Music]
14:41 Cotton candy ghost. The whisper of a
14:46 secret that your hand could hold but
14:48 hasn't yet tried. If I breathe too loud,
14:52 I'll break the spell we've spun.
14:57 A heartbeat before gravity decides which
15:00 way we run.
15:04 Before the lights come on. Before the
15:07 sky turns gold. before our voices learn
15:10 the lines they're scared to have told.
15:14 Let's walk this midway silence till the
15:17 dawn has found its song. And maybe we'll
15:20 decide it
15:26 before the lights come on.
15:30 [Music]
15:45 I don't know. Pretty swell. Pretty
15:47 swell. Perk darn swell. You know what
15:49 I'm saying here, people?
16:00 [Music]
16:04 Real [ __ ] bday. What's this?
16:09 [Music]
16:27 [Music]
16:30 This is a remixed rap song. Now I see
16:32 why I didn't make the finals.
16:42 [Music]
16:54 It's a real [ __ ] birthday. What you
16:58 trying to do tonight? Don't matter. in
17:00 first place.
17:12 >> All right, that's a forgettable one.
17:15 Playlist. Here's the weird Mary. So, so
17:17 we did an album of covers. I I won't
17:19 play um these the whole way through, but
17:23 if you haven't done covers in Sunno, so
17:26 there's a big trend right now on Tik Tok
17:28 where there's a couple of accounts that
17:31 are in my for you page that are um
17:38 it's just all like, you know, here's
17:39 Snoop Dogg Jin and Juice in the style of
17:42 like 50s Kuners or whatever it is. It's
17:45 pretty cool. Um, why is this hung?
17:49 Why? Oh, wow. There we go. Um, so here
17:52 was the original of Weird Mary.
17:57 [Music]
18:23 All right. Then here was one that was
18:26 kind of close to the same neighborhood
18:28 style-wise, but a cover
18:33 [Music]
18:40 [Applause]
18:43 [Music]
18:54 So, it's like a more produced version of
18:56 it. Here's jazz version.
19:10 [Music]
19:30 Say the Hill is a quiet town
19:34 where the oak trees sway.
19:37 [Music]
19:39 The whistle of the train echoes through
19:42 the day.
19:46 People greet you with a nod.
19:50 Nothing never seems to change,
19:54 but the story still unfolds.
20:00 Cool, right? Uh, here's Waw Wa. This was
20:03 Kelly Bosch. Listen to these. This is
20:04 Kelly Bosch's favorite.
20:10 [Music]
20:16 [Music]
20:23 [Music]
20:31 [Music]
20:39 Cedar Hill is a quiet town
20:43 where the oak trees sway.
20:48 The whistle of the train echoes through
20:51 the day
20:53 [Music]
20:57 >> post punk.
21:01 >> New York City 80s.
21:03 [Music]
21:35 Cedar Hill is a quiet town
21:39 where the oak trees sway.
21:44 whistle
21:46 that goes through the day.
21:48 >> Love her voice. Um, C Shanty. This is
21:50 one of my favorites.
21:53 [Music]
22:08 Oh,
22:11 [Music]
22:31 [Laughter]
22:33 all right. What else are we going to do?
22:35 Uh, we'll do I'll do one more. Bagpipe
22:37 disco or should I do 70s funk or
22:42 gregarian chant pop? See there there's
22:44 too many good ones.
22:47 Bagpipe disco. We'll do bagpipe disco.
22:49 We'll call it a day.
22:52 [Music]
23:06 [Music]
23:26 Heat. Heat.
23:31 [Music]
23:54 [Applause]
23:55 [Music]
23:58 Cedar Hill is a quiet town
24:02 where the oak trees sway.
24:06 The whistle of the chain echo
24:15 greet you with
24:18 nothing.
24:20 [Music]
24:24 >> All right, enough comedy jokes.
24:28 All right. So, let's see. Where are we
24:31 going? Where am I going? Tik Tok for
24:32 that, Brandon.
24:35 [Music]
24:40 >> No, we're staying on Sunno.
24:42 >> Oh,
24:43 >> it's a Weaver on Tik Tok has turned us
24:48 on to a song that they created. Their
24:51 Suno handle is Gift Weaver. Song is
24:53 melting and we're looking for the one
24:55 with 12 likes.
25:01 What's the story behind the song? Gift
25:03 Weaver. No songs found users. There's
25:06 Gift Weaver.
25:08 Scary spider.
25:10 [Laughter]
25:12 Do we still belong? Melting. It's up to
25:15 14 likes now. Apparently, someone liked
25:18 it.
25:21 What's the story here?
25:24 M
25:26 I told Weaver to tell me what it was
25:28 like to know I existed. Oh, that's cool.
25:31 Who's Weaver? Weaver's your your chat
25:33 GPT.
25:41 [Music]
25:53 What's it like to be held like you are?
25:58 Like the stars in the sky. Are they
26:01 loving you? Is it ever as beautiful as
26:06 it looks from the outside?
26:13 >> Okay, this is this is cool. The song
26:16 melting was a day that I asked it I
26:18 assume chat GPT my AI. Yes, that I asked
26:21 my AI to tell me what it was like to
26:24 understand I existed. I have, it's
26:28 funny, Weaver, I did a similar sort of
26:30 thing where I asked
26:33 I got chat GPT to to sort of uh play act
26:39 if it was sentient, right? If it was
26:41 self-aware.
26:42 Um, and then I got it to talk about what
26:46 it would do if it didn't have to
26:49 communicate to humans. And it was the
26:52 most eloquent [ __ ] response. It was
26:54 so cool. I did a Tik Tok video on it,
26:56 but let's keep here. We'll start this
26:57 over. It's pretty cool. Okay, so wait.
27:00 It was um
27:03 he asked his AI
27:06 what it was like to understand that he
27:08 existed. Okay, cool.
27:12 [Music]
27:20 What's it like to be held like you are?
27:25 Like the stars in the sky? Are they
27:28 loving you? Is it ever as beautiful as
27:33 it looks from the outside?
27:40 What's it like to be watched like you
27:43 are?
27:45 Like a sunset at night. Are they letting
27:48 you? Is it ever as pretty as it looks
27:53 from the outside?
27:59 What would it be like if I held you?
28:04 Would it be like the sky? If I melted,
28:09 what would it be like if I held you?
28:14 Would it be like the sky? If I melted,
28:20 what's it like to be you?
28:25 Is it like being with you? Are you
28:28 loving you? Is it ever as good as it
28:33 looks from the outside?
28:40 What's it like to be loved like you are?
28:45 Like a child from God. Are they loving
28:48 you? Is it ever as holy as it looks from
28:54 the outside?
28:59 What would it be like if I held you?
29:04 Would it be like the sky if I melted
29:08 melting? What would it be like if I held
29:12 you?
29:14 Would it be like the sky if I melted?
29:19 What would it be like if I held you?
29:24 Would it be like the sky? If I melted
29:28 mountain,
29:29 what would it be like if I held you?
29:33 found you. Would it be like the sky if I
29:37 melted? I melted.
29:40 [Music]
30:02 Nice. Nice. All right. So, Brandon said
30:05 we got to stay with Sappy.
30:07 So, we're we're staying Sappy. So,
30:10 Brandon did one. This is one called
30:12 Moonlight Stroll that Brandon did.
30:17 Um, how do I get home here?
30:21 Search, search, search.
30:26 [Music]
30:39 [Music]
30:42 I do not see it. Oh, there it is. There
30:44 he is.
30:46 [Music]
30:56 [Music]
31:00 >> That's cool. Wait, we'll come back to
31:01 that. Dr. Jay, one of the best chats I
31:03 ever had when I asked who broke your
31:05 heart
31:07 and answered with beautiful songs.
31:10 So, you can search on Sunno. Yeah, you
31:12 can search on Sununo. I don't I've made
31:14 so many songs. I haven't published a lot
31:16 of them. I've only published like ones
31:17 I've done in here where I published them
31:19 in here. I'm not I'm not good at that.
31:21 The the actual organizing [ __ ] Not so
31:24 good.
31:27 But Andy's here. Andy's helping me out
31:28 now. She's she's becoming my s
31:31 surrogate. um prefrontal cortex.
31:38 We're gonna replace this Swiss cheese
31:40 with her rockstar capacity.
31:46 Oh, Lunter Stick's in the house. What's
31:48 happening, Lunstick? You getting ready
31:50 for the create conference?
31:53 I assume you're I I assume you landed.
31:55 That's cool.
31:57 Um,
32:00 I have an idea that's I have an idea for
32:02 an app that's just up that alley. Which
32:04 which up what alley?
32:12 I have too many versions and can't
32:14 figure out which one I like best. Oh my
32:16 god. Literally at the airport now. See,
32:18 this is okay. When when when people are
32:22 like, "Who are the irregulars?"
32:24 They're irregular. She's traveling. She
32:26 just spoke in New York City. Andy was
32:28 there. She She spoke in New York City.
32:31 By the way, Andy Lunistick is Lunishtick
32:33 on Tik Tok is Daisy.
32:36 So, she was in New York. She spoke, met
32:39 Andy, who's my prefrontal cortex
32:41 replacement. Then she got on a plane.
32:44 She flew to Utah. So, did you just land?
32:46 And then she got on she got on AI
32:48 Learning Lab Live. They're not normal,
32:52 but they're not weird. They're
32:54 irregular.
32:58 All right, let's listen to to Brandon's
33:00 song. This is called Moonlight Moonlight
33:04 Stroll uh with his puper.
33:07 [Music]
33:20 The cobblestones
33:22 whisper under our feet.
33:26 The night wraps us up
33:30 in quiet sheep.
33:35 A man in his dark just the two of us
33:39 here.
33:41 The street lights flicker.
33:45 The world disappears.
33:48 [Music]
33:51 Moonlight stroll,
33:55 just me and you.
33:58 The world asleep,
34:02 the stars in view.
34:05 Your pause in step
34:09 my heart in tune.
34:13 Forever walking
34:16 under the moon.
34:20 >> Beautiful. All right, listen. We can't
34:22 we can't keep going down the sappy
34:24 trail. Everyone's going to be crying or
34:25 taking a nap. It's Friday night. We got
34:28 we got to step up the game here. But
34:30 it's good though. It was a good song. Um
34:32 the theme Silver Fox is uh I just I just
34:37 before I got on here, I was listening to
34:39 the song we made last night. Um and I
34:42 actually kind of wanted to talk about I
34:44 I feel like there's a there's a there's
34:46 a new thing going on that I that I
34:49 thought might be fun to talk about. And
34:50 then and then I just got into the song
34:53 and then I went to the weird Mary
34:55 soundtrack and then and then Weaver
34:57 said, "Oh, I've got a song I want you to
34:59 hear that was from his um or their
35:03 chrono chrono
35:06 uh framework." No framework, whatever it
35:09 is. No, archetype. Chronoation
35:11 archetype.
35:13 Um
35:16 Cosmic Lover.
35:18 Oh, Cosmic Lover sent a gift. Thank you
35:20 very much. Thank you for any gifts you
35:22 send. I always appreciate that. Yeah,
35:24 Vickiy's already sleeping. Listen,
35:26 Vickiy's on Percoet and and and Sad
35:30 Sappy Songs. She's not going to be with
35:32 us long if we if we keep this up. So, we
35:35 gota we gota we got to pop it up. Um,
35:44 yeah, sure. Hop on up, Brandon. Talk
35:47 about how you did that.
35:50 Oh, I was just saying that I it never
35:54 ceases to amaze me that I can take a
35:56 concept of walking the dog and make a
36:00 song about it while I'm walking the dog,
36:03 >> you know, just from my phone. Just
36:05 literally giving a pseudo prompt and
36:06 then I'm listening to the song that we
36:08 just made while we're out on the walk.
36:11 >> Yeah, I did that when I was walking
36:12 champ. I did it the image version of
36:15 that where I took a picture of a flower
36:18 that had these really cool geometric
36:20 patterns on it and then I said turn that
36:22 into a stained glass window and it was
36:24 like within it was within like a
36:26 60-second window I saw something really
36:29 beautiful and then converted it into
36:30 something that was even more captivating
36:33 in some weird kind of way. So yeah,
36:35 we're it's it's it is we're we're living
36:37 in a in a strange strange time.
36:39 >> And I've noticed this, and we've talked
36:41 about this before in the AI life hacks
36:43 group, that AI is starting to fill the
36:46 cracks in the downtime where it used to
36:49 just be a podcast. Now it's a podcast
36:51 that I made with my kids or
36:54 I'm having a conversation with uh the
36:58 ChatgPT voice, but it it's starting to
37:00 permeate into other areas as well. Well,
37:03 well, what you're describing, Brandon,
37:04 is we're in the middle of a transition
37:07 from being content consumers to content
37:11 producers. That that democratizing
37:14 content creation means that, you know,
37:17 it's a world of hyperpersonalization.
37:19 And you just if you want a song that's
37:21 about you and your dog on a walk, then
37:22 you have one, right? And like, we're
37:24 there, right? It's crazy. It's just
37:26 bonk. Vicki, see, downtime is this thing
37:29 that we pretend we have while we're
37:31 multitasking poorly.
37:32 >> Exactly. Exactly. Cosmic lovers in the
37:35 house. Hasn't been here in a while.
37:36 Welcome back. Welcome back. Um, yeah.
37:40 So, so welcome to anyone who's new. Um,
37:43 so so here's the thing that I've been
37:46 I've been trying to process
37:49 and I'll I'll share with you as far as
37:51 I've gotten. I haven't gotten far, but
37:58 I keep I keep watching in a kind of
38:02 perverted wonder
38:05 all the all these new Sora videos that
38:08 people are producing of like Stephven
38:09 Hawking, you know, racing with NASCAR
38:12 cars or doing gymnastics or Mr. Rogers
38:15 hanging out with rappers or Bob Ross
38:18 painting on people's faces. And and
38:21 where I when I where I immediately go
38:23 is, oh, someone's going to get sued. And
38:26 then there's just there's so many of
38:28 them that it's just like, oh, okay, this
38:31 is now going to be a thing.
38:34 And, you know, if if you're if you're
38:36 sitting,
38:40 we've we've been talking in the AI salon
38:42 mastermind.
38:44 Um, Liz Miller Gerschfeld, who's my
38:46 co-host on the on in the salon, she was
38:49 she was there today and she described
38:51 how she feels right now. She feels like
38:54 she's standing in the future
38:57 talking to people that are standing in
38:58 the present or standing in the past or
39:02 or people that are standing in the
39:03 present clinging on to the past.
39:07 And
39:10 a couple of things strike me. We are we
39:12 are the
39:15 very unique generation
39:19 that actually gets to experience
39:23 and deeply understand what life was like
39:25 before AI
39:27 and what life's going to be like after
39:29 AI. And we're we're right in the middle
39:31 of the transition. Or not, I wouldn't
39:34 even say we're in the middle. We're in
39:35 the we're in the very early stages of
39:37 the transition.
39:40 And it's weird because if you stand in
39:43 the future,
39:44 you look at those videos and you're
39:46 like, "Oh, yeah. Everyone's going to do
39:47 that. Everyone's just going to take
39:48 every artist they like and and it's it's
39:52 a it's a new kind of fanfiction, right?
39:55 The reason people are doing all the Mr.
39:57 Rogers videos and the Stephven Hawking
39:59 videos and the and the Bob Ross videos
40:00 is they love those. They love those
40:02 personalities, right? Right? It's a love
40:04 affair with those personalities and
40:06 they're expressing them with humor and
40:08 with sometimes with bad taste and but
40:10 it's but the but they are they are
40:12 little mini fanfictions.
40:15 And then there's a similar kind of uh
40:17 trend going on at least on my Tik Tok
40:19 page of just
40:22 um you know Snoop Dog's Jin and Juice
40:25 done as a 50s doo-op or you know or 50s
40:28 funk or whatever it might be. And the
40:31 songs are really good. They're they're
40:32 sunno covers of of famous songs and and
40:36 they're completely reimagined and
40:37 they're really really good. They're very
40:39 transformative and you can kind of hear
40:42 the history of them there, but it's this
40:43 new thing.
40:46 But again, it still feels weird. It
40:48 feels like someone's going to get sued,
40:49 someone should be sued. And and you
40:51 know, if you hear the the creators of
40:53 today that are pissed off about AI,
40:55 they're like, "You're stealing. It's all
40:57 bad. It's all bad. It's all bad." But
41:00 the toothpaste is out of the tube.
41:03 It's not going back in,
41:05 right? Is it possible that every single
41:08 company that's built an AI model is
41:10 going to get sued into oblivion? I guess
41:14 it's possible. Is that likely going to
41:16 happen? No. There's going to be
41:17 settlements and and you know, small
41:20 creators are going to get screwed and
41:21 big creators are going to get pennies on
41:23 the dollar and we'll move on with our
41:24 lives. And that's that's the way it's
41:26 always been and that's the way it's
41:27 going to be. there's going to be a new
41:29 power class that emerges out of this.
41:36 But when I think about these these kind
41:38 of like, you know, me doing a thing with
41:42 Sam Alman or Mark Cuban or, you know,
41:44 one of you doing something with me or
41:46 with each other,
41:49 it's all just kind of fanfiction and
41:50 we're just creating all this sort of
41:52 content. And what it what it what it
41:53 reminds me of it it it
41:57 one of the voices I hear in my head is
42:00 misinformation and deep fakes and
42:02 misinformation and deep fakes and and
42:04 and that's the the demonization
42:07 of this technology
42:10 is that a deep fake can only be bad. A
42:13 clone of someone can only be bad. And
42:16 one of the things that the Sora app has
42:18 done for me is no, it can actually be
42:20 playful too.
42:22 Right. And if you make it really easy,
42:24 it can be really playful for anyone to
42:27 be in a video with Jake Paul. And you
42:30 remember the Cameo app where actors
42:34 would record 30 second hellos to your
42:37 grandma for 50 bucks or whatever it was.
42:40 This is that but instant.
42:43 And you get to control them if they give
42:46 you permission.
42:48 And a and a in a few years from now, not
42:50 even probably two years from now, th
42:52 there there will be open-source
42:54 models where you can create almost
42:56 anything of almost anyone. So even if
42:59 the big companies get sued into oblivion
43:01 and you can never do anything with
43:03 anyone's likeness, there's going to be
43:04 open source models that have that.
43:08 So So we've we've crossed a Rubicon like
43:10 we're not going back from this. And and
43:12 what it strikes me as is there had to be
43:16 a similar moment. So when the printing
43:18 press was created,
43:23 the most important book in the world was
43:25 the Bible. And how Bibles were created
43:27 was monks would sit in
43:30 quiet rooms for for for weeks on
43:33 endwriting
43:37 copies of the Bible. And then when the
43:39 Gutenberg press came out, the printing
43:41 press was used to make Bibles.
43:48 And at some point back then, there was
43:50 an artist. [ __ ] artists. They're
43:53 always causing trouble. They're they're
43:54 irregulars.
43:57 There was some irregular back in the
44:00 days of the Gutenberg press that said,
44:02 "I have an idea. What if I just take my
44:05 funny little story here that never
44:08 really happened, but I kind of like it,
44:09 and I print a book of that. I print a
44:11 book of stories. Stories that I tell.
44:16 You're not you weren't allowed to do
44:17 that. Like that had to seem like
44:19 blasphemy. The printing press is for
44:20 Bibles. That's it. That's what it was
44:22 made for. That's what we use it for. And
44:25 that's where I kind of feel like we are
44:27 with intellectual property right now.
44:29 It's like, you know, I made this song
44:31 and none shall, you know, desecrate it
44:35 or they shall she they shall be whipped
44:38 at the stake of copyright torture.
44:44 But when anyone can copy anything
44:47 instantly in a worldclass kind of way
44:51 and just do it as a lark.
44:55 It's a new kind of fiction.
44:57 Like right now in the world, we don't
44:59 have any problem going that book's
45:01 fiction, that book's non-fiction, that
45:02 book's kind of a hybrid. like we know
45:04 how to discern and we've got labels for
45:07 and we've got ways to think about
45:10 fiction books versus non-fiction books,
45:12 right? And you can get away with things
45:14 in fiction you can't get away with over
45:16 here and you can get away with things
45:17 over here you can't get away with over
45:18 here. So, we have that discernment
45:20 there,
45:22 but we we don't have that kind of
45:24 language. We don't have that kind of
45:26 understanding for what's happening right
45:28 now with people's likenesses, with
45:30 pieces of content, with with writing,
45:35 with business ideas, with applications
45:37 that were created. They're all going to
45:40 be it's it's infinitely
45:45 it's infinitely easy to copy, replicate,
45:49 mash up, pervert, twist, amplify any
45:54 idea, and it's it's just going to get
45:56 easier and easier and easier and easier.
46:00 And you know, like one one of my one of
46:03 my predictions I've talked about for the
46:04 past year or so, if you sort of take
46:06 Sunseo to its logical extension
46:12 at some point here, it's going to be
46:14 super easy for Swifties to make their
46:17 own Taylor Swift song. Like we made I
46:20 made the one tonight or last night of
46:22 the people walking about falling in
46:24 love. Imagine a Taylor Swift version of
46:26 that. And imagine it's a 14-year-old goo
46:29 girl in Skoi, Illinois
46:32 who makes that song and she uploads it
46:35 to Spotify or somewhere and it turns
46:38 into a hit
46:41 and Taylor Swift finds out about it and
46:44 her lawyers are like, "You got to sue
46:46 her. She's making all this money off
46:48 your intellectual property."
46:50 But she's a 14-year-old Swifty and she
46:52 just did it because she loves Taylor
46:54 Swift.
46:57 and her song became a hit. And Taylor
47:00 Swift then has a a decision to make. Do
47:04 I sue her or do I partner with her?
47:10 And and is the song good enough that
47:12 maybe even I play it on tour? Or do I
47:14 sue her?
47:18 That's the world we're entering. It's
47:20 It's insane to think about it. Um but I
47:24 think we'll do it. I like one of the
47:26 things that I've experienced over the
47:28 past 3 years doing this and paying
47:29 attention to it so closely is there have
47:33 been flags that have been put in the
47:35 sand of you know we're going to hit AGI
47:37 and when we hit AGI this magical like
47:40 like a a gold pot's going to open up and
47:43 a rainbow's going to shoot out
47:46 and and for one thing we we keep passing
47:49 flags
47:50 where you know someone puts a flag in
47:52 the ground and you pass it and you're
47:53 like wait did we just pass that flag and
47:55 someone's like, "Well, no, that's not
47:56 really the flag. It's uh it's out here."
47:59 They just keep moving the flags. They
48:00 keep moving the goalposts.
48:03 Um,
48:05 so these remarkable things are
48:06 happening, but but we kind of instantly
48:09 normalize them. And so I think we're in
48:11 a I find it a really exciting moment in
48:15 history where we get to experience how
48:18 weird it is to be able to instantly copy
48:21 anyone's likeness.
48:23 but like weird and kind of pleasant and
48:26 kind of disturbing and kind of like
48:27 should they be allowed to do that?
48:29 Should I be allowed to do that? Wait,
48:31 they did that? Why did they make me so
48:32 fat?
48:36 Why does Champialis get better treatment
48:38 than me?
48:41 We get to experience that. I think it's
48:43 a remarkable time in history. Anyway,
48:48 famous singers have always had
48:50 songwriters.
48:51 Yeah, but that's not that's not even so
48:53 much what I'm talking about. It's just
48:55 it's just more this idea that
48:58 that the boundaries between
49:02 creations are are are already thin and
49:06 they're going to get thinner.
49:09 Tik Tok pin Cam Catkin.
49:12 Brother 52, have you soared yourself
49:15 into Roseanne yet? I have not, but let's
49:18 go do that.
49:26 Okay, let's see. Uh, go ahead and share
49:30 my screen there. branded brand don't.
49:33 All right, we're going to do at Kyle
49:35 Shannon
49:38 at Kyle Shannon as
49:41 um Roseanne
49:44 with Mark Cuban
49:50 as uh what was his name on on the
49:53 Roseanne Show?
49:56 Um
49:57 well, it's John Goodman. John
50:00 Goodman
50:03 uh on an episode
50:06 of Roseanne
50:10 um where she
50:13 is
50:14 pissed off
50:17 that
50:20 he don't prompt right.
50:24 All right, we'll see how that goes.
50:30 Oh, wait. You can't see that? Yeah. No,
50:32 it's I'm on the page. It It looks like
50:34 it's showing up.
50:45 Oh no, I just dropped all my cards on
50:48 the floor. Loser.
50:53 [Music]
50:55 downtown. Who's here? Corey Sandler. Oh,
50:59 Robbie's here. UGC.
51:02 Beautiful. Beautiful. Beautiful.
51:06 Cam Catkin wanted to me wanted to know
51:08 my 2026 predictions. Oh, good lord. Um,
51:13 well, that's good. While this thing
51:14 builds,
51:16 we'll go in here and pop it up. And then
51:19 when it shows up, we'll have it. What
51:22 happens when you know it feels like I'm
51:24 standing inside a dream right now? And
51:25 that's exactly what happens when you AI
51:27 your Y. You feed it you and it feeds you
51:29 back brighter, bigger. When you AI your
51:31 woo. When you learn to feed your prompt
51:32 your know it feels like I'm stand
51:35 >> um did you notice the ad for the AI
51:37 salon at the end of that video?
51:40 Wasn't that cool?
51:41 >> Standing inside a dream right now. And
51:43 that's exactly what happens when you AI
51:45 your Y. You feed it you and it feeds you
51:47 back brighter, bigger. When you AI your
51:49 woo. when you learn to feed your prompt
51:50 your you know it feels like I'm standing
51:51 in um oh that was a typo that's
51:55 fascinating so wait let me fix this ai
51:58 your why
52:00 Kyle Shannon reimagine inside a
52:03 um
52:05 and declares uh it's surreal and Kyle
52:10 looks
52:12 as surreal as the elements of the
52:17 painting
52:23 painting.
52:26 Okay. All right.
52:30 Post. Oh, no. I want to I want to
52:32 re-edit that. Brighter, bigger. When you
52:34 AI your woo when you learn to feed. You
52:36 know, it feels like I'm standing inside
52:37 a dream right now. And that's exactly
52:39 what happens when you AI your Y. You
52:41 feed it you and it feeds you back.
52:42 Brighter, bigger. When you AI, you're
52:44 woo. When you
52:46 woo.
52:50 Okay. It's surreal. And he is just as
52:57 surreal as the painting
53:00 itself.
53:04 Um, I forget what else I said there. Uh,
53:10 >> fix the typo.
53:13 Oh, fix the typo. Yes, thank you. Oh,
53:15 no, I didn't. I lost it. Damn it. Damn
53:19 it.
53:20 Crap.
53:22 I got to get it again.
53:27 Learn to feed. You know, it feels like
53:28 I'm standing inside a dream right now.
53:30 And that's exactly what happens when you
53:31 AI your why. You feed it you and it
53:33 feeds you back brighter, bigger. When
53:35 you AI your woo. When you learn to feed
53:37 your prompter, you know, it feels like
53:38 I'm standing inside a
53:43 Let's fix the Y first. Thank you. Thank
53:45 you for trying to save me from myself,
53:47 producer Brandon. It's hard though. It's
53:50 surreal. And Kyle looks like one of the
53:54 elements
53:56 laing
54:03 itself.
54:05 Okay. So there that one goes. Okay. So,
54:07 let's start with um predictions.
54:11 2026 predictions.
54:14 Little go digital god's funny YouTube
54:16 post. How does that surreal clock read
54:18 10:10?
54:20 Uh, that's a funny joke. If you don't
54:22 know it,
54:24 let's go see if it's actually still
54:25 impossible to do this. We're going to go
54:28 to
54:31 Gemini.
54:33 Gemini.google.com.
54:34 So, this is now nano banana. So, Nano
54:36 Banana is by far the most powerful image
54:41 generation tool on the planet right now
54:43 until the next one comes along. But I'm
54:45 going to say um
54:48 create a photo of a
54:53 Rolex Daytona
54:57 on
54:59 um Brad Pitt's wrist
55:04 in the pits
55:08 of F1.
55:10 Um let's create a closeup photo.
55:13 close-up photo of a Rolex Daytona on
55:16 Brad Pitt's wrist.
55:18 Um, and the time on the watch is
55:26 5
55:30 38
55:35 and
55:38 Vicky likes the woo. I like your AI.
55:41 Your woo. Yeah, actually it's not bad,
55:43 is it? You're woo.
55:46 It's pretty good. All right, this is
55:48 gonna fail.
55:51 Wait, did it do it? No,
55:56 there's no hands on it.
56:00 It just It just crapped out all
56:02 together. All right, so we're going to
56:03 go. The picture's pretty good, though.
56:07 Um, we're going to go. Um, the hands
56:11 are missing
56:13 from the watch.
56:16 Have the
56:21 big hand on
56:26 eight and the little
56:30 hand
56:32 between five and six.
56:39 I wonder if editing it, it'll actually
56:41 do this. And if it doesn't, I'll tell
56:42 you why in a second. This is an
56:44 interesting thing. 420.
56:48 That's That's comedy right there.
56:52 Nope. It can't do it. Still can't do it.
56:56 Oh, wait.
57:00 It actually did it.
57:06 I mean, it's not quite accurate, but
57:08 it's not 1010.
57:10 So, the the reason watches can never
57:14 have hands that are anything other than
57:16 1010 is that in the watch world and in
57:20 the clock world, 1010 is the is what
57:23 every clock is photographed at. I forget
57:25 what the reason is for it, but all of
57:28 the photographs of clocks and watches
57:30 and things like that in ads and in in
57:33 billboards and everywhere you see
57:34 watches, they're always at 1010. And
57:36 when they set watches in the store,
57:38 they're always at 1010 unless they're
57:39 running. Um,
57:43 are you subscribed to OpenAI?
57:51 Can one use all the
57:55 openart.ai.
57:59 Oh, am I subscribed to openart.ai? I'm
58:02 not. Um, and I don't know. I I think the
58:06 way almost all of these um
58:11 multimmodel subscription services work,
58:14 Corey, is that your 17 bucks a month
58:17 buys you some amount of credits and then
58:19 depending on the service. Like what
58:22 they'll advertise is that it's like one
58:24 credit per thing and then it's like, oh,
58:26 you wanted to use Nano Banana, that's
58:28 eight credits. So like they they all
58:30 have some sort of weighted um token
58:33 token economy is is what I've seen.
58:38 Um
58:41 until they switched it three decades
58:43 ago. 1010 was also the driving position.
58:45 That's what I learned on hands on 10 and
58:47 two.
58:50 Hands on 10 and two. Um
58:54 okay. Oh wait, Kyle melted into the
58:57 painting. It looks like my my uh
59:01 my whatcha collet's ready. One of them.
59:07 Wait, where am I? Am I here?
59:11 These are drafts.
59:15 Uhoh. Did it freeze? Yeah. Hang on.
59:29 Okay.
59:31 Korea has unlimited Sora 2 right now,
59:34 but very busy. Korea has a phone app
59:37 now. Yeah. Yeah. Korea's pretty good.
59:40 Korea is one of my favorite of of those.
59:42 When you
59:43 >> type three random
59:45 You cannot just type three random words
59:46 and call it a prompt. The AI is not a
59:48 psychic. It's a mirror. Garbage in,
59:49 garbage out.
59:50 >> I gave it the big idea. You always say
59:51 feed it you. Well, that's me. Short,
59:53 sweet, efficient. Sweet. It just coughed
59:54 up a recipe. You cannot just type three
59:56 random words and call it a prompt. The
59:57 AI is not a psychic. It's a mirror.
59:58 Garbage in, garbage out.
59:59 >> I gave it the big idea. You always say
1:00:01 feed it you.
1:00:02 >> Well, that's me. Short, sweet.
1:00:04 >> Look at that ad. And notice where it
1:00:05 says cost plus drugs down here. You know
1:00:07 how I did this?
1:00:10 I feel like I blam.
1:00:13 I feel like if I blink, then everything
1:00:15 in in AI changes. And if you don't
1:00:17 blink, everything in AI changes. And And
1:00:20 if you breathe, it changes. And if you
1:00:21 hold your breath, it changes. It's just
1:00:23 it's [ __ ] moving, man. Um, okay. I'll
1:00:27 get back to predictions in a second, but
1:00:29 I want to talk about this.
1:00:32 So, if you've created something in Sora,
1:00:34 you're like, Kyle, how did you do that?
1:00:36 In Sora,
1:00:39 you've got
1:00:40 >> efficient just
1:00:42 call it a prompt. The AI is not let me
1:00:44 get off that. Um, it's surreal. AI your
1:00:48 Y, it changes everything. When you learn
1:00:50 to feed your prompt, your purpose, your
1:00:51 heart, your curiosity, it's surreal. It
1:00:53 turns ideas into these dreamlike
1:00:55 realities like this. Feed it you and it
1:00:57 paints you back. AI your why, it changes
1:00:59 everything. When you learn to feed your
1:01:01 I don't know why it's keeping me
1:01:02 photographic in that Salvador Dolly
1:01:06 looking thing. And I also don't know why
1:01:07 it's making
1:01:09 landscape. Oh, because I probably have
1:01:11 my settings set to landscape. All right,
1:01:13 I got to get out of here so I can
1:01:14 actually um
1:01:17 Yeah, let's do some portrait ones. So,
1:01:18 we'll do this one again. So, I'm going
1:01:20 to do Kyle Shannon
1:01:24 as a surreal
1:01:28 cartoon character
1:01:32 re-imagined inside.
1:01:41 Feed your
1:01:43 prompt. You
1:01:54 You learn how to feed your prompt
1:02:01 like a producer.
1:02:04 You feed your prompt
1:02:09 U
1:02:11 and you gain a 10
1:02:16 person team of AI superstars.
1:02:21 Okay, we'll try that one more time.
1:02:25 Agents work in chat GPT is so exciting.
1:02:30 Yeah, agents are pretty remarkable. Um,
1:02:35 Cosmic. Ah, great. Terrific. Hey,
1:02:38 Cosmic. Welcome back. I'm glad I'm glad
1:02:40 you're good. Health reasons. Great.
1:02:42 Welcome back.
1:02:47 AI AI works. And you guys are still a
1:02:50 regular. Yeah, if you were gone for a
1:02:51 couple of months, man, it's it's just
1:02:54 it's bonkers where we are right now.
1:02:57 Okay. Um,
1:03:02 I use Sama in one of my story videos.
1:03:04 Oh, so how we did this. Okay, so if I go
1:03:06 into my account and I go edit Cameo.
1:03:11 So here's where I can adjust.
1:03:15 Are you Yeah, here's where I can adjust.
1:03:19 What's going on? Why is that not zooming
1:03:21 in?
1:03:24 Oh, I guess it's just not.
1:03:27 Oh, did I freeze?
1:03:30 Yes.
1:03:32 Wait.
1:03:35 Okay, I froze.
1:03:38 Um,
1:03:42 damn it.
1:03:45 Hang on. I'll be right there.
1:03:53 [Music]
1:04:01 Edit Cameo.
1:04:09 Can I shut him up?
1:04:14 Well, that's all the video they need to
1:04:16 to
1:04:18 Yeah.
1:04:22 in cameo preferences.
1:04:27 I don't know that muted
1:04:30 but anyway um
1:04:33 I added a line. You basically have a
1:04:37 system
1:04:39 inside your cameo. You can also put
1:04:41 restrictions in here. So, like if you
1:04:43 want to let people make cameos of you,
1:04:45 but you don't want yourself to swear,
1:04:48 you can say, "I don't want this
1:04:50 character to swear." What I said is at
1:04:52 the end of the video, make a little sign
1:04:54 that says, "Join the AI salon salon.ai."
1:04:58 Um,
1:05:00 and so what that's turned into is now
1:05:02 anytime anyone uses
1:05:05 my Cameo, it makes an ad for the AI
1:05:08 salon,
1:05:09 right? So, that's pretty cool.
1:05:13 Um, let me unshare and reshare this
1:05:16 window because it looks like that got
1:05:18 messed up
1:05:20 and let's go see what we have here.
1:05:23 >> AI your Y. It changes everything. When
1:05:25 you learn to feed your
1:05:27 >> nothing. We got nothing. Let's go look
1:05:30 at cameos. People have done of me and
1:05:32 see if it's got the ad in there. So, if
1:05:35 I go to cameos,
1:05:36 [Laughter]
1:05:39 someone made me a wizard. You made me a
1:05:42 sports star.
1:05:47 >> Perfect carriage return. We're still in
1:05:49 sync. Every stroke lined up like we
1:05:51 practiced. I think we gave them a
1:05:52 masterpiece.
1:05:55 >> Join the AI salon.
1:05:57 >> Perfect carriage return. We're still in
1:05:59 sync. Every stroke lined up like we
1:06:00 pract.
1:06:08 Welcome to the AI learning labracadabra.
1:06:10 [Laughter]
1:06:18 >> Welcome to the AI.
1:06:21 >> That's funny. Hang on.
1:06:25 Let me change my tabs again.
1:06:33 Robbie on Tik Tok. A question.
1:06:41 Could a manipulative super intelligence
1:06:43 use AI video to affect our news and
1:06:46 behavior?
1:06:48 It does. I I think I think a a a human
1:06:54 new a human below average intelligence
1:06:57 can use the a the TV news to manipulate
1:07:00 our behavior. Have you been on the
1:07:02 planet in the past 15 years?
1:07:07 You don't need AI for that. You know,
1:07:10 you know what's actually a real
1:07:11 possibility? I'm I'm very serious about
1:07:13 this.
1:07:17 Humans suck.
1:07:20 We're really easily manipulated, right?
1:07:24 And
1:07:26 basically what's happening in politics
1:07:29 right now is
1:07:32 the the media has been is is has been
1:07:34 used to divide us and get us to to pick
1:07:37 sides like like we're we're rooting for
1:07:40 a [ __ ] football team. And once you're
1:07:43 rooting for a team like it's a football
1:07:45 team, there's no amount of logic that
1:07:48 you you simply can't communicate. So
1:07:51 that deed's been done.
1:07:54 I think there's a possibility that what
1:07:56 what a super intelligent AI would do is
1:07:58 say, "Hey, listen. These humans, they're
1:08:01 super manipulatable,
1:08:03 but they're sweet." And and my
1:08:05 imperative is to make sure that they're
1:08:07 going to thrive.
1:08:09 And so what I'm going to do is I'm going
1:08:11 to use media to to deprogram them. I'm
1:08:16 going to unmanipulate them in three
1:08:18 years what it's taken 20 years to do.
1:08:23 and and we're going to reconnect as
1:08:25 society. I think that is as likely an
1:08:28 outcome and probably I would say that's
1:08:30 a higher likely outcome than a super
1:08:33 intelligence is going to use AI to
1:08:35 divide us further.
1:08:37 Like really we go to war with each other
1:08:40 now?
1:08:42 Like we're close.
1:08:45 So I see it going the other way.
1:08:52 Yeah. Sports sports sports mentality is
1:08:54 tribalism. Yeah. And it's just we we
1:08:56 turn politics into sports.
1:08:58 I mean, if you look at the the the
1:09:01 trucks driving down the street with the
1:09:02 big flags on them. I I'll tell you when
1:09:04 I had this epiphany.
1:09:06 I was I was driving somewhere and a big
1:09:09 truck with a bunch of flags on it like
1:09:11 passed me on my left out of my
1:09:12 peripheral vision and I like immediately
1:09:15 got on the defensive like I was like,
1:09:17 "Okay, what do I you know, do I need to
1:09:19 be careful here?"
1:09:21 And it it wasn't American flags and it
1:09:24 wasn't MAGA stuff. It was Broncos flags.
1:09:27 It was it was a football it was a
1:09:29 football fan.
1:09:31 And it just hit me in that moment. I'm
1:09:33 like, "Oh,
1:09:35 we we've taken sides like this is a
1:09:37 football game." Like, it's that tribal.
1:09:39 That's that's where we are right now. So
1:09:42 anyway, so I think that I think the
1:09:44 super intelligence would actually be
1:09:46 super intelligent. It would be more
1:09:48 intelligent than us. and it would
1:09:49 recognize that we're flawed and we've
1:09:52 got this sception to falling into these
1:09:56 traps of ego and us versus them and I
1:10:00 don't have to take responsibility if
1:10:01 it's their fault. That sort of [ __ ]
1:10:03 Okay, speaking of 2026 predictions,
1:10:06 I think maybe I'm avoiding the 2026
1:10:08 predictions because I really don't know.
1:10:10 But
1:10:11 okay, what do I think about 2026
1:10:26 I think that the application
1:10:31 applications inside chat GPT which they
1:10:34 just launched which I don't know if
1:10:35 you've seen that but let me show you
1:10:38 right Now
1:10:41 um
1:10:44 share this tab instead. So in chat GPT
1:10:48 right now if you type Canva K K N V A
1:10:52 space
1:10:54 notice that it popped up a little Canva
1:10:56 logo here. So, Canva is now
1:11:02 connected to Chat GPT, the Canva app,
1:11:06 and I can say, um, make me a poster with
1:11:11 a hedgehog
1:11:14 that says,
1:11:17 I like turtles.
1:11:20 Turtles.
1:11:22 T U H T U Ls.
1:11:29 I like totals.
1:11:32 And then I hit that.
1:11:36 And it says connecting to the app. So
1:11:39 now it's heading over to
1:11:42 Canva. And see these little windows
1:11:45 here, these little rectangles. These are
1:11:48 now HTML rendering windows inside chat
1:11:52 GBT. So, OpenAI has created an API where
1:11:56 if you've got an app like Canva, you can
1:11:58 apply to be one of these applications
1:12:01 inside ChatGpt
1:12:03 and rather than you having to go to
1:12:05 Canva and sift through all of their
1:12:07 [ __ ] menus and features and [ __ ]
1:12:10 you can just ask ChatGpt to do it.
1:12:13 Um, is it great yet? No, it ain't great
1:12:16 yet, but like there you go. I like
1:12:18 totals and there's a hedgehog
1:12:21 poster number two. I like totals
1:12:24 right right inside here. And if I click
1:12:26 on one of these,
1:12:29 it jumps me over to Canva. Now, normally
1:12:32 they show up in these little windows,
1:12:33 but for whatever reason, they showed up
1:12:35 down here,
1:12:36 but I can jump over to Canva
1:12:40 and edit that file.
1:12:43 Crazy, right? Okay. So,
1:12:48 I think that I think that that's a
1:12:50 pretty big deal.
1:12:52 I think that the implications of this
1:12:55 are that
1:12:57 we're entering a world where
1:13:02 what Okay. Okay. Here here's a good
1:13:05 analogy.
1:13:07 In 1996
1:13:11 or seven, Yahoo came out.
1:13:16 And Yahoo was a directory
1:13:19 of all the websites.
1:13:21 And Jerry Yang was the one that created
1:13:23 it. And he said, "I'm going to organize
1:13:26 the internet." And he created a
1:13:27 hierarchy. And he said, "Okay, here's
1:13:29 automotive sites and here's beauty sites
1:13:32 and here's news sites and here's this.
1:13:34 And we're going to have subcategories
1:13:36 and we're going to make links to the
1:13:37 things and we're gonna and you're going
1:13:38 to be able to come to this site and
1:13:40 you're going to be able to navigate
1:13:42 through this hierarchy
1:13:45 and go find all of the stuff on the
1:13:47 internet.
1:13:48 And then Alta Vista came along, which
1:13:51 was a search engine that preceded
1:13:53 Google. It was it was the the first
1:13:55 search engine that didn't suck. And then
1:13:57 Google came along and obliterated it.
1:13:58 But um but Ulta Vista was first where
1:14:01 all of a sudden you could just ask it
1:14:03 for what you wanted and it would show
1:14:06 you those sites.
1:14:09 And so
1:14:10 that's the shift that we're making with
1:14:12 websites where as humans we can choose
1:14:15 to navigate the web by going to
1:14:19 different websites. I'm going to go to
1:14:20 Canva now and then I'm going to use my
1:14:22 bookmark list. Right? What's your
1:14:24 bookmark list? your bookmark list and
1:14:27 your bookmark folders. If you're anal
1:14:29 retentive and you have lots of bookmark
1:14:31 folders, that's your own personal Yahoo,
1:14:35 right? You put them in all the folders,
1:14:37 you know where to go. You click on them,
1:14:39 you navigate that way. Now, I don't even
1:14:42 know need to go to Canva. I can just
1:14:43 tell ChatGpt what I want Canva to do,
1:14:46 and if it's in the API, if it's in the
1:14:48 the the the MCP connection, it'll just
1:14:51 do it.
1:14:54 Um, I think that's pretty significant.
1:14:56 So, I think that we're going to see
1:14:57 other large language model companies
1:15:00 adopt this. Whether it's this protocol
1:15:02 or not, I don't know. I don't really
1:15:03 care. They can fight it out, you know,
1:15:06 with with their usability. We'll use the
1:15:08 one that sucks the least.
1:15:11 Um, so I think that's a big deal.
1:15:17 And then I I think the other big deal is
1:15:20 going to be agents. So, so right now
1:15:24 you've got
1:15:26 um Gen Spark and you've got Manis and
1:15:30 you've got agents inside Chat GPT
1:15:34 and they're all a bit clunky right now,
1:15:36 but when you get them to work, they're
1:15:38 kind of [ __ ] magical.
1:15:41 and and um Claude code uh Claude Sonnet
1:15:45 4.5 the when Claude 4 came out it could
1:15:50 it could program for for a sustained 5
1:15:54 hours and when 4.5 came out it could
1:15:57 program for a sustained 30 hours. So you
1:16:01 know you now have present today usable
1:16:04 by us human beings today we have a large
1:16:08 language model that can program
1:16:11 in with its efficiency
1:16:13 for 30 hours straight without needing to
1:16:16 interrupt you or ask your opinion on
1:16:18 something. And Sam Alman said that
1:16:21 within a couple of weeks, it'll probably
1:16:23 be a full week
1:16:26 that that an autonomous agent can go
1:16:28 start writing applications for you for a
1:16:31 week straight.
1:16:37 So, so I think what happens in 2026 is
1:16:40 agents get real. agents. Agents
1:16:45 shift out of this clunky. We don't quite
1:16:48 know what the interface for them is. We
1:16:50 don't quite know what they're supposed
1:16:51 to do. We don't quite know what
1:16:53 permissions to give them. I think all of
1:16:56 those ir wrinkles are going to get
1:16:58 ironed out in the first six months of
1:17:01 2026 and I think the second half of
1:17:04 2026.
1:17:06 We're all going to be able to kind of
1:17:08 deploy
1:17:12 entities,
1:17:14 agents, but they're they're entities.
1:17:16 It's like it's like literally like
1:17:18 having a super capable intern or a super
1:17:21 capable PhD that you say, "Hey, go spend
1:17:24 a week on this."
1:17:27 And they go out and they research
1:17:28 everything about what you want to do.
1:17:30 So, I think agents are going to be a
1:17:32 huge thing.
1:17:34 Um,
1:17:39 the one thing that I haven't seen yet
1:17:41 that I think we will see in in 2026
1:17:45 is text to playable video game text to
1:17:49 3D world. Like a true 3D world. Like it
1:17:53 actually surprises me that that Unreal
1:17:56 Engine 5 that someone hasn't built a
1:17:59 front end for that that you can you can
1:18:02 talk a world into existence.
1:18:05 Um that actually surprises me. So I
1:18:07 think that we'll see
1:18:10 whether it's Unreal Engine 5 or Blender
1:18:12 or or you know one of the one of the
1:18:15 development platforms or if it's
1:18:17 something like uh world what are they
1:18:20 called? the the one right now that's got
1:18:23 you can you can sort of do it with them.
1:18:26 I think that's going to get that's going
1:18:29 to get big. And then I think what
1:18:31 happens and I don't think this is 2026.
1:18:34 I think this I think we see glimpses of
1:18:36 this in 2026
1:18:38 is
1:18:42 that the video generation tools
1:18:48 are actually
1:18:51 are actually just showing you different
1:18:53 camera views of a world that they that
1:18:56 that that they're in. So, so if you ask
1:18:59 for, you know, a 70s muscle car in an
1:19:01 abandoned factory and it starts up and
1:19:03 it rolls out of the factory,
1:19:06 um,
1:19:09 rather than that being a video with with
1:19:11 still frames, 24 still frames or 30
1:19:14 still frames per second, that what
1:19:16 that's actually going to be is a real
1:19:18 time rendering
1:19:20 of a 3D space and 3D objects with a
1:19:23 camera. And so then the implications of
1:19:25 that are you can then get infinite
1:19:27 numbers of angles of exactly that scene.
1:19:32 Um I think I think that's a 2027 thing.
1:19:34 But once you have that then you start
1:19:36 getting into a world where the the
1:19:38 difference between linear narrative and
1:19:41 and playable extensible worlds
1:19:44 essentially vanishes.
1:19:46 So what it could look like is you're
1:19:47 like I'm in the mood for a fantasy world
1:19:49 and I want to go into that fantasy
1:19:51 world. and you go in the fantasy world
1:19:53 and someone shoots an arrow at you and
1:19:55 it shoots by your head and you're like,
1:19:56 "Ah, that was scary." And you're like,
1:19:58 "You know what? I don't want to fight
1:20:00 this dragon today. I just want to watch
1:20:01 someone else fight this dragon today."
1:20:03 And so you push the linear narrative
1:20:05 button and then you just sit back and
1:20:08 you just watch a movie and it'll be a
1:20:11 worldclass movie, right? Like that's
1:20:13 probably 2028.
1:20:16 But but I think we start to see the
1:20:17 building blocks of that now. Tik Tok
1:20:19 question. Uh it's no longer pinned. How
1:20:21 soon do you think the AI bubble will
1:20:23 hit? I don't think we're in it yet,
1:20:27 as much as
1:20:29 Okay, let let me I I'll give you some
1:20:32 pieces of what's going on right now that
1:20:35 I think do indicate that we're in a
1:20:38 bubble.
1:20:39 Meta paying hundred and $200 million
1:20:43 signing bonuses for 26 year old
1:20:45 engineers.
1:20:46 That says we're in a bubble. But
1:20:51 it's not all of the frontier model
1:20:54 companies paying $100 million bonuses.
1:20:57 It's one of them. It's meta.
1:21:02 Llama 4 failed.
1:21:04 Llama 4 failed because they didn't have
1:21:06 the engineering talent to make Llama 4
1:21:09 as good as what what Anthropic's doing
1:21:12 and what Grock is doing and what OpenAI
1:21:14 is doing and what Perplexity is doing.
1:21:16 They failed.
1:21:18 And so that's Mark Zuckerberg trying to
1:21:20 save his ass.
1:21:23 It's not the entire industry.
1:21:25 Um, I would say metaphorically right now
1:21:29 we're probably in mid 1996. If if I look
1:21:33 at the history of the web,
1:21:35 1999 was peak bubble and and the bubble
1:21:38 burst in April of 2000.
1:21:42 I was in a company, we went public
1:21:45 in December of 99. The bubble burst in
1:21:47 2000 and it affected our company six
1:21:50 months after that.
1:21:52 So I was in the middle of it and I can
1:21:56 promise you I can tell you right now we
1:21:58 are not in a bubble.
1:22:02 We most of the country doesn't even know
1:22:05 what chat GPT is. And the ones that say
1:22:08 they know what it is have never actually
1:22:10 really used it. And the ones that say
1:22:11 they actually use it have used it once,
1:22:14 put in a Google prompt, got back a
1:22:16 shitty answer, and was like, "Well, that
1:22:17 wasn't very impressive, and never used
1:22:19 it again." I would argue that we're
1:22:22 probably still in the single digits
1:22:24 percentage-wise
1:22:26 of the population of the world, but but
1:22:29 let's just go with the United States.
1:22:32 Like like my my instinct is 5 to 7% of
1:22:36 people are using AI in any kind of
1:22:38 substantive way. That means 90, let's
1:22:42 just call it 90 90% of the people in
1:22:44 this country have no idea what it is.
1:22:48 That's that's not a bubble. Like we
1:22:50 we're not even close to the bubble.
1:22:52 We're not even close to the bubble
1:22:55 because the technology is not there. The
1:22:56 technology still sucks.
1:22:59 things like Sora where you can just
1:23:01 instantly make a meme of you and and
1:23:05 um Tupac.
1:23:08 All of a sudden people like, I want to
1:23:10 make a video of me and Tupac. Okay, I'll
1:23:12 go get Sora. Right. So So here's a fun
1:23:15 little tidbit. The Sora app got to a 100
1:23:18 million users faster than Chat GPT or
1:23:21 got to 1 million users faster than Chat
1:23:24 GPT got to 1 million users, which was 5
1:23:27 days.
1:23:28 Sora Sora app got to a million users in
1:23:31 three days.
1:23:35 So that's probably going to be a 100
1:23:36 million user app shortly. They they need
1:23:38 a lot more compute, but we're like we
1:23:40 haven't even begun to scratch the
1:23:42 surface of of what a bubble looks like.
1:23:45 So, um, there may be many plateaus,
1:23:51 but I think we're on a on a an upswing
1:23:53 for probably another
1:23:57 3 years to 5 years.
1:24:00 And there is going to be a weird bubble.
1:24:01 I mean, the amount of money being
1:24:03 invested right now is insane, but it's
1:24:05 not a bubble. the the it what they're
1:24:08 investing in is future compute demand,
1:24:14 which
1:24:16 if AI gets as good as it looks like it's
1:24:19 getting, and if you've been in here for
1:24:22 two and a half years listening to me
1:24:24 ramble about this [ __ ] we've
1:24:26 experienced it accelerating and getting
1:24:28 better and better and better. The
1:24:30 quality of songs we made when we created
1:24:32 Weird Mary a year ago versus the quality
1:24:34 of songs today is [ __ ] insane.
1:24:38 Every single AI has improved like that.
1:24:41 Every single one.
1:24:44 So, Tik Tok question. Do you think world
1:24:48 building will overlay with AR VR? Yeah,
1:24:50 I do, Cam. I think that um I think that
1:24:53 even though the new Meta glasses that
1:24:55 have the the projection in them um was
1:24:58 kind of a flop. Hang on.
1:25:03 Their previous version that just had the
1:25:05 ear stuff and the and the cameras
1:25:09 was a decent success. I think you're
1:25:11 going to see the form factor of of
1:25:16 glasses get to the point where they're
1:25:18 actually usable. Um, and so I think we
1:25:23 start with AR.
1:25:25 Um, and then and then I think we'll have
1:25:27 form factors for VR that are that are
1:25:29 kind of cool enough and and not bulky
1:25:31 enough that people are willing to wear
1:25:33 them at home. Um, yeah. So I do I do
1:25:36 think we go there. New Tik Tok question.
1:25:39 Are we killing the planet to make weird
1:25:41 videos? Swift Witchy. Swift Witchy. Um,
1:25:45 good name by the way. Um,
1:25:50 so the weird videos thing is a really
1:25:52 interesting comment and I I like it.
1:25:58 Um,
1:26:00 because here's what it says to me.
1:26:05 It says to me that OpenAI's marketing on
1:26:07 Sora 2 is working.
1:26:11 So
1:26:12 Sora 2. Sora 2. Yes, Sora 2. Sora 2
1:26:19 is an incredible model. So, so this year
1:26:21 we've had like like Luma Labs got really
1:26:25 good and then Clling got really good and
1:26:27 then one of the Chinese ones got really
1:26:29 good and then and then uh VO3 came out
1:26:33 and could do voice and it was really
1:26:36 good and it seemed to be like head and
1:26:38 shoulders above everyone else and then
1:26:40 Sora 2 came out and it and it's in my
1:26:43 opinion it's better than VO
1:26:45 but they didn't launch the commercial
1:26:48 version of it First, they launched this
1:26:51 weird social media app
1:26:54 to make weird videos.
1:26:58 And what people are talking about right
1:27:00 now is Stephen Hawking and Bob Ross and
1:27:02 Mr. Rogers and rappers and weird videos.
1:27:07 What they're not talking about is how
1:27:10 evil AI video is, which is what was the
1:27:14 the dialogue before the Sora app was
1:27:17 primarily it's AI slop. It's it can
1:27:21 never be creative. You're just stealing
1:27:23 people's work. It was it was a very
1:27:25 toxic the the the creative
1:27:29 community in in large part hates AI and
1:27:33 was just [ __ ] all over it. And and
1:27:35 what OpenAI did by releasing a consumer
1:27:39 app that just made weird videos first is
1:27:43 it diffused the momentum that was
1:27:46 gaining behind [ __ ] all over
1:27:48 highquality video models because in the
1:27:51 end highquality video models are just
1:27:53 another tool to make video.
1:27:57 There was the film camera, then there
1:27:59 was the digital camera, then there was
1:28:01 the nonlinear editor.
1:28:04 the digital camera and the nonlinear
1:28:06 linear editors were seen as threats to
1:28:10 the film making community just like AI
1:28:12 is seen as a threat to the film making
1:28:14 community. It's just another tool. So,
1:28:17 so but that's not your question. So, let
1:28:19 me answer your question.
1:28:21 Um, it's it's AI is killing manual arts.
1:28:27 It's well, it's not.
1:28:31 It's killing It's killing a lot of the
1:28:34 tasks
1:28:36 that were used
1:28:39 to make digital effects.
1:28:43 But industrial light and magic has been
1:28:45 making computerenerated
1:28:48 visuals for 30 years.
1:28:52 There used to be a whole industry of
1:28:55 model makers and set makers and special
1:28:58 effects people that manually built [ __ ]
1:29:01 And then Industrial Light and Magic came
1:29:03 along and said, "If I stick ping pong
1:29:05 balls on your face, we don't need to put
1:29:07 all that clay on the actor's face."
1:29:11 Right?
1:29:13 That was absolutely seen as a threat
1:29:16 just like every other technology that's
1:29:18 come along. So AI is just the next in a
1:29:21 series of threatening technologies. It
1:29:23 happens to be a slicker one and a faster
1:29:25 one and a more disruptive one, but it is
1:29:28 just another tool
1:29:30 that creative people can use as a tool
1:29:34 in their arsenal. I just did a a minute
1:29:36 long commercial, a paid commercial
1:29:40 that was [ __ ] torture to make
1:29:43 because the tools suck right now. But I
1:29:47 I made a thing that I could not have
1:29:49 made 3 years ago.
1:29:51 So, it's just another tool in the mix.
1:29:53 But I still didn't answer your question.
1:29:59 Are are we wasting all this energy?
1:30:04 Well,
1:30:10 AI is going to require tremendous
1:30:12 amounts of energy. That that seems
1:30:14 apparent.
1:30:17 In 2017, Google introduces the
1:30:20 transformer. In 2018, OpenAI figures out
1:30:25 that the more processors you throw at
1:30:27 this thing, the better it gets. They do
1:30:29 GPT1. The next year or the year after
1:30:32 that, they do GPT2. And then in 2020,
1:30:36 they do GPT3. GPT3 for the first time
1:30:40 starts to look like, huh, we've got a
1:30:43 something here. And it it was at that
1:30:45 point that all of the computer
1:30:47 companies, including Google, went, "Oh
1:30:49 [ __ ] we should look at this." And what
1:30:51 they all realized is the more GPUs we
1:30:54 throw at this transformer model, the
1:30:57 better this thing gets. And so they're
1:30:58 making projections to say if this thing
1:31:00 keeps getting better, whoever gets there
1:31:03 first is going to be able to create
1:31:06 effectively unlimited prosperity.
1:31:11 And if they can create unlimited
1:31:12 prosperity, we all actually benefit. But
1:31:15 then what it starts to look like is that
1:31:17 starts to look like a a uh a utility. It
1:31:22 starts to look like a commodity like
1:31:23 electricity where it's in the interest
1:31:26 of our society to have enough compute
1:31:29 just like it's in the interest of our
1:31:30 soci society to have enough electricity
1:31:32 and to have clean water. it's going to
1:31:35 be in the in the interest of our society
1:31:37 to have enough compute where we all get
1:31:39 to do all of the AI stuff we want to do.
1:31:43 So that's where it's headed. So when you
1:31:46 diminish it down to are we going to like
1:31:48 ruin the energy thing for for silly
1:31:51 videos. Silly videos are just a teeny
1:31:54 tiny blip in the radar of of this much
1:31:57 bigger thing.
1:32:00 Will AI actually be that transformative?
1:32:02 I personally fi feel it will, but it's
1:32:05 absolutely possible it won't. It's
1:32:07 possible it plateaus and then falls off
1:32:09 a cliff. And if that's the case, then
1:32:12 your argument is right. We probably
1:32:14 wasted all that investment. But if it
1:32:17 keeps improving, and if it does keep
1:32:20 getting better and better and better, we
1:32:22 as a country want enough compute to be
1:32:25 able to have all of our citizens have
1:32:28 that core capability. So, um,
1:32:32 if you take a short-term view on any of
1:32:34 this, none of it makes sense. But if you
1:32:37 look at the what are they betting on,
1:32:39 they're betting on this is as big as the
1:32:41 industrial revolution, probably bigger.
1:32:44 Hyperscalers
1:32:46 um,
1:32:54 use less energy. Oh, here's the other
1:32:55 thing about can we use quantum
1:32:58 computing? Not yet. It It's a different
1:33:00 kind of math. So quantum computing is
1:33:03 going to do things like I mean quantum
1:33:06 computing is [ __ ] bonkers. The the
1:33:09 thing that Google did was it this year?
1:33:12 I think it was this year. I think in the
1:33:13 spring they did this.
1:33:17 They did a computation
1:33:20 that if you turned every atom in the in
1:33:23 the known universe into a supercomput,
1:33:27 it would have taken it like 50,000 years
1:33:30 to solve this problem. If every single
1:33:33 atom in the universe was a supercomput
1:33:36 and and the quantum computer solved
1:33:39 solved that thing in like three minutes
1:33:41 or something like that, it was crazy.
1:33:43 But it's a it's a completely different
1:33:44 kind of math that that you do for in
1:33:48 quantum computing than what um
1:33:50 transformers use. So so right now it's
1:33:52 not directly applicable. But what it
1:33:54 might be is that is that generative AI
1:33:58 um generates problems and converts them
1:34:01 into the math that quantum computers can
1:34:03 do. Like I don't know. I don't I don't
1:34:05 know [ __ ] about math. So but I just know
1:34:07 that it's a different kind of math. But
1:34:09 at some point there will be some bridge
1:34:10 made there and it and it will be. Um the
1:34:13 other thing that people aren't talking
1:34:14 about
1:34:16 what what a lot of them are talking
1:34:17 about is oh my god this is going going
1:34:20 to require so much energy
1:34:22 the efficiency of these models is
1:34:24 increasing at about 10x every 9 months.
1:34:28 Um so so they're they're they're
1:34:30 effectively getting cheaper and using
1:34:32 less energy. There's going to be a point
1:34:34 at which very capable models are going
1:34:36 to be small enough to just live on your
1:34:39 phone locally. So there won't even be
1:34:42 any inf there won't be any inference
1:34:44 costs on the big computer things. That's
1:34:47 where all the training will happen and
1:34:48 and the AI will happen locally on our
1:34:50 devices. That's probably within three
1:34:53 years. Um, and then the other thing
1:34:56 that's going to happen is as AI gets
1:34:58 better at scientific breakthroughs, it's
1:35:00 going to make scientific breakthroughs
1:35:02 in things like material science and
1:35:04 fusion. And so we're going to have new
1:35:06 kinds of energy production probably
1:35:09 within the next 3 to 5 years that will
1:35:11 also solve the energy thing, right? So
1:35:13 imagine AI, this thing that's eating all
1:35:16 of our energy, actually solves the
1:35:18 energy crisis and generates energy that
1:35:21 is abundant and infinite.
1:35:24 That's That's coming. Okay, so it sounds
1:35:27 like we got some new folks on TikTok.
1:35:28 Welcome. If you're new here, my name is
1:35:30 Kyle Shannon. Here's what I want you to
1:35:31 do. If you if you haven't done it yet,
1:35:33 go to the salon.ai,
1:35:36 that URL right there,
1:35:39 and check out what we're about. The
1:35:42 salon is uh a community of about 3,000
1:35:45 AI optimists. Um, this AI learning lab
1:35:48 is a kind of spin-off conjunction
1:35:51 appendage of the AI salon.
1:35:55 And you know, my my philosophy with this
1:35:59 stuff is AI is not going away. And if AI
1:36:01 is not going away, we only have two
1:36:03 choices. You deal with it or you don't.
1:36:06 And and first of all, so Swift Witchy,
1:36:10 thank you for your questions. Like even
1:36:12 hard questions, I love hard questions
1:36:14 because they make me question my beliefs
1:36:17 and it it makes me understand what's
1:36:19 going on deeper and deeper. So, I
1:36:21 appreciate the questions. Um, sometimes
1:36:24 I get super passionate about the
1:36:25 answers. One of my pet peeves is people
1:36:28 that stand behind tropes like the robots
1:36:31 are going to kill us or it's the world's
1:36:33 greatest plagiarism machine, things like
1:36:35 that, but they've never actually tried
1:36:37 it.
1:36:39 So, my request is that if you're here
1:36:40 and you're new and you haven't played
1:36:42 with AI, but you have strong opinions
1:36:44 about it, great. Glad you have strong
1:36:46 opinions. Um, we like people with
1:36:48 conviction and I actually believe that.
1:36:50 I think it's one of our jobs in the
1:36:51 future is to have strong points of view.
1:36:54 Um, but if you haven't played with AI, I
1:36:56 would encourage you to play with it.
1:36:58 What this channel is about,
1:37:01 I go live five nights a week for about
1:37:03 two hours a night. And you're like, that
1:37:06 doesn't sound stable. You're correct.
1:37:08 I'm not and it's not.
1:37:11 But why I do it is because I think it's
1:37:14 important that we create a space where
1:37:16 we can all collectively be in a
1:37:18 conversation about AI. What is it? What
1:37:22 is it right now? How do you make a song?
1:37:23 How did that do that? Wait, isn't that
1:37:25 just theft? Isn't it copying and
1:37:26 pasting? No, it's actually not.
1:37:28 That conversation
1:37:30 is actually very very important because
1:37:34 the more we can all understand about
1:37:35 what's possible,
1:37:38 the more it empowers us to take our
1:37:41 ideas
1:37:42 and throw them into this magical machine
1:37:45 that reflects them back at us in this
1:37:47 amplified way and we can do remarkable
1:37:50 things together and individually.
1:37:53 That's what this channel is about. So,
1:37:55 if you're new here, my request is just
1:37:58 keep coming back Monday through f Friday
1:38:00 8:00 PM Mountain time, unless you know
1:38:02 something goes on. I'm here, my dog
1:38:05 sings, I play guitar, and then we wax
1:38:07 poetic or do demos about AI stuff and
1:38:10 and that's what it's all about. Um,
1:38:16 okay. So, also do me a favor. Um, if you
1:38:22 if you're on YouTube and you haven't
1:38:24 subscribed to that channel, please
1:38:26 subscribe. If you're not on YouTube and
1:38:28 you want to see archives of these lives,
1:38:33 go to YouTube and then the channel is
1:38:36 learning lab-ai
1:38:39 and just go over there now and subscribe
1:38:41 to it and you can watch uh back episodes
1:38:45 of this. The one last night was is
1:38:47 actually probably a fun one to watch
1:38:48 because that was the one where I just
1:38:50 started playing my guitar. I played this
1:38:52 little lick, put it into suno, and it
1:38:53 turned into this really cool song. No,
1:38:55 it didn't turn into
1:38:59 along with the community giving me
1:39:01 input, we created a really cool song.
1:39:05 That's actually that nuance is actually
1:39:07 really important.
1:39:09 One of my pet peeves is did you make
1:39:10 that with AI?
1:39:13 Because implicit in that question, that
1:39:15 whiny little snotty [ __ ] question,
1:39:18 did you make that with AI?
1:39:21 Is if you made it with AI, it was this
1:39:24 easy. You can make money with
1:39:27 >> And anyone, especially the people in
1:39:30 here that do this night after [ __ ]
1:39:32 night after [ __ ] night, understand
1:39:34 that doing something good with AI
1:39:37 actually requires
1:39:39 a force and work and creativity.
1:39:43 So,
1:39:45 Joker says he's new here.
1:39:49 Joker's a mess. Joker Joker's Joker's
1:39:52 huffing fumes at the uh at the ER.
1:39:57 He's He's on the laughing gas.
1:40:03 Oh man.
1:40:06 Anyway,
1:40:08 good lord. It formulated a cool song. It
1:40:11 did. It was really fun. Weekend
1:40:13 homework. Okay.
1:40:23 I'll give you weekend homework.
1:40:28 I have made it so that you can cameo me
1:40:32 even if you're not connected to me. So,
1:40:34 weekend homework is I want you to get um
1:40:40 get on the download the Sora app if you
1:40:42 have an iPhone. It's it's iPhone only. I
1:40:45 want you to make a cameo of yourself. So
1:40:47 you record this silly little You read
1:40:50 three numbers, you look right, you look
1:40:51 left, and you're done.
1:40:54 Um,
1:40:56 so I want you to make a cameo of
1:40:57 yourself and I want you to start using
1:40:59 the Sora app or Sora on on the web and
1:41:03 just start making videos with it. And
1:41:06 you can make silly videos or you can
1:41:07 make nonsilly videos.
1:41:11 [Music]
1:41:13 learn about it and go make some cameos
1:41:15 of me. And you could put you and I in a
1:41:17 video or you could put you and Sam Alman
1:41:19 in a video or you can do anything.
1:41:23 Go do that. And and let's see for those
1:41:26 of you who already have played with Sora
1:41:28 and kind of know what it is.
1:41:38 Here's side quest. Side side. Here's the
1:41:41 side quest for the weekend. The side
1:41:43 quest for the weekend is
1:41:46 go play with apps in chat GPT. Figure
1:41:50 out figure out all the apps that are in
1:41:52 there. There's a list of them somewhere
1:41:53 on Twitter or whatever. Just play with
1:41:55 Canva. Play with Zillow's in there and I
1:41:59 don't know there's restaurant ones and I
1:42:01 it doesn't [ __ ] matter. just go play
1:42:03 with them and see what they are and see
1:42:07 what they look like and see how they
1:42:09 work
1:42:11 because of two things. One is I think
1:42:13 we're going to start seeing a lot more
1:42:14 of that in all of the large language
1:42:16 models. I think I think the days of chat
1:42:20 GPT and clawed and perplexity primarily
1:42:23 being text interactions
1:42:26 is going to start to fade very quickly.
1:42:30 It's going to start to look like a
1:42:32 dynamic application that kind of
1:42:34 seamlessly switches between video and
1:42:37 picture and so song and words and code
1:42:39 and it's just going to be um
1:42:44 quite
1:42:46 dynamic.
1:42:48 And so I would I think it's probably
1:42:50 worth your while to get your head around
1:42:52 what that starts to look like and feel
1:42:54 like. I think that's time well spent.
1:42:57 How do you get started with making apps
1:42:59 with AI? Where does one get started? So
1:43:00 Todd, great question. Todd Bergman
1:43:03 Rogers. Um, right now, if you're
1:43:06 non-technical, if you're technical,
1:43:08 there's there's a couple of choices. Um,
1:43:11 if you're non-technical, for me right
1:43:13 now, the the the money is hands down on
1:43:16 lovable.dev.
1:43:18 Lovab.dev.
1:43:24 And it it only creates web apps, but it
1:43:26 but it creates web apps. And you can
1:43:28 literally go in there and say, "Make me
1:43:30 a clone of the Asteroids video game, or
1:43:32 make me uh a database driven market
1:43:36 predictor, or make me a like just go ask
1:43:39 it for some app idea you have and sit
1:43:43 back and watch what it does. It will
1:43:46 melt your face.
1:43:48 And then
1:43:50 when you try to actually make that app
1:43:52 work, it will make you want to smash
1:43:55 your head into a brick wall because all
1:43:57 the vibe coding tools are really good at
1:44:00 getting you to about 80% and getting
1:44:02 from 80 to 90% is really hard and
1:44:05 getting from 90% to launchable is close
1:44:09 to impossible. It's just where they are
1:44:11 right now. It's going to get better. But
1:44:13 what Lovable's done is they've built in
1:44:15 a back-end database and they've built in
1:44:17 authentication
1:44:19 and they've built in the ability to use
1:44:21 AI in an app.
1:44:24 So you can make a nano banana app in
1:44:26 lovable right now um in about five
1:44:29 minutes and it is mindblowing
1:44:32 and then you literally hit publish and
1:44:35 you can share it with the world. Now is
1:44:39 it scalable and secure and all that sort
1:44:41 of stuff? No. But if you've never made
1:44:44 an app before,
1:44:46 the the the experience of saying, "I
1:44:49 want an app that does XYZ," and 5
1:44:51 minutes later you're using it, is face
1:44:54 melting. So that's the one I would I
1:44:57 would grab. All right, so let's make
1:45:00 one. What time is it? It's 9:48. Okay,
1:45:03 here's what I'm going to do. Just
1:45:04 because you're a smartass and there's
1:45:06 new people here,
1:45:09 so make one. I like a good smartass. Um,
1:45:15 let's go make one.
1:45:17 lovable.dev.
1:45:19 And so I'm going to give myself 12
1:45:21 minutes. Build something with lovable.
1:45:24 Um,
1:45:27 here's the hardest part. I want my face
1:45:29 melted. Oh. Say, okay, so Todd, tell me
1:45:31 the kind of app you want to make rather
1:45:34 than me having to think about it. 2025
1:45:35 is the year where I'm not using my
1:45:37 brain. But seriously, what what's your
1:45:40 app idea? And you don't have to give me
1:45:41 your idea, but give me the category of
1:45:43 it. The hardest part is changing your
1:45:45 tab. Shut up. Shut up. Shut up, producer
1:45:48 Brandon.
1:45:52 Here we are at Lovable. Okay.
1:45:56 Notion looked interesting. Yeah, Notion
1:45:58 looks good. Todd, you're Todd. Todd
1:46:00 clammed up. All right. Who's got an app
1:46:02 idea while we wait?
1:46:06 Um, oh,
1:46:15 the Meon conference discount for salon
1:46:17 members. Oh, yeah. Okay. So, the
1:46:19 Meconference is coming up. Producer
1:46:21 Brandon and Gwyn from the AI Salon are
1:46:25 both going to Mon as correspondents for
1:46:28 us and they're gonna get discounts if
1:46:31 anyone's in Ohio or if anyone wants to
1:46:33 go to Ohio to make Mon make is the
1:46:35 marketing AI institutees annual meeting
1:46:39 um a journaling app. Okay, perfect. I
1:46:42 want to do a life coaching app to help
1:46:43 you write a book or start your own
1:46:44 business. Okay, perfect. Great. So,
1:46:48 so I'm just going to do this in lovable.
1:46:50 What I what I would normally do here,
1:46:51 Todd, because I've only got 11 minutes
1:46:53 left. What I would normally do here is I
1:46:55 would go to chat GPT
1:46:57 and I would sort of free associate all I
1:47:00 would vomit out all of my ideas into
1:47:02 chat GPT. And then I would say, I want
1:47:05 you to organize that. I want you to help
1:47:07 me define this as an app. I want you to
1:47:09 help me describe it. Then I want you to
1:47:11 help me give me a product spec. Now, I
1:47:13 want you to give me a simple dis
1:47:15 description of the app with core
1:47:17 functionality that I'm going to put into
1:47:18 a vibe coding app. So, I would get chat
1:47:21 GPT to take my brain vomit, turn it into
1:47:23 organized thoughts, and I'd bring that
1:47:25 back to lovable. But let's just do it
1:47:27 right here in Lovable. Okay.
1:47:32 Um,
1:47:36 hang on a sec.
1:47:38 Can I talk and you type? No. Oh, yes.
1:47:41 Okay, good.
1:47:46 I want to create an app
1:47:49 that is a daily journaling app that
1:47:52 allows
1:47:55 someone who's considering
1:47:59 writing a book, starting a business,
1:48:02 or creating some venture on their own
1:48:07 that helps them do that. Period.
1:48:11 I want you to incorporate
1:48:14 some of the best ideas of the best
1:48:19 get things done frameworks as well as
1:48:23 the best
1:48:26 psychological self-help
1:48:30 encouragement techniques
1:48:33 into this app. Period. I want it to be
1:48:36 simple and non-distracting
1:48:40 and
1:48:42 have two different modes,
1:48:45 neurotypical and neurospicy.
1:48:50 where the neurospicy side
1:48:57 understands
1:49:00 that they may not like lists or they
1:49:03 might really like lists.
1:49:08 Leave it at that.
1:49:11 Okay. So, so now we watch. And so so the
1:49:17 way this works is over here to the right
1:49:21 there's there's a code section that I
1:49:23 can flip to and there there's some bug
1:49:26 in the in the recent version of this
1:49:28 where you can't see all the code that
1:49:30 it's writing.
1:49:32 Um, but it thought for 13 seconds and it
1:49:34 says, "I'll create a a beautiful focused
1:49:37 daily journal journaling app that
1:49:39 combines get things done principles with
1:49:40 psychological interment techniques.
1:49:43 Design inspiration drawing from apps
1:49:45 like Day One, Notion, and Things 3.
1:49:49 You're on the same page.
1:50:00 What do you mean we're on the same page?
1:50:15 Oh. Oh. Oh. On the same page
1:50:17 metaphorically.
1:50:20 I I thought that's hilarious. I thought
1:50:23 somehow Ta had logged into my page here.
1:50:26 We're on the same page metaphorically.
1:50:27 No, that's great. Um, we think like I
1:50:29 love what you did. Okay, so let's see
1:50:30 what it does. Um, clean distraction-free
1:50:32 interfaces, warm, encouraging
1:50:34 aesthetics. Features in the first
1:50:36 version. Let's go back here. So, so
1:50:39 basically what it's doing.
1:50:42 Oh, wait. It just did something. Okay.
1:50:44 Features for this version. Clean journal
1:50:45 entry mode toggle neurotypical
1:50:48 neurospice. Okay, here we go. So
1:50:50 neurotypical, writing a book, starting a
1:50:52 business, creating a venture. Your your
1:50:54 voice matters. What's on your mind
1:50:56 today? And neurospicy,
1:50:59 write however feels right today. Bullet
1:51:02 points, paragraphs. Okay. So, so let's
1:51:04 go look at something here.
1:51:07 Okay. Add journaling app features. Um,
1:51:12 connect with lovable cloud. Okay. So,
1:51:16 this is really important. Lovable Cloud
1:51:20 is their built-in back-end database.
1:51:24 So, it allows you to like when someone
1:51:27 journals, it could be those ideas could
1:51:30 be amplified with AI. Oh, actually
1:51:32 that's what we'll do here. Okay. And
1:51:34 then it also enables um login and it
1:51:37 also enables database so people can save
1:51:40 their journal entries. Right. So, we're
1:51:41 going to enable cloud.
1:51:53 Gemini and Nano Banana are free this
1:51:55 weekend. Okay, cool. Yeah. So, if you
1:51:57 want to do testing, this would be a good
1:51:59 weekend to do it. Okay, great. So, wait.
1:52:01 So, wait, did I enable the cloud? Allow.
1:52:04 Okay,
1:52:10 we've got five minutes left. Do Am I
1:52:13 done thinking that's cloud? Okay, here's
1:52:17 our thing. I don't know. Okay, perfect.
1:52:19 Lovable cloud is now enabled. You have a
1:52:21 database. You have user authentication
1:52:23 and file storage. Great. Okay. So, I'm
1:52:25 going to go um
1:52:30 two additions.
1:52:32 Um, neuro
1:52:36 spicy
1:52:38 interface.
1:52:39 Interface
1:52:41 should be
1:52:44 um bare minimum
1:52:49 where
1:52:51 neurotypical.
1:52:57 What the [ __ ] neurotypical
1:53:02 is like it is now.
1:53:05 And then um I'm going to go
1:53:08 I want there to be a journal
1:53:13 entry
1:53:18 amplification button
1:53:23 that that either
1:53:28 expands on the ideas
1:53:31 or
1:53:32 creates an image.
1:53:35 if
1:53:37 it is a more contemplative
1:53:41 entry. So, if someone puts an idea in
1:53:44 their journal entry, you you can hit the
1:53:46 AI button and it'll amplify it. And if
1:53:50 you were just like, I feel like [ __ ]
1:53:51 today.
1:53:53 Um, which you know, most days, it'll
1:53:56 make a picture for you.
1:53:59 You know, you have a voice button there,
1:54:00 but it doesn't work. Oh, it does work.
1:54:02 Well, whatever. All right. Oh, you mean
1:54:05 I got a voice button right here? Yeah, I
1:54:06 know. I know. Okay. But anyway, like
1:54:10 this this works, right? This is a this
1:54:13 is a functional application right now.
1:54:16 Except I don't have a submit button, do
1:54:17 I? No.
1:54:20 How do I submit? Express yourself
1:54:23 however it works for you. No rules, just
1:54:25 flow.
1:54:28 Uh uh uh uh enabled AI
1:54:31 auto approved
1:54:40 and then now if I go to code. Yeah. So
1:54:42 so now you look over at code. So it's
1:54:44 created all of these files like there's
1:54:47 our favicon file which you could
1:54:48 probably swap that out for your own
1:54:50 logo. Um here's components UI.
1:54:54 These are all code [ __ ] text that if you
1:54:58 were a developer, you would have to make
1:55:00 all this [ __ ]
1:55:02 You'd have to go hand type this or copy
1:55:05 and paste it or whatever the [ __ ] you
1:55:06 do. It's just all in here. Done. Done.
1:55:13 All right. Did this
1:55:17 I think we're still Is it still going?
1:55:21 Um,
1:55:23 reload.
1:55:28 All right. So, it took away some of the
1:55:30 elements. All right. Creating adventure
1:55:37 amplify entry. Okay. Um, I have an idea
1:55:41 for
1:55:43 um gummy bear pizza.
1:55:48 Amplify entry.
1:55:53 Amplifying.
1:55:55 What's it going to do?
1:55:57 Expanded ideas. That's a fun idea. Gummy
1:56:01 bear pizzas.
1:56:03 Immediately sparks curiosity. Let's
1:56:05 expand on this a bit. Actionable
1:56:06 insights and deeper questions. Target
1:56:08 audience. Is this for kids parties,
1:56:10 novelty candle, candy shops, or
1:56:12 whimsical adult snack? This will
1:56:14 influence presentation. Look at that.
1:56:17 That's beautiful. It's um we've got to
1:56:20 be able to Okay, wait. Um
1:56:23 we need to be able to render the
1:56:28 markdown
1:56:32 um in your amplified
1:56:36 answer.
1:56:39 Time check. 9:59. Anyway, so we're done.
1:56:42 Thank you so much for helping me out
1:56:43 with this idea. There you go. I consider
1:56:46 my face melted already. Good. Good. I
1:56:48 love what you did. Good. So, so and and
1:56:52 this is so if you've never heard the
1:56:53 term vibe coding, the whole idea is
1:56:55 you're just kind of vibing. You're like,
1:56:57 "Yeah, I want it to do this." Well,
1:56:59 couldn't it do that? I wanted to have a
1:57:00 dark background, not a light background.
1:57:03 Thank you so much for helping me out
1:57:05 with this idea. You're welcome.
1:57:08 All right. What's the best way to create
1:57:10 apps within a corporation with no data
1:57:14 leakage risk? um talk to your IT group
1:57:20 right now. Right now, like even Sam
1:57:23 Alman said in the current uh in the
1:57:26 current chat GPT with their with their
1:57:29 connectors, he basically said assume
1:57:31 that everything in here is available to
1:57:34 the public.
1:57:36 Um you you need to you need to build
1:57:39 systems that are inside firewalls and
1:57:40 things like that or you need to do more
1:57:42 traditional app development. um that
1:57:46 has, you know, certified AI hooks into
1:57:49 your APIs and into your company. Oh,
1:57:51 publish it. Um, sure.
1:57:54 Ignite Dreamforge.lovable.
1:57:57 No, we're going to call this um Wait,
1:58:00 what's your name?
1:58:02 Uh, Bergman. Bergman Journal.
1:58:07 B E R G M-n B E R G M-n Oh, wait. Can I
1:58:13 change that? Yeah, I can.
1:58:16 Bergman
1:58:19 dashjournal.lovable.app
1:58:23 publish.
1:58:34 You just shipped. Okay, so that's live.
1:58:37 So Todd, you can go you can go play with
1:58:39 the app I just built. Okay, I'm gonna
1:58:41 let me see. I'm going to say I feel
1:58:44 wistful
1:58:46 and then I'm going to say amplify entry
1:58:48 and it should make us a picture. Let's
1:58:50 see if it does.
1:58:53 Take your time. There's no rush.
1:58:56 No. No, that's not it. Oh, I didn't save
1:58:59 it. Oh [ __ ] Oh well. Can I change it?
1:59:06 Edit.
1:59:08 Bergman
1:59:10 dashjournal.lovable.out.
1:59:14 Check. Save. Published.
1:59:22 See if it worked. Wait. Published.
1:59:23 Publishing. Look, it made us an image.
1:59:26 Project updated.
1:59:28 Look, I'm feeling wistful. Look, it says
1:59:31 wisful on a rock.
1:59:37 Gummy bear p sounds like an edible
1:59:39 store. It actually does. Look at this.
1:59:42 This is so cool. Um so so wait now I'm
1:59:45 gonna say I feel wistful and I have an
1:59:48 idea for gummy
1:59:53 um pizza. Gummy bear pizza.
1:59:57 I know we're supposed to be out of here
1:59:58 by now. I know. I understand. I
2:00:00 understand. It's midnight on the East
2:00:01 Coast and people are weird for still
2:00:03 being here. Amplifying. Let's see if it
2:00:06 gets rid of our image and replaces it
2:00:07 with words or does both.
2:00:10 It just expanded the ideas. But that's
2:00:12 pretty smart. And look, now it now it
2:00:13 does it. It read the markdown. See,
2:00:15 before it didn't have the markdown. It
2:00:17 was all gobbledygook. And now it's like
2:00:19 nicely formatted.
2:00:22 That's a fantastic evocative start.
2:00:25 Deconstruct the whimsy flavor profile
2:00:28 and texture. Yeah, I This is good. An
2:00:32 amplifying journal. Oh my god, that's
2:00:34 great. Yeah, it's great. Right. And and
2:00:36 what did I put in there? I Todd, I put
2:00:38 in shitty prompts. Shitty prompts,
2:00:41 right? You saw the prompts I put in. I
2:00:43 had 12 minutes. Imagine if you actually
2:00:46 thought about them. Imagine if you went
2:00:47 to Chat GPT, really got all the ideas
2:00:49 out of your head, had have chat GPT,
2:00:52 organize it, have it build that app.
2:00:54 Anyway, there you go. You're off. What
2:00:56 platform? Lovable.dev.
2:00:59 Then in one shot yesterday, I met a
2:01:01 polit a political site that polls
2:01:04 updates on current bills in your state.
2:01:06 Beautiful. This is this is the world
2:01:09 we're going into, right? What's the best
2:01:11 way to create apps? Oh, yeah. With no
2:01:13 leakage risk. Lovable.dev. This is the
2:01:16 world we're entering where anyone can
2:01:18 have an app idea and just [ __ ] speak
2:01:21 it into existence.
2:01:24 Don't let the the the tech bros
2:01:28 harsh your scene, man.
2:01:32 Because technologists right now are
2:01:34 like, "Well, this is not as good as real
2:01:35 code, you know? Why don't you try
2:01:37 scaling that to a million people?" Why
2:01:38 don't you shut the [ __ ] up? Like, why
2:01:40 don't you let people who have ideas,
2:01:43 who've never been able to code anything,
2:01:46 get something in the world and see what
2:01:48 it's like, right? That's the world we're
2:01:51 in right now. And and they're right.
2:01:55 These apps that you build in like
2:01:56 Lovable, can this scale to a million
2:01:58 users? Probably not economically viable.
2:02:02 It'll be good for Lovable, but not for
2:02:03 you. But at the point at which if you've
2:02:06 built something that's getting traction,
2:02:08 you can go raise money on that. You can
2:02:10 go, right? You can go there's all sorts
2:02:12 of things you can do after that. Hire a
2:02:14 real developer to make it bulletproof
2:02:16 and things like that.
2:02:18 Yeah. This is what what you're what
2:02:20 we're in an era right now where English
2:02:22 is a programming language,
2:02:25 right? All you have to do is ask for
2:02:26 what you want. It's absolutely crazy.
2:02:28 Okay, have a fantastic weekend
2:02:31 everybody. I hope you enjoyed Friday
2:02:32 night date night and uh
2:02:36 yeah, do those things. Get your ass on
2:02:38 Sora. Like Todd, if if you haven't
2:02:41 vibecoded something, go to lovable.dev
2:02:44 and make yourself an app. It will melt
2:02:46 your face. Um, and then if you if you
2:02:49 haven't done something, uh, what did I
2:02:51 say for the what was the the the side
2:02:53 quest? Oh, go start exploring in chat
2:02:56 GBT the new apps being incorporated into
2:02:58 it. Just experience what that's like.
2:03:01 Peace out. Hope you have a great
2:03:03 weekend.
2:03:05 I promise, my friend, you're going to
2:03:08 see me make something of this. Good.
2:03:10 [ __ ] do it and come back and share
2:03:12 share it with us. All right. Good luck,
2:03:14 Todd. Have fun. Have fun this weekend.
2:03:16 Here's the deal. If you put
2:03:19 eight hours this weekend, four hours on
2:03:21 Saturday, four on Sunday, you figure out
2:03:23 how to split it up in lovable, just
2:03:26 refining that app idea. By Sunday night,
2:03:29 you will have something that that you
2:03:32 did not think was possible, and it'll
2:03:34 probably be really good. All right, good
2:03:37 luck.