3/20/2025 - OpenAI's New Text-to-Speech: Creating Unique & Disturbing Voices

Description

Chapters

Transcript

0:00 Go go go go go go go go
0:05 [Music]
0:19 go.
0:24 [Music]
0:31 Feeling nervous, aren't
0:35 you? Gentle
0:39 voice in the
0:41 [Music]
0:43 sky. Don't ever let them steal your
0:48 joy and chill away.
0:51 to keep them from running
0:55 wild. They can kick dirt in your face,
0:59 dress you down, and tell you that your
1:01 place is in the middle when they hate
1:04 the way you
1:09 shine. I see you tugging on your
1:13 shirt, trying to hide inside of it. Hide
1:17 how much it hurts.
1:20 Thank you,
1:21 Winston. Let them
1:24 laugh while we
1:27 can. Let them
1:30 spin. Let them scatter in the
1:35 wind. I've been to the movies. I've seen
1:40 how
1:41 [Music]
1:44 it land.
1:47 [Music]
2:01 [Applause]
2:03 Good evening, good people. I went out to
2:06 dinner. Had a lovely dinner at Root Root
2:08 Down here in
2:10 Denver. I did not go to Whiskey Club, so
2:13 I am not drunk. I had a beer. So, if
2:17 you're expecting me to go get twerking
2:19 tonight, not going to be the night for
2:22 it.
2:26 Um,
2:33 wait. Walt Grace desperately hating his
2:37 old place. dreamed of discover a new
2:41 space and buried himself
2:44 alive inside his basement tongue on the
2:48 side of his facement. He's working away
2:51 on
2:52 displacements and what it would take to
2:55 [Music]
2:59 survive. When you're done with this
3:02 world,
3:05 wait. Wait.
3:08 When you're done done with this
3:13 world, you know the next is up to
3:18 you. And for once in his life it was
3:27 quiet as he learned as he learned as he
3:31 as he learned learned how to wait. Learn
3:35 how
3:36 to
3:39 [Music]
3:48 tide. I should probably learn that song
3:50 before I try singing it.
3:51 [Music]
3:54 Huh? Not even a twlet.
3:57 [Music]
4:09 [Applause]
4:15 [Music]
4:49 I just needed to hear Champy going off.
4:52 He went off there, didn't he? Did you
4:54 like that, Champy? You like that? Pretty
4:56 old. Good.
4:58 [Music]
5:06 [Music]
5:14 Sit in this lonely
5:16 town. Wonder when things are going to
5:21 change. Dream my life
5:24 away. Seems these dreams have turned to
5:27 a
5:29 bunch. Get my nerve up, but my past is
5:34 pulling me down.
5:38 Wondering how
5:40 long this black sheep going to stick
5:43 around.
5:45 [Music]
6:05 Um, fair amount going on today. One of
6:08 the big
6:10 things side hustle me a day early for
6:14 Friday night date night. That's a rough
6:15 one. I I'll tell you what, if you think
6:17 it's a Friday, you come in here on a
6:19 Thursday and I'm not even drunk. Ah,
6:21 that's rough. Sorry about that.
6:23 [Laughter]
6:27 It's Friday for Steo. Yeah,
6:30 exactly. It's Friday day. It's Friday
6:33 morning date night for Steo.
6:36 [Music]
6:55 Standing
6:57 between you and a hard place is
7:02 insane. Standing too
7:05 near you in a fire makes it
7:12 clear your trouble to me.
7:15 [Music]
7:18 Real trouble, can't you
7:23 say? Leaning in
7:26 close. Smell of your perfume scares me
7:32 most. Leaning
7:35 away, it feels farther every
7:41 day. You're trouble to me.
7:48 Real trouble. Can't you
7:51 [Music]
7:58 see? I got that dog doing good
8:02 tonight. Oh man.
8:06 [Music]
8:10 Okay, CEK Tacoma. Hope everyone's having
8:14 a nice
8:15 day. Well, thank you for that. Very
8:18 [Music]
8:21 nice. Um, so today, what did we get?
8:24 Claude is now connected to the internet.
8:26 You know, I've been bitching about that
8:28 for a year. Like, what the [ __ ] is up
8:29 with that? They connected it to the
8:32 internet today.
8:35 So, as as one of the major, you know,
8:38 model companies, the fact that they
8:40 weren't connected to the internet was
8:41 just starting to get weird. So that's
8:43 over. So if you go to claude now
8:44 connected to the internet,
8:47 um, OpenAI today had a kind of
8:51 spontaneous tech announcement for
8:54 developers
8:56 um that they've got a new speech to no
8:58 two new speechtoext models that they say
9:01 are better than Whisper. Uh, and Whisper
9:04 was pretty good. It was already
9:04 state-of-the-art. Um, so these are
9:06 better than that and
9:08 cheaper. And uh and then there's a new
9:13 texttospech model. Um and then there's a
9:16 new um there's some new things in the
9:19 API to make it very easy to do voice
9:21 agents. So um I assume I assume sorry
9:27 scratching the eye. I assume this means
9:29 we may see some something akin to a
9:33 custom GPT builder that's an agent
9:35 builder in the next couple of months. I
9:38 would hope. Tik Tok
9:41 pin. Winston game for Champion. I really
9:44 enjoy you also,
9:45 Kyle. Champion is incredible. Surreal to
9:48 me. Helps my soul. You know what,
9:50 Winston? Me, too. I got to tell you,
9:52 there's something about, you know, it's
9:55 when dogs bark or when, you know, babies
9:58 cry or kids scream or whatever it is,
10:00 there's a visceral thing that happens to
10:03 you, right?
10:05 And champ singing is absolutely this
10:08 visceral thing that I don't care what
10:11 kind of mood I'm in. When I hit a chord
10:13 on the guitar and he sings or or you
10:16 know something comes on the TV and he
10:17 sings like I can't help smiling. It's
10:20 like when I'm playing like a lot of
10:22 times I'm laughing. It's just the joy of
10:25 him just sitting over there totally in
10:28 it.
10:28 Just he is just in the
10:33 world. It is pure joy and there's you
10:37 can't escape it. So thank you for
10:39 noticing that. That is that is
10:41 absolutely what he is. Yeah. Feel the
10:43 same. Yeah. It's it is there's it's an
10:46 involuntary response and like when when
10:49 he barks I have an involuntary response
10:51 of like
10:53 It's not good. Uh, but when he sings,
10:56 man, it is just this involuntary
10:58 response of
11:00 joy. Right, Champy? Are you a good dog?
11:04 Are you a good big fat dog? Yeah. Are
11:06 you fat? Are you so fat? Ah, it's a fat
11:10 [Music]
11:16 dog. Oh, yeah.
11:20 [Music]
11:24 in a
11:29 westerly
11:33 direction. This car is my
11:37 train. I've been driving. I've been
11:41 wondering what it is I'm running from
11:44 [Music]
11:48 again.
11:51 Feel like an 80y old man
11:55 holding on to
11:58 29. And up ahead on that
12:03 horizon is a California
12:06 [Music]
12:13 line. Up ahead of trucks carrying a wide
12:16 load. Prefab house cut in
12:21 half. Cute little front door and two
12:23 windows. My
12:25 lung ain't sure whether a cry should
12:29 last. You see, I broke a home of myself
12:33 once. As I stumbled to that
12:37 door, I read a note by the dawn
12:41 light said, "Don't you come around
12:44 here."
12:49 anymore. I've had
12:53 enough. This freedom on the
12:57 rock. Never was good with
13:00 decision. At least that's what I've been
13:03 told.
13:07 [Music]
13:12 Okay, let's get this show on the road.
13:14 Let's get this thing started. Does he
13:16 ever do anything here? Are we ever going
13:17 to learn anything? It's called the AI
13:18 learning lab and all I hear is a dog.
13:20 It's apparently a fat dog. I don't know
13:22 why we're fat shaming shaming animals
13:24 now. I thought I thought we stopped
13:27 doing that years ago. And so So now you
13:30 can't fat shame people. You're going to
13:31 fat shame your pets. Absolutely
13:33 unacceptable. Absolutely
13:35 unacceptable. All right. And that kids
13:38 kids is what is known as a copyright
13:40 strike.
13:45 Oh man. Okay. Um I'm going to show you
13:50 something that is going to be disturbing
13:52 and then I'll explain what it is and
13:54 then I'm going to do a tutorial on how
13:56 we made it.
13:57 So, one of the things that OpenAI
14:00 launched today is this new texttospech
14:04 model and they made it so that you can
14:09 similar to 11 Labs. You can just
14:11 describe a character and it will invent
14:13 a voice character for
14:15 you. There's 10 base models you can
14:18 choose
14:19 from and then you can write prompts and
14:22 they have a website for it. And then
14:25 they also announced in the announcement
14:27 today that for the next 24 hours they're
14:29 going to have a little contest and
14:30 whoever comes up with the most creative
14:32 voice they're going to give a slick
14:34 radio to which I and it's it's pretty
14:36 cool. It's teenage engineering. It's
14:38 really cool looking. Um so what I'm
14:40 going to show you is my entry for for
14:43 that contest and then we'll we'll
14:46 uh we will I'll tell you how I made
14:50 it. So yes, she's disturbing. Oh, wait.
14:53 Let me do this. All
14:59 right, here we go.
15:23 Damn it. Audio isn't working on
15:27 YouTube. Oh,
15:32 [Music]
15:40 you have a name, don't you? A fine one,
15:44 too.
15:46 Strong,
15:48 sharp. I have eaten many
15:51 names. Some sweet, some bitter, some so
15:56 old that even the stars have ceased to
16:00 remember them. I wonder, will you guard
16:04 it tightly? Or will you in a moment of
16:09 trust tell me? Speak it aloud, little
16:13 one. Let me taste it. Just
16:17 once. Just enough to know how it feels.
16:24 [Music]
16:27 So, you have a name. So that voice uh is
16:32 done um in the new uh texttospech model.
16:38 If you go to here, let me change my tab
16:41 sharing so you can see where I'm headed.
16:44 If you go
16:47 to
16:49 openai.fm, that will take you to this
16:52 site where you can build these
16:57 things. Okay.
17:00 And this is it here. It's pretty [ __ ]
17:03 hideous, but it it's functional. So,
17:06 what they've got is they've got like mad
17:08 scientist, calm, dramatic, oldtimey,
17:12 chill surfer. And then for each one,
17:15 they've got this structured
17:17 prompt. So, we'll do chill surfer. And
17:20 then there's 10 voices across the top.
17:23 So, you pick a voice and then down here
17:26 is the description of that voice. Hang
17:29 on a sec. Black bar. Um, and then here's
17:34 the script of what they're going to say.
17:36 And then you play it. And it's pretty
17:39 quick. Oh, wait. I got to
17:42 change to change my my audio
17:45 thing just so you can hear it. I think
17:48 it's worth hearing the examples.
17:51 Um, so this this voice is Aloy or Alloy.
17:59 Whoa, dude. Sounds like a gnarly
18:01 situation with your card. But hey, no
18:04 worries. I got you. All right. And then
18:08 I can choose a different voice, but it's
18:11 still going to be a chill surfer.
18:14 Whoa, dude. Sounds like a gnarly
18:17 situation with your card.
18:23 Whoa, dude. Sounds like a gnarly
18:25 situation with your card.
18:29 Or I can change the personality of that
18:32 same voice. So, you've got the base
18:34 voice and then the personality on top of
18:36 it. So, here's that same voice with an
18:38 olden
18:40 timey description.
18:44 Ah, ladies and gentlemen, a most warm
18:47 welcome to ye grand internet service
18:49 company. And so like the description
18:51 here, tone, the voice should be refined,
18:53 formal, and slightly theatrical,
18:55 reminiscent of a charming radio
18:57 announcer from the early 20th century.
18:59 Pacing, the speech should flow smoothly
19:04 um at a steady cadence, neither rushed
19:07 nor sluggish.
19:09 um pronunciation words should be
19:11 enunciated crisply and elegantly right
19:14 so and then it's got emotion inflection
19:16 and word choice so you can basically
19:18 just tell this thing how you want it to
19:20 behave and you can create new ones so
19:24 what I did again back to my commitment
19:27 of not using my brain in 2025
19:31 um let's have chat GPT help us write
19:34 some of these things and so I did Um,
19:38 let me close this. Let me open
19:44 this tabs. Thank you.
19:52 [Music]
20:01 [Music]
20:08 creating voice prompts.
20:11 Okay. Is this
20:13 it? Yeah. Okay. So, here's So, my
20:18 um that character that I created is
20:21 called the eater of
20:23 names. And
20:25 then here's the
20:31 um here's the structured prompt that
20:33 matches the prompt that they have over
20:35 there. So So how I got to
20:39 this I'm pretty happy with and I'll I'll
20:42 share it with you because it it took me
20:44 a couple of attempts. So the first thing
20:46 I
20:47 did was I just copied over
20:52 um one of the prompts. What did I say? I
20:54 said, "I'm using a new app that lets me
20:56 control the voice personality of text it
20:59 creates. Here's an example prompt." And
21:01 I showed that. And then I said, "Uh, can
21:05 you write a users's manual on how a
21:07 creative would write a prompt for a new
21:09 original voice?" So rather than just
21:11 say, "Copy the prompt," I said, "Write
21:15 me a user manual." Because I tried it
21:17 just copying the prompt and it got all
21:19 confused. So, I said,"I want you to
21:22 write me an instruction manual for how
21:24 you would tell someone how to write one
21:26 of these
21:26 prompts." And so, it did that and it
21:29 came up with some stuff. And then I I
21:32 said I said, you know, make the
21:34 descriptions longer, things like
21:37 that.
21:41 Um, and then I said, "Okay, invent a new
21:44 character using it." and it didn't quite
21:46 get it right. So, I kept changing
21:50 it until I got
21:53 it. I kept making small little detail
21:56 changements for the output it gave until
21:58 I was happy with it. And then I said
22:00 rewrite the user's manual. And so then
22:02 it rewrite the rewrote the user's
22:04 manual. Right? So identity who's
22:06 speaking provide only the name and the
22:08 title that encapsulates the character's
22:10 essence. So identity the clock maker the
22:13 forbidden forbidden forgotten king and
22:16 then so it gives instructions for each
22:18 of the structured little pieces. Then it
22:21 gives an
22:22 example. Um and then I said make the
22:24 examples longer and more detailed. And
22:26 so it did that and then that I was
22:28 pretty happy with. And so now it's a
22:31 pretty good manual. And now I just said
22:33 you know
22:35 um write me a prompt. Perfect. Now, now
22:38 create an original character that is
22:41 unforgettable and
22:43 incredible. And so it came up with the
22:46 Eater of Names, which I think is just an
22:49 absolutely disturbing character
22:52 name. The Eater of Names. It's like
22:56 [ __ ] scary and brilliant, right?
23:01 Um
23:04 and and then so a effect whispered and
23:07 insidious yet strangely inviting like a
23:11 secret only they can tell. Their words
23:13 slide like silk but carry the weight of
23:16 something ancient and hungry. Tone
23:19 darkly poetic with the cadence of a
23:21 bedtime story told too late at night
23:23 when the shadows stretch a little too
23:25 long. emotion, a mix of quiet amusement
23:28 and predatory curiosity, as if savoring
23:31 the taste of every word spoken to them.
23:34 Pauses, lingering, drawn out silences
23:37 that make the listener lean in only for
23:40 the next word to land like a drop stone.
23:42 Pronunciation smooth and deliberate as
23:46 though savoring each syllable except
23:48 with when saying a name, which is
23:50 devoured quickly, almost as if they were
23:53 stealing it away. And then I had it
23:55 write a script for each one. So, so
23:58 we've now got this thing that actually
24:01 maybe what I'll do here. Why don't I do
24:03 that? I'll I'll take that instruction
24:05 manual. We'll create a custom GPT. I'll
24:08 publish it and then I'll just give it to
24:10 you and you all can go play. But don't
24:13 if you win the [ __ ] radio because you
24:16 used my custom GPT, send me the
24:19 radio, [ __ ] because I want the
24:22 radio.
24:25 It's got an OpenAI logo on
24:29 it. But I'll take a bottle of Bllandons.
24:32 Um, that's that's an
24:35 alternative. Um, and it and it wrote the
24:38 script. Ah, you have a name, don't you?
24:40 A fine one, too. Strong, sharp. I can
24:42 hear I can hear it beneath your breath,
24:44 tucked in behind your teeth. But names
24:47 are such fragile things, aren't they? a
24:49 slip of the tongue, a whisper in the
24:51 wrong ear, and suddenly gone, forgotten.
24:55 Um, so just just really cool. I I have
24:57 eaten many names, right? So cool little
25:00 script.
25:02 So let's
25:05 grab our
25:11 instructions, user manual.
25:16 I don't think we need
25:18 that. We just need
25:22 this Tik Tok pin.
25:25 Kyle, save your time and effort. Just
25:27 tell it to sound like Mr. Rogers from
25:30 Mr.
25:31 Rogers. Yeah, exactly. Well, that too.
25:35 But I just thought I I just thought this
25:36 was fun. I thought it was a cool
25:37 creative writing exercise
25:41 um to get it to write really creative uh
25:44 versions of those things. All right, so
25:47 we're going to go we're going to go
25:50 new. We're going to go custom GPT
25:53 explore GPTs. If you have if you've
25:55 never made a custom GPT, it's relatively
26:00 straightforward. I'm going to go into
26:02 configure. We're going to call this
26:04 the openai
26:11 umfm. I don't know if I can put a URL in
26:13 the name. Um
26:18 uh let's
26:19 see. Voice
26:23 designer and
26:27 prompt
26:29 writer. Okay. So, uh description uh
26:36 creates
26:40 structured prompts
26:43 and
26:45 sample
26:48 scripts and
26:53 original
26:56 characters for use
26:59 in open
27:03 AI's TTS
27:06 playground. Uh uh uh uh uh uh uh uh uh.
27:10 Okay. What does this D GPT do? So, we're
27:13 going to say
27:18 um follow these
27:25 instructions. And then I'm going to
27:27 paste in all that [ __ ] we put in there.
27:33 And then I'm going to go
27:41 um if the user wait
27:44 if the user
27:49 says get started.
27:56 Generate three
27:59 unique,
28:01 original,
28:07 unforgettable,
28:12 remarkable
28:14 characters based on the
28:21 instructions. And then conversation
28:24 starters. I'm going to write get
28:25 started. So we'll only have one
28:27 conversation
28:34 starter. And we should say make me an
28:37 image.
28:41 [Music]
28:47 Use this all making
28:50 sense. Y'all following along? All right.
28:54 So, let's say get
28:56 started. Let's see if it just yanks out
28:59 three
29:03 characters. Identity. The astral
29:05 cartgrapher. Affect curious and
29:07 reverent. As if in awe. Yeah, it works.
29:10 Okay,
29:16 perfect. This is awesome. All right,
29:18 stop.
29:20 Create. We're going to put this in the
29:22 GPT
29:23 store. It's called the
29:25 OpenAI.fm Voice
29:27 Designer. Where are we going to put
29:29 this? Programming. Yeah, this is
29:31 programming. Damn it. We're [ __ ] vibe
29:34 coding here, people. You understand
29:36 what's happening? Save.
29:41 [Music]
29:46 unable to publish
29:49 GPT. It violates usage policies. May
29:53 contain content that requires previous
29:55 authorization. I told you I couldn't put
29:57 that URL in there. So, I'm going to put
30:00 open
30:01 AI brackets do bracket
30:05 FM. Let's see if Let's see if we can get
30:09 around it like that. Boom.
30:12 GPT store. Oh, wait. It was the wrong
30:15 category. Damn it.
30:19 Programming unable to
30:23 publish. Oh, brand names. Oh, open AI.
30:26 That's what it is. Okay, let's see.
30:30 um voice designer and prompt
30:34 writer
30:36 for
30:38 open
30:41 AI
30:43 TTS. Let's see if I can do it by
30:46 splitting the
30:50 words
30:52 create. Boom.
30:55 programming
31:01 save. God damn it.
31:27 Do I have it somewhere else? Hang on.
31:30 Let me see if I I don't think I have it
31:32 in
31:37 here.
31:42 No, must just be in the title. All
31:44 right.
31:45 Create GPT
31:48 store programming.
31:57 Save published. View GPT. Okay. So now
32:01 I'm going to save this. Brandon, throw
32:02 up the
32:05 uh Source
32:07 Camp. How late are
32:09 you? We Champy sung like six
32:28 songs. And yeah, pop up.
32:31 Uh, we started and we started an hour
32:34 late. Yes, producer Brandon is even
32:37 riding you on this one.
32:44 Um, okay. So, we're going to share this
32:46 in the AI salon. So, if you go
32:49 to the
32:51 salon.ai, where am I going to put this?
32:53 This is a GPT. Oh, no. We'll do it in
33:02 irregulars. Okay. So, here we'll go. Uh,
33:05 open
33:07 AI text to
33:10 speech. I guess I should put
33:13 text to speech because TTS doesn't make
33:17 sense to people that don't know what the
33:19 [ __ ] that means. Open AAI text to speech
33:24 prompt
33:26 writing
33:29 GPT. Um
33:35 uh prompts and scripts for
33:40 open no prompts and script
33:46 generator and I'm going to go open
33:49 aai.fm
33:56 FM and then we'll paste
34:07 it. I could put in the the scary the
34:11 scary chick. Yeah, it's good. We'll just
34:14 post it. Don't notify everyone. All
34:16 right. So if you go to the irregulars
34:18 channel in the AI salon mighty
34:21 networks, you can go play with the GPT
34:24 we just
34:25 created. Even source camp source camp,
34:29 you can go play
34:32 too. So you just click get started and
34:35 it'll generate three voices for you.
34:38 Boom. Identity the lantern bearer.
34:44 Nice. and link is in the YouTube channel
34:46 as well. There you go.
34:50 Beautiful. Identity. The choir of
34:53 dust. Layered and echoing as if many
34:56 voices speak in the haunting memory,
34:58 never quite in unison. I love that.
35:00 Let's go try this
35:01 one. Oh, but it didn't write. Oh, which
35:05 one of these do you want to turn into a
35:08 full script or sample monologue? Number
35:10 two.
35:14 Excellent
35:18 choice. Oh, this is good. Now it's doing
35:20 a more
35:21 detailed prompt and it's g like like
35:24 like chat GPT lately. It's giving us
35:28 lots of
35:33 emojis tells me not found in red. Am I
35:36 late, Mr. It? Sample
35:39 monologue. This place was not always
35:42 quiet. There were songs here once. Heavy
35:44 with joy, swollen with grief. I don't
35:46 know why it's creating such such scary
35:49 things. Were the other ones scary? Choir
35:52 of dust, the gilded
35:55 parasite, the lantern bearer. I don't
35:57 know. Well, whatever. I told it to make
35:59 them memorable, didn't
36:00 I? Okay. Identity, the choir of dust.
36:06 Let's copy all
36:09 this and we'll go over
36:12 to I guess I'll put it in music and
36:21 audio. Open
36:28 aai.fm. All right. So, there's that.
36:33 And then let me go back here and get the
36:38 monologue. Copy that. Go back to
36:45 here. Paste
36:48 it. And then what's cool is so now I can
36:52 pick different voices to to try this
36:55 personality out out on. Well done,
36:58 squirrel Tik Tok pin. Um, how do you
37:00 make money with AI? Uh, I'll get back to
37:03 you on that one. Uh, Charry. Oh,
37:05 Chardonnay. Char, wait, what is it? Oh,
37:07 Chatty
37:08 Chad. Nice name, Chatty Chad. Uh, I'll
37:11 get back to you on how do you make money
37:13 with with uh ChatGBT in a second here.
37:16 You can make money with Chad GBT. That's
37:18 a question that get asked gets asked so
37:21 much that we have a button for it. Okay,
37:25 I'm gonna have Alloy. I'm going to try
37:27 the 10 voices with this personality.
37:30 We'll see how they are. Oh, wait. I
37:31 gotta share my tab, right? Hang
37:35 [Music]
37:40 on. Ta.
37:42 [Music]
37:49 This place was not always quiet. No,
37:52 that's boring.
37:55 This place was not always quiet. Um, I'm
37:59 going to lose all this [ __ ] I think the
38:02 identity is too much. Oh, and you know
38:03 what I'm going to lose? I'm going to
38:04 lose the the emojis. I bet they're
38:06 [ __ ] it
38:08 up.
38:11 So, like all stuff with AI right now,
38:14 everything's janky. Assume nothing will
38:18 work because in most cases, things don't
38:24 work. It's not you. You're not
38:27 stupid. Don't apologize. Don't feel
38:31 shame. Blame it on the AI. You can blame
38:35 it on Pate.
38:36 Pate works at Google. He he accelerates
38:41 uh tensor processing units. He makes
38:44 them more
38:45 efficient.
38:47 Apparently, them being less efficient is
38:50 why we're getting crappy results. So,
38:53 you can absolutely blame him. All right,
38:55 let's see if this is any better.
38:59 This place was not always quiet. Yeah,
39:02 it's a little There were songs here
39:04 once, heavy with joy, swollen with
39:07 grief. We remember not What did P
39:12 say? I just got here and I don't
39:14 appreciate this
39:18 slander. Well, well, we also used Open
39:21 AI to to write the prompts. So, so
39:24 they're not even using Google hardware.
39:26 So, you you are you are absolved of all
39:29 responsibility for the crappy prompt
39:31 writing.
39:34 This place was not always quiet. That's
39:38 a shitty voice. I don't like it.
39:41 This place was not always quiet. There
39:48 this place was not always quiet. All
39:51 right. It sucks. Okay. So, we're going
39:53 to go back to chat jpe tag and let's say
39:56 um make me a really funny voice and
40:03 write
40:04 the
40:06 monologue for it.
40:12 Oh, and we got to say,
40:15 um, no more
40:22 emojis. Identity. The mirror
40:26 gremlin. Okay. Effect, frenetic, and
40:29 theatrical. Okay, this is good. The
40:31 mirror gremlin. Okay. Uh, let's see.
40:41 Copy. Let's go back over
40:44 here. Throw it in there.
40:47 Paste. Let's go back over here. Sample
40:51 monologue. Ah, there you are. Finally.
40:54 I've been waiting for you in the mirror.
40:56 I'll be watching because I literally
40:58 can't leave. This is not funny. This
41:01 sounds scary.
41:06 All right, back to Alloy. Let's see if
41:08 this one's funny.
41:11 Ah, there you are. Finally. I've been
41:13 waiting in this mirror for three whole
41:15 hours. Yeah.
41:19 Ah, there you are. Oh, wait. Let me get
41:22 rid of all this
41:28 description. Ah, there you are. Finally.
41:31 I've been waiting in this mirror for
41:33 three whole hours. All right. I've been
41:36 waiting on this sticky note for three
41:38 whole hours. What's going on?
41:41 What about
41:43 it?
41:49 Oh. Oh, just that I need to get back to
41:51 that Tik Tok pin. I will for Chatty
41:53 Chad. I'll get back to Chatty
41:55 Chad. Where? No, there's there's another
41:58 one from Side Hustle.
42:01 that's been pinned for like personalized
42:03 wedding ceremonies and charge clients
42:05 250 bucks. There you go. Okay, so let me
42:08 chatty chat. Let me give you the answer.
42:10 So here's the here's the you can make
42:12 money with why this button exists is
42:16 when chatpt first came out Tik Tok was
42:19 littered with idiots that were like you
42:22 can make money with chat and I don't
42:24 they always said it like that and I
42:26 don't know why. I think there's
42:28 something in the algorithm that someone
42:31 has figured out that you have to talk
42:32 like an idiot for people to pay
42:34 attention to
42:36 you with her. And honest to God, do
42:40 people listen to these people? Anyway,
42:42 that's neither here nor there. Here's
42:45 the problem with that statement is that
42:50 CHBT is simply a means to an end. It is
42:53 simply a tool. It is it is no different
42:57 making money with chatbt than it is
42:59 making money with Photoshop or Microsoft
43:02 Word or a shovel. You can make money
43:05 with a
43:06 shovel. Well, how? Well, you find people
43:11 that need holes and you tell them you
43:14 can make them a hole and then you say,
43:16 "I will charge you $5 for every hole I
43:18 make." Okay, you can make money with a
43:21 shovel. So similarly with chat GPT it is
43:25 just a tool. Now it happens to be a tool
43:27 that's incredibly
43:29 powerful but you chatty Chad um likely
43:34 have some expertise in some area of your
43:38 life. It could be in baseball cards. It
43:40 could be in I don't know mowing lawns.
43:43 It could be in um um writing clever
43:47 poems on the street with a typewriter.
43:50 You know, like everyone's got some
43:52 things that they've got some expertise
43:54 in. So anywhere where you've got some
43:56 expertise, you can imagine, huh, I
43:59 wonder who's got problems that if I
44:02 could solve them better, faster, or
44:05 cheaper than how they currently get
44:07 those problems solved, maybe I could do
44:10 something
44:11 here, right?
44:14 And so that could be
44:16 like, hey, there's probably a bunch of
44:18 restaurants out there with boring
44:21 menus and I could probably go in and
44:24 just say, hey, in half an hour, I'll
44:27 take your menu and I'll deboringize it.
44:30 And you walk in with your chat GPT app
44:33 and you take a photograph of their menu.
44:37 You take a snap of their menu and then
44:39 you say, "Analyze this menu and make it
44:42 less boring.
44:44 And then you copy and paste that into a
44:47 word document and you print it out and
44:49 you come back the next day and you
44:50 charge them $200 for a new menu. Like
44:54 solve the problem is how you make the
44:56 money. The tool is kind of
44:58 irrelevant. It just happens to be a
45:00 really amazing tool. But but don't think
45:02 like that. Don't think how can I make
45:04 money with chat GPT because you'll just
45:06 you you get in this [ __ ] endless loop
45:08 of frustration. Um, Pete M. They say,
45:12 "No, you asked me to build too many
45:15 holes. Can you ask me again
45:18 tomorrow?" Oh, then you say, "You ask me
45:21 to build too many holes. Ask me again
45:23 tomorrow." Yeah. Yeah. You can also use
45:24 the shovel to fill the holes back in if
45:26 you made too many. If you're not careful
45:29 careful, you're a you can AI a little
45:31 too hard and get job offers, too. Oh,
45:33 that's the other thing. the especially
45:37 if you use chat GPT to solve a
45:41 problem for people that don't know what
45:45 chat GPT is or what it makes possible,
45:48 you will you will occur to them like a
45:51 [ __ ]
45:52 magician. And as Pate tells us, this is
45:55 not magic. It's math. And and we have
45:58 come to a compromise. It's mathematical.
46:01 But if you're dealing with chat GPT at
46:04 any level of competence and you solve a
46:07 problem for someone who's never seen it
46:09 before, never experienced it, it will
46:12 blow their [ __ ] mind. And and side
46:15 hustle Mimi is right. You just might end
46:16 up with a a a job offer just because you
46:20 came in and solved a problem. What I
46:21 would also encourage you to do is, you
46:25 know, find a charity or a community
46:27 group or friends of yours and just
46:30 experiment. say, "Hey, I I have this
46:32 idea for being able to whatever help,
46:36 you know, people do marketing copy or
46:38 marketing strategy much more quickly
46:40 than they did before. Can I do it for
46:42 free for
46:43 you?" And and maybe you'll give me a
46:46 testimonial, right? But again, start
46:48 with the problem. Start with things you
46:51 know, challenges that people in that
46:53 field deal with, and then can you use
46:56 chat GPD to solve it faster, better, or
46:59 cheaper?
47:05 He finally revealed how to make chat
47:07 GPT. I've been waiting all these years.
47:10 Yeah. Well, the the way you make money
47:11 with chat GBT, it's got nothing to do
47:13 with chat GPT. That's the that's the
47:15 irony. The irony in
47:19 it. Oh, man. All
47:22 right. Um.
47:26 Ah, there you are. Finally.
47:32 Ah, there you are. Finally. I've been
47:35 waiting in this mirror for three whole
47:37 hours. That's not bad.
47:40 Ah, there you are. Finally. No.
47:46 Ah, there you are. Finally. I've been
47:48 wait I didn't cut my hair. Oh, no. I
47:51 just It's just I don't have enough
47:54 product in it, apparently.
48:00 Ah, there you are.
48:04 Ah, there you are. O, awful.
48:08 Ah, there you are. No.
48:13 Ah, there you are. Was it Fable?
48:17 Ah, that. No, it was Echo.
48:21 Ah, there you are. Finally. I've been
48:23 waiting in this mirror for Who was it?
48:26 Alex. Ah, there you are. Finally. No
48:29 quarrel.
48:32 Ah, there you are.
48:37 Ah, there you are. Finally. I've been
48:39 waiting in this mirror for three whole
48:41 hours. And let me tell you, watching you
48:43 scroll aimlessly through your Yeah, I
48:45 guess that one's okay. All right. Well,
48:46 so there you go. There you got it. You
48:49 can style your hair with Chad
48:52 GBT. It couldn't possibly look worse.
48:59 Oh man. All right.
49:03 Fantastic.
49:05 Um, hey me. It's a huge
49:10 [Music]
49:14 Okay. After over two years, I missed
49:17 it. You're getting bold. You mean bald?
49:24 Uh uh uh
49:26 uh I got the IRS to concede to my to
49:29 concede my case. They were behaving bad
49:32 and wrong. Chat
49:37 GPT. Sorry about that law degree. Yeah.
49:40 I mean,
49:45 listen, these tools are getting better
49:47 and better and better and better and
49:50 better and better and better and you're
49:53 going to be able to do things like, you
49:56 know, um, deal with the IRS or deal with
50:00 lawyers or understand
50:02 contracts. I mean, you can take very
50:04 complicated contracts. I'll tell you a
50:06 fun thing to do if you're new to Chat
50:09 GPT. You know how when when you're
50:11 surfing the internet and you sign up for
50:13 something new and they're like, "Here's
50:15 our terms and conditions." And you
50:17 always just check the box and hit
50:19 submit. Have you ever wondered what
50:21 you're agreeing to, but you're like,
50:23 "I'm not going to read through the 47
50:25 pages of that stupid [ __ ] thing."
50:29 copy and paste the terms and conditions
50:31 from some piece of software you just
50:33 signed up for and paste it into chat GBT
50:36 and said, "Explain this to me. What What
50:39 am I What rights am I signing away
50:41 here?" It's pretty It's pretty
50:44 wild. Pretty
50:46 wild. All right. Yeah, to be fair, you
50:49 have to work with it to how to how to
50:50 get it to prompt better. Yep. It will
50:52 not it will not without work. Yeah,
50:54 exactly. All these
50:56 things, all of these tools right now
50:58 require us to be the human in the loop
51:02 to make them not fully suck. Um,
51:06 probably two years, three years. Um,
51:08 they're going to be they're going to be
51:10 fairly impeccable
51:12 um, beyond our capabilities. I would
51:15 argue right now that most of these tools
51:17 are significantly beyond my
51:18 capabilities. like significantly. You're
51:21 like, "Well, you know, yeah, I I mean,
51:24 you seem to be wholly
51:29 [Laughter]
51:33 unqualified." Oh man.
51:37 Um, what else was I going to show? Oh,
51:39 the other thing that I did here, let me
51:42 show you one other
51:44 [Music]
51:47 thing cuz I thought this was kind of
51:49 fun. Th this is this starts to get into
51:52 the neighborhood of have, you know, just
51:53 play with chat GBT.
51:56 So,
51:59 my Oh, that's the voice designer. Wait,
52:03 let me go to
52:14 That's not what I
52:19 [Music]
52:23 wanted. Yeah. Okay. So, it chat GPT came
52:29 up with this character, the eater of
52:31 names, right? Which I just loved. So, I
52:34 thought, well, I wonder what the eater
52:36 of names looks like. And then it, you
52:39 know, it's got the description of the
52:40 voice
52:42 here. And so what I did was I I copied
52:46 the description of the voice and I went
52:49 to
52:53 midjourney. Oh, look at me. That's
52:55 that's the French lady
52:58 me.
53:01 Um, and so I pasted in
53:07 I pasted in the prompt, the full prompt
53:10 of the voice description into
53:13 Midjourney. And then the other thing I
53:15 did
53:18 um where do I have
53:21 it is I went to my
53:24 mind where I store my
53:27 um srefs that I think are cool. And so I
53:32 found this sref that I thought was just
53:34 kind of kind of cool. So I copied that
53:38 sref. If you don't know what an SREF is,
53:40 it's a style reference. So on Twitter,
53:43 there's I don't know if you've seen
53:44 these things, but people put out these
53:47 style references. You can just copy and
53:48 paste the tweet. I store them in this
53:50 thing called my mind. I mean, I've
53:52 literally got hundreds of
53:55 these and they're just these are just
53:57 ones that I like that I'm like, "Oh,
53:58 that might be fun to go back and use
54:00 that sometime." So, I just took one of
54:01 the recent ones that I liked, this
54:04 one, and then I went
54:07 into
54:09 midjourney and I put in that style
54:15 reference. And
54:23 oops. And these are all the things that
54:25 came
54:26 out. Like I almost went with that one.
54:29 That one was pretty cool. But like they
54:31 they all had this kind of just [ __ ]
54:35 crazy creepy
54:37 ass
54:39 thing. French lady Kyle is too
54:42 unbothered. Silly Kyle. The hair I was
54:44 referring to was your post. Oh, you had
54:47 long hair. Now you don't. Oh yeah, that
54:50 one. I thought you meant
54:53 my Oh, you know what are you gonna do?
54:55 Um, that's beautiful. That's beautiful.
54:58 Aren't these beautiful? They're just
54:59 [ __ ]
55:00 gorgeous. Like any one of these could
55:03 have been something to do something
55:04 with. That one's
55:08 gorgeous. Gorgeous.
55:16 Stunning. And then that one I almost
55:18 went
55:19 with. And then in the same That one I
55:23 almost went with. That one I almost went
55:25 with. That's the one I did go with.
55:28 Then I did a couple of more. Then I did
55:31 some variations of
55:35 her. Like that one I almost went with. I
55:38 kind of like the
55:48 red. So then I took the audio recording
55:52 that she said. I took this image, put
55:55 that into Hedra that with the character
55:58 new character animator thing, and then I
56:01 went to Suno and I wrote the creepy
56:02 music, and then I stuck it all together.
56:05 They're beautiful, aren't they, Lori?
56:06 Yeah, they're just gorgeous. And it's
56:09 like this is, you know, when we talk
56:11 about vibe coding and vibe storytelling
56:14 and this is, you know, vibe creative or
56:16 vibe artistic
56:18 directing. The prompt here was a prompt
56:22 for a voice.
56:24 But I was like, I wonder, you know,
56:26 Midjourney is good at interpreting
56:29 [ __ ] And, you know, just combined that
56:31 with a style reference I thought was
56:33 cool. And it just made this really
56:35 unique aesthetic out of
56:38 nowhere. Like instant
56:43 bang. Look at
56:47 that. That's midjourney.
56:50 Yep. I just did it, Kyle. I added the
56:53 prompt to open a FM. Now what? Um, okay.
56:57 So, if you're in OpenAI
57:02 FM in the bottom left, yeah, this is
57:05 this is this is such shitty interface.
57:07 Okay, so you put the you put the prompt
57:10 in the bottom left, right? The one that
57:12 says emotion, pauses, things like that.
57:15 You put the script in the script box
57:17 here, and then you select a voice up
57:20 top. So, so the way it works is think as
57:23 the top voices are your base voice and
57:26 then the lower left hand prompt is the
57:29 personality of that voice and then the
57:31 script is the script. So you pick you
57:34 know the ballad character and then you
57:36 just hit play.
57:51 [Music]
57:56 Yes, I did all that. How do you send it
58:00 to Open AI? Oh, for radio. Oh, you click
58:04 the share
58:05 button right
58:07 here and then copy it and then go to
58:15 Twitter socials. Twitter
58:20 And then you go search for
58:25 um OpenAI developers is the
58:29 account. And if I go there, I assume
58:32 it's pinned or if it's not
58:37 pinned, new state-of-the-art model.
58:40 Yeah. So, it's about the the fifth post
58:43 down is the
58:46 contest. We're also holding a radio
58:49 contest and so they say respond to
58:51 this respond to this
58:56 tweet with the URL. Now, most people put
58:59 in the URLs. I put in the creepy
59:03 lady, but you can just put in the URL.
59:06 And I might get disqualified because I
59:08 put in I put in the creepy lady, but
59:10 [ __ ]
59:12 it. Oh, yeah. Here's someone that did a
59:19 video. Yeah. So, I did creepy
59:24 lady. All
59:27 right. Make
59:29 sense. In a
59:31 nutshell, what were those four guys
59:33 showing us this afternoon? Oh, in a
59:35 nutshell. Okay. So, this most
59:40 awkward presentation of these awkward
59:43 dudes that honest to God, it was like
59:46 every one of these is like an SNL skit
59:48 that no one
59:55 would. That guy, that guy you could
59:57 understand, the other three you
59:58 couldn't. Okay, so here's what they
59:59 have.
1:00:03 There are probably more components than
1:00:05 this, but there there were
1:00:08 basically two
1:00:10 models that they announced today.
1:00:14 GPT40 transcribe and GPT40 Mini
1:00:18 Transanscribe that you talk into it and
1:00:21 it turns it into text. So, just like
1:00:23 Whisper, but they're new models. They're
1:00:25 cheaper and they're more accurate than
1:00:27 Whisper was. So, they have a new
1:00:29 voicetoext model.
1:00:32 Then the thing that we're playing with
1:00:33 on OpenA FM is a texttovoice model,
1:00:37 right? So you take the text and you take
1:00:40 the personality prompt and the base the
1:00:43 base voices and you can generate a
1:00:46 thing. And then the final thing that
1:00:49 they announced
1:00:51 um was some calls for the API. So if
1:00:56 you're coding, if you're writing your
1:00:57 own
1:00:58 agent, rather than having to say take,
1:01:02 you know, when when voice comes into the
1:01:04 model, turn it into text and now I'm
1:01:06 going to take that text and I'm going to
1:01:08 send that to OpenAI and then it's going
1:01:09 to give me an answer and then I'm going
1:01:11 to send that to a texttovoice
1:01:15 model. They've basically just made that
1:01:17 a single call. So as a programmer, you
1:01:19 just put in a single line and it does
1:01:21 those those three stages. Um, so it's
1:01:24 basically a way for programmers to
1:01:27 design voices and then be able to create
1:01:30 voice agents where we'll just be able to
1:01:33 talk to applications that go off and do
1:01:35 [ __ ] for us. So that's that's in a
1:01:37 nutshell it was programmers can use
1:01:40 voices easier when they're developing
1:01:42 autonomous
1:01:44 agents. All
1:01:47 right, there's a text limit in the
1:01:49 little box on the left. Probably. Yeah,
1:01:53 probably
1:01:55 probably the blue man crew without the
1:01:58 makeup. Yeah, it was it's the these
1:02:01 these are painfully awkward awkward
1:02:04 presentations, but you know um they're
1:02:06 changing the world, so give them a
1:02:09 break.
1:02:10 Um I don't give them a break, but you
1:02:16 should. Oh, man. All right. So there you
1:02:21 go. That's stuff that's stuff in
1:02:24 nonsense.
1:02:30 Um I had
1:02:33 um I had manis
1:02:35 [Laughter]
1:02:37 today. I had manis today.
1:02:40 Um fill out my um my uh March Madness
1:02:46 bracket thing at work.
1:02:49 I said, "Go research March Madness and
1:02:52 just give me a prediction for, you know,
1:02:54 for all the levels." And it did. And
1:02:57 everything looked good until the at the
1:02:59 very end it got two of the games
1:03:00 switched up and it didn't quite do it
1:03:02 right. But basically, it has Duke taking
1:03:05 everything, which Brandon, you got that
1:03:07 same response, too. So, it looks like
1:03:10 looks looks like Duke is
1:03:13 favored. Um, okay. This Sunday at 400 pm
1:03:18 central time or 300 p.m. Mountain my
1:03:22 time,
1:03:23 um there's a learn out loud in the AI
1:03:26 salon. So if you go to the AI salon and
1:03:28 go to the events calendar at the top
1:03:30 left, um this Sunday is a learn out
1:03:33 loud. And actually, Pate, are you doing
1:03:36 a are you doing a uh demystifying
1:03:39 AI? Do you teach people about AI? Teach.
1:03:43 Teach is a very generous word in that
1:03:47 sentence. Yeah. So, the whole purpose of
1:03:51 this channel, I go live five nights a
1:03:53 week. Hang on, I got to do some physical
1:03:55 dexterity challenges here. Um, I go five
1:03:59 nights a I go live five nights a week
1:04:01 just talking about AI.
1:04:04 Um, one of the things that we call this
1:04:06 channel is chat add. Um, so if you're
1:04:09 expecting an agenda and a curriculum,
1:04:11 this ain't the place. Um, but if you
1:04:13 want a place that will allow you to just
1:04:16 kind of be in the AI conversation and
1:04:18 learn things along the way, that's what
1:04:20 it's about. Um, you can also join the AI
1:04:22 salon. So, if you go to the salon
1:04:26 um.ai, you can join there. Um, I do do
1:04:30 one on one um coaching sessions. It's in
1:04:33 my it's in my Tik Tok bio if you just
1:04:36 click right under the bio. Um, you know,
1:04:39 for for a fee per hour. Um, so you can
1:04:42 book time there. Uh, if you want to do
1:04:44 that, if you want to do one-on-one
1:04:45 stuff, I'm happy to happy to talk about
1:04:47 that. But, you know, I've been giving it
1:04:49 away for free here for I don't know
1:04:51 what, two years
1:04:52 now, something like that.
1:04:56 So,
1:04:58 yeah. Yeah. Yeah.
1:05:01 [Music]
1:05:05 Yeah. And Murphy, what's happening? An
1:05:08 Murphy, what's shaking? What's going
1:05:10 down?
1:05:11 My voice is getting
1:05:18 crispy. Two and a half years. Yeah.
1:05:20 Well, the the Tik Tok channel's been
1:05:22 around for two and a half years. I've
1:05:23 started I started the lives I think
1:05:26 April April 30th, two years ago was the
1:05:30 first the first
1:05:32 live.
1:05:33 [Music]
1:05:35 Um, yes. See out all Sunday for the
1:05:38 learn out loud. Yep. That's going to be
1:05:40 a really good
1:05:42 one. Kyle, whatever your rates, it's a
1:05:45 steal. Oh,
1:05:46 Ann, you're the best. Thank you. Very
1:05:50 generously giving it away. Thank you.
1:05:53 Very generously or very foolishly.
1:05:59 No, but listen, the
1:06:01 the to be quite serious for a moment,
1:06:04 the the reason I do this
1:06:09 is I find it
1:06:11 terrifying. Like
1:06:14 terrifying how many people are not
1:06:17 prepared for what's coming. Like not
1:06:18 only not prepared, but like
1:06:21 actively putting their fingers in their
1:06:23 ears going, "La, I don't want to hear
1:06:25 about AI. AI's evil.
1:06:29 Yeah, this [ __ ] is coming fast and hard.
1:06:34 And what I've learned over doing this
1:06:37 for the past couple of years is that
1:06:40 even the most bitter, hardened,
1:06:45 angry people that are like AI
1:06:49 haters, when they actually experience
1:06:52 it, it's like the [ __ ] skies open up
1:06:54 and they're like, "Oh my god, I didn't
1:06:57 realize I could do that. I didn't
1:06:59 realize I could be that powerful or that
1:07:02 like people like rediscover things about
1:07:06 themselves. It is truly remarkable. And
1:07:11 and there's just so many people with
1:07:12 their arms [ __ ] crossed like they
1:07:15 make that face like like they just stuck
1:07:18 stepped in some dog
1:07:19 [ __ ]
1:07:23 AI. Have you tried it? Ah, it's the
1:07:25 world's greatest plagiarism machine.
1:07:28 Okay. Have you tried it? It
1:07:31 hallucinates. Okay. Have you tried
1:07:34 it? You want to see something
1:07:38 cool? And you show them. They're like,
1:07:40 "Wait, what? It can do that? I didn't
1:07:44 know it could do
1:07:45 that." They have their Kevin Mallister
1:07:50 moment. That's a pretty magical moment.
1:07:53 So that's why I do this because, you
1:07:55 know, I want as many people as
1:08:01 possible, learning this
1:08:03 [ __ ] you
1:08:05 know, it's been so much fun over the
1:08:09 years. But Kyle, you have changed
1:08:11 countless lives. Seriously. Ah, thank
1:08:13 you, Source Camp. I appreciate that. I
1:08:16 showed chat to a 70-year-old supply
1:08:18 chain manager. He started his journey
1:08:20 today. That's the other thing. Age
1:08:22 doesn't [ __ ] matter. Technical
1:08:24 background doesn't
1:08:26 matter. You know what matters? You have
1:08:29 an idea and you have an intention to get
1:08:31 it the [ __ ] out of your head. That's
1:08:34 it. Oh, I once had an idea for an app,
1:08:37 but I didn't get a degree in
1:08:39 programming. I never really learned
1:08:41 programming. I just do
1:08:45 it. Okay, you don't need to know
1:08:47 programming anymore. You can go make
1:08:48 applications. What? Really?
1:08:53 I didn't get that art degree. My My
1:08:55 stepdad told me I was untalented, so I
1:08:58 don't draw anymore. But I really like
1:09:00 art. Well, you don't need to draw. You
1:09:02 can just go make art.
1:09:05 Really? It's like any limitation you
1:09:08 think you had.
1:09:12 Any if they haven't already been
1:09:14 obliterated, they're about to be
1:09:16 obliterated.
1:09:18 And then and then you get to ask
1:09:20 yourself the question, if I could do
1:09:22 anything, if I could do anything, what
1:09:25 would I choose to do? What difference
1:09:27 would I want to make in the
1:09:30 world? And then you go do that. That's
1:09:33 what this is. That's what this is. It's
1:09:36 not all the [ __ ] everyone talks
1:09:39 about. Robots are going to kill us.
1:09:42 Might they kill us? Well, they might,
1:09:43 but I'll tell you what, I'm going to be
1:09:45 fully self-expressed when they do. Tik
1:09:48 Tok pan after hours. Also, Tik Tok cam
1:09:50 is off center. Oh, sure is. That's
1:09:55 embarrassing. I will never forget I told
1:09:58 you I was too old and you said, "Oh, get
1:10:00 over
1:10:02 it." That sounds like
1:10:05 me. I'm too old for AI. Ah, get over
1:10:10 it. I'm glad I said that, Becky. And it
1:10:13 was shortly after that that we we
1:10:15 started building custom
1:10:18 GPTs. And I remember the night you came
1:10:20 into the live and you were like, Kyle,
1:10:22 I'm I I made this quilt GPT. Like, I
1:10:26 love quilting and I've got this GPT
1:10:28 that's making me quilt patterns. And it
1:10:30 was like a week after you were like, I
1:10:32 don't think I could do this AI stuff.
1:10:35 [ __ ] awesome. [ __ ] awesome. So
1:10:38 anyway, Champ, what are you craving
1:10:41 cheese?
1:10:42 Cold, please. What is your
1:10:49 deal? The
1:10:56 hell? Come on. All
1:11:03 right. I think Champ had to go. Poop.
1:11:06 The hardest thing for me post abuse is
1:11:08 believing in myself. Do you help me,
1:11:10 Kyle? Oh, thank you very much.
1:11:12 Becky, thank you. Oh, thank you. Um,
1:11:15 yeah, it's, you know, it's funny, Becky.
1:11:19 Um, very early on in the AI salon, one
1:11:22 of the things that
1:11:26 that I noticed a lot. I don't know. I
1:11:29 don't know why it hit me because this
1:11:31 probably occurred back in the early days
1:11:33 of the worldwide web, but what I noticed
1:11:35 was whenever people would would kind of
1:11:39 step up on stage to show their work or
1:11:41 or you know, I'd talk to them in the AI
1:11:44 salon, they would apologize. There was a
1:11:47 lot of
1:11:49 apologizing of, "Oh, well, I'm not
1:11:51 really good at this. I'm sorry. I'm not
1:11:53 really good at this. I'm not very
1:11:56 technical. I'm sorry." There was just
1:11:57 lots of apologies from men, from women.
1:12:01 Um, and and we kind of went on a crusade
1:12:04 for like six months of like no
1:12:07 apologies. Like, stop
1:12:09 apologizing because the reality is, you
1:12:13 know, I am I am hopefully bright and
1:12:16 shiny evidence of this. There are no
1:12:20 experts, right? the these tools are
1:12:23 changing so fast and and they're
1:12:27 accelerating away from us so fast that
1:12:29 you literally can't keep up with them
1:12:31 and and what you were genius at 6 months
1:12:35 ago is no longer relevant today. So
1:12:38 there are no experts and so let's just
1:12:41 lose the apologies. And the minute
1:12:43 people sort of started dropping the
1:12:45 apologies, they just stepped into this
1:12:47 power of self-expression.
1:12:51 Oh, here's this thing I did without the
1:12:53 apology. It was incredibly powerful. And
1:12:55 it's like I feel
1:12:57 like if we can maintain that as the
1:13:00 salon expands and as this channel, you
1:13:03 know, continues on there, there's such
1:13:05 power in that. So anyway, so thank you
1:13:07 for acknowledging that. Ann Murphy
1:13:09 backstage and Murphy, I know. Get up
1:13:13 here. What are you doing?
1:13:16 Okay. Well, I should forewarn you that
1:13:19 you know what? This is how how we know
1:13:21 we're all friends and fam. Yeah. There's
1:13:23 only a couple of people as until today
1:13:26 who have ever seen this robe,
1:13:30 which could literally get up and walk to
1:13:34 the Mall of America. Exactly. Which And
1:13:38 what what what is you do you know your
1:13:39 distance from the Mall of America? I
1:13:41 know it's I think 232 days of pogo
1:13:45 sticking. Okay, that's good. Perfect. or
1:13:47 it might be 2,32 days. I one of those.
1:13:53 Okay. So, I had I wanted to to dive in
1:13:56 on a little something. Oh, good.
1:14:01 So, yesterday on the podcast, we were
1:14:05 briefly talking
1:14:07 about was it yesterday we were talking
1:14:09 about the I'm sorry thing. Yeah. Oh,
1:14:11 yeah. Yeah.
1:14:14 And like I feel like we're all ready to
1:14:18 go to the next level of that kind of
1:14:22 uh way we talk about ourselves.
1:14:26 And I do think that one of the pieces is
1:14:29 the AI readiness component.
1:14:32 Like
1:14:34 that's that's what it is. And and the
1:14:37 the thing that I wanted to say
1:14:41 um that kind
1:14:43 of follows what you were saying about
1:14:46 the people who have their head in the
1:14:48 sand is that I've been starting to think
1:14:51 more and more about how AI readiness has
1:14:54 a ton to do with communication
1:15:00 like hard conversations.
1:15:03 Yeah. and
1:15:05 and translation
1:15:09 and empathetic conversations where
1:15:12 you're like dying inside but you just
1:15:15 have to keep listening, you know. Well,
1:15:17 you said it and you you started out the
1:15:20 the the conversation yesterday. you're
1:15:22 working with these these big
1:15:24 organizations and you're looking at the
1:15:26 impact of AI and you're like, "Hey, it
1:15:29 looks like about 75% of your consultants
1:15:33 hours are not necessary, right?" Like
1:15:36 that's that's pretty rough, right? That
1:15:38 there's there's implications to that and
1:15:40 that's you're not causing that. You're
1:15:43 just saying, "Hey, if I just look at
1:15:45 what this stuff does, right? Some shit's
1:15:49 gonna happen, right?" Yeah. Exactly.
1:15:52 Yeah. Yeah. And you know, so I delivered
1:15:55 that report today to them and
1:15:58 Yeah. And you know, as I did, I felt
1:16:05 these mixed feelings about it, you know,
1:16:08 because it's going to that transition
1:16:11 from where they are to where they're
1:16:13 going. Yeah. And figuring out what
1:16:15 they're going to do with the found time.
1:16:17 Is it cost savings? Is it another layer
1:16:19 of excellence? Is it business
1:16:20 development? Is it time off? Blah blah
1:16:22 blah. Y that's going to be brutal for
1:16:27 their their contractors, their
1:16:29 consultants, their staff. Y and well, so
1:16:33 so there's there's a silver there's a
1:16:36 silver lining to it though, which is
1:16:37 this for the people. So so so one of the
1:16:41 thing is okay, if 75% of the hours goes
1:16:44 away, then some people are going to, you
1:16:45 know, lose their their gig, right? Yep.
1:16:49 For the 25% that
1:16:51 stay, they are likely going to be the
1:16:53 ones that step into this AI world and
1:16:56 they're going to amplify what they do.
1:16:57 And you know, for the they're going to
1:16:59 crush. Yeah. For Exactly. For the people
1:17:01 that get curious now, when those
1:17:04 decisions get made, if this is a big
1:17:06 institution, it'll probably be six
1:17:08 months or a year until, you know, your
1:17:10 report ends up in some actionable thing.
1:17:12 If you're involved, it'll probably
1:17:13 happen faster. But but you know, but but
1:17:16 let's say six months down the line,
1:17:18 people still have time to get curious
1:17:20 about this stuff. And if you do, you
1:17:22 will likely be one of the ones that
1:17:24 survives, you know, the the the
1:17:26 brutalness, right? Oh, a million% a
1:17:30 million%. And uh they serve like
1:17:34 children's hospitals and stuff.
1:17:37 So when you think about like we talk
1:17:39 about like the ripple effect of this
1:17:42 like they're using AI to get smarter
1:17:46 about how to advise their fundraising
1:17:49 clients means they'll raise more money
1:17:52 for the mission of different hospitals
1:17:55 and children's hospitals and and [ __ ]
1:17:58 like that. Yeah. Yeah. And you know
1:18:00 hopefully fingers crossed they can do it
1:18:02 more efficiently and they can so so the
1:18:05 money goes more directly to programs
1:18:07 than administrative or marketing or all
1:18:09 the other [ __ ] that happens. Right.
1:18:11 Yeah. Let me say one factoid that people
1:18:15 might find interesting and then I'll
1:18:17 I'll hop off because you've got you've
1:18:19 got stuff to share with the nice people
1:18:21 or crispiness levels that you need to
1:18:23 take care of. It's the crispiness level
1:18:25 but yeah what's the nugget? Okay. Here's
1:18:28 here's a tiny short use
1:18:31 case. It
1:18:33 is in our industry. We all do lots of
1:18:37 feasibility studies and so you know you
1:18:40 may have 50 hours of conversations that
1:18:42 now most people
1:18:44 have
1:18:46 recorded you know but some people write
1:18:50 out in hand. I think I told you this
1:18:51 part. And so for every hour that they're
1:18:55 in an interview, which the interviews
1:18:57 take way too long because they're
1:18:58 underprepared every single time because
1:19:00 they haven't done the research, right?
1:19:03 They allocate an hour to
1:19:06 process the notes from the hourlong
1:19:10 meeting. An hour. Wow.
1:19:13 So that's why you know if you take those
1:19:16 you take that two hours and you cut that
1:19:19 second hour down by what I mean it takes
1:19:23 seven minutes to process the transcript.
1:19:25 Seven seven minutes on the long side. On
1:19:28 the long side if you really you use
1:19:30 right. Yeah. Exactly. And how
1:19:32 perfectionist you are and what prompts
1:19:34 you have blah blah blah. So then when I
1:19:37 turn around and say that
1:19:39 like it sounds crazy. I sound like a
1:19:42 crazy person, but until they try it and
1:19:44 they'll be like, "Holy [ __ ] we're
1:19:46 paying the people all this money." Yep.
1:19:48 Yep. Exactly. Exactly. So, lots of
1:19:51 opportunities. Lots of Yeah. Massive.
1:19:53 Massive ones. Um, thanks for hopping up
1:19:56 here. That was so cool. Yeah. This was
1:19:58 fun. And the robe is not that dirty.
1:20:01 It's
1:20:05 no comment. Okay, I'm gonna Can you kick
1:20:08 me off? I don't know how. We'll take it.
1:20:10 Yeah, we can figure it out.
1:20:12 Okay, bye.
1:20:16 Oh, that's
1:20:18 [Music]
1:20:21 awesome. Boom. Gone. All right, I got to
1:20:25 get out of here. Valerie Cox, I got
1:20:27 mixed reviews mostly. I get weird looks.
1:20:29 I end up feeling nerdy. So, Valerie Cox,
1:20:33 that that feeling of feeling
1:20:36 nerdy, just keep living in the weirdness
1:20:39 of it. Just keep putting yourself out
1:20:40 there. Keep keep learning out loud. Uh
1:20:43 because at some point all the people
1:20:45 that are giving you weird looks right
1:20:46 now are going to realize they've got to
1:20:48 deal with this. And guess who they're
1:20:50 going to come to? They're going to come
1:20:51 to you. Um we see that over and over
1:20:53 again here. Um source camp. Um I also
1:20:57 got booked today to speak to a senior
1:21:00 retired church group and they are paying
1:21:02 money. Yes. Yes. Yes. Go tell the
1:21:07 people. Just keep putting yourself out
1:21:09 there. Kelly Camp has been um she's had
1:21:11 this AI agency for two years now. She's
1:21:14 figuring it out. She is making a
1:21:17 reputation for herself in the business
1:21:18 community and people are starting to
1:21:20 write checks. The you know,
1:21:25 she's out there solving problems. Solve
1:21:27 problems. The AI is just a tool to get
1:21:30 you there. All right. Corey Sandler
1:21:33 uploaded
1:21:35 her Oh. voice to the irregulars tab.
1:21:39 Cool. Let's go look at
1:21:43 that new activity. Let me share my share
1:21:47 my
1:21:57 screen. All right. This is Corey Sandler
1:22:00 saying, "Oh, this is going to go to a
1:22:01 different tab, isn't it?" It did. Hang
1:22:04 on. Damn it.
1:22:08 All these new tabs. Everybody's got to
1:22:11 spawn a new window to screw up my
1:22:14 evening. Okay. Uh, which voice? So,
1:22:17 Sage. Let's see. A voice that only
1:22:19 exists between dreams and
1:22:21 departure. She tends to stories the way
1:22:24 others tend to prayers. Oh, I like it.
1:22:29 Sh. Don't worry. I'm right here. I'll
1:22:33 stay until you're asleep. Tonight, we're
1:22:38 counting sheep. But not just any sheep.
1:22:42 These ones, they have names and tiny
1:22:46 little stories just for you. Are you
1:22:50 ready? Then close your eyes. Let your
1:22:53 breath fall heavy in your chest and
1:22:57 listen. One, her name is Lynden. She
1:23:01 wears a pale blue ribbon tied round one
1:23:04 ear, and she always walks sideways when
1:23:08 she's nervous, which is often. But
1:23:11 tonight, she jumps right over the moon
1:23:16 silently. The ribbon trailing behind her
1:23:20 like a comet's tail. Two, that's
1:23:23 Thimble. He was born with one black hoof
1:23:27 and a thousand questions. Very cool.
1:23:30 Beautiful.
1:23:33 Don't worry. Love it. Love it. Love it.
1:23:36 Love it. And then apparently if I open
1:23:40 this, she's got the prompt in here as
1:23:43 well,
1:23:46 right? Yeah. There you go.
1:23:49 Beautiful. Oh, not
1:23:57 sharing. I don't know. Okay. So, let me
1:24:01 get out of here. Um, so go check out
1:24:02 Corey's uh Corey's voice there. She
1:24:05 shared her prompt as well. Thank you for
1:24:07 sharing that. Generous. Appreciate it.
1:24:11 Um, so tomorrow,
1:24:14 um, if you're a fan of if you're a fan
1:24:17 of,
1:24:19 uh, the learning lab of the salon, um, I
1:24:22 do this thing on, uh, on LinkedIn at
1:24:26 11:00 a.m. Mountain time on Fridays
1:24:29 called AI office hours. It's a really
1:24:31 remarkable group of people that show up
1:24:34 um and we just kind of shoot the [ __ ]
1:24:36 about what's going on with AI and who's
1:24:38 doing cool work and you know some people
1:24:41 are really advanced, some people are
1:24:43 just trying to figure it out. There's
1:24:44 everyone you know at every level in
1:24:46 between. It's just kind of the way it
1:24:48 is. I would argue that the group that
1:24:50 shows up for AI office hours are some of
1:24:52 the smartest people in the game today
1:24:55 with this generative AI stuff. So come
1:24:57 hang out with us. Um, so if you just go
1:24:59 to my LinkedIn profile, Kyle Shannon,
1:25:02 um, and click on events, um, you'll see
1:25:05 a bunch of them there. We've been using
1:25:06 the same URL for two years, so just grab
1:25:08 one of the old ones. I'll send out a
1:25:10 reminder in the morning. Um, tomorrow
1:25:14 night is Friday night date
1:25:16 night. If you got a date, bring your
1:25:18 date. If you don't have a date, I'll be
1:25:20 your date. Just come
1:25:21 on, air fry the
1:25:24 nachos, microwave the the hot pockets,
1:25:27 do what you need to do. And then what do
1:25:29 we do different on Friday night? Date
1:25:31 night.
1:25:33 Nothing. Except it is
1:25:38 Friday. Tik Tok pin. All right, hang on.
1:25:41 Let me see what we got here. Um, does
1:25:44 everybody know that Joy Perie is in
1:25:46 partnership with Luma? Yes, she's she's
1:25:48 in the creative partner program. Joy
1:25:51 Perty is a talented filmmaker. And uh
1:25:55 yeah, by the way, if you have not
1:25:57 applied for the creative partner
1:25:59 programs of all the video tools, if you
1:26:01 want to make video, it's really
1:26:03 expensive. Apply for the creative
1:26:06 partner programs. I've gotten in three
1:26:07 of them now. Um just by doing it. Just
1:26:10 put it out there. Put yourself out
1:26:12 there. H But I'm not talented enough.
1:26:15 Neither am I.
1:26:18 I'm just pushy tick. Thank you, Kyle.
1:26:21 Great meeting. Thank you, mods. Thank
1:26:22 you, Kake. Great. All right, let me get
1:26:25 up on out of here and uh so I'll see you
1:26:28 tomorrow. So, come hang out at office
1:26:29 hours and then I'll see you tomorrow
1:26:30 night. All right, and we will figure out
1:26:33 some stuff to do. All right, peace out
1:26:36 YouTube and others. We will see you
1:26:39 soon.