Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
A.I. Music | Grammys Outline New Rules #AITownHall
Recorded: June 20, 2023 Duration: 1:16:20Space Recording
Short Summary
The discussion primarily revolved around the implications of AI in music creation, touching on the potential for AI to surpass human creativity by analyzing vast data sets. Participants debated the role of AI as a tool versus a creative force, with some expressing concerns about the loss of human touch in music. The conversation also highlighted the rapid advancement of AI technology and its impact on the music industry, including the potential for AI-generated music to become mainstream.
Full Transcription
Hei folk, vi er på å få alle opp.
Stelman, jeg hører at du har hatt en marathon de siste få tider.
Ja, det har vært galet. Vi har bare vært arbeidende nonstop for denne større kvisten.
Det kommer til å være stort.
Jeg tror at vi kommer til å løse traileren.
Very soon, I think people will be significantly impressed, and yeah, the special guest is a major, major guest.
In my opinion, the biggest guest we've ever had.
Obviously, Mario may disagree, may think it's Elon Musk, but just based on the circumstances, who he is, what is happening, it is going to be huge and major.
That is insane. So somebody who could even compete with Elon Musk, that's a pretty bold claim.
So I think we're all waiting with bated breath out here in the audience.
So yeah, we're just bringing everyone up, so just hold on for a few minutes.
Once we get the panel up, then we'll get it started.
Awesome.
Ok, så EYC, hva er den nyeste AI-nytene fra kommunitetet?
Sjå, sjå.
Jeg tar den av med en ny headline, og så går jeg gjennom noen høglighter.
Den store headlinen for i dag er om gen-AI-musikk.
Grammierne for nære dager har utsett nye regler.
and there's actually been a bit of controversy, right?
Is this saying that people can use AI for songwriting Grammys?
The answer seems to be yes, to some degree.
So, in other words, AI is somewhat authorized to be used.
But at the same time, there's controversy
because of the fact that it's very limited in its use, right?
So, trying to keep...
mostly humans, you know, in the business of winning artistic awards.
This is near and dear to my heart because I've got a musical family,
you know, married to a singer-songwriter, my father's an opera singer,
and I myself, my day job is actually as an artist, a writer, and a film director.
That's literally what I do during the day, you know, to pay the bills.
And so this creativity in Gen. AI is definitely something near and dear to my heart,
so I'm excited to dive into that.
Before we go to the...
The main banger though, lots of really interesting stuff.
So, for example, the House is starting to introduce legislation
that will require Congress and the White House to appoint, you know,
20 people across government, industry and civil society,
as well as CS, thankfully, so a few computer scientists,
in an AI commission where people are going to be able to...
I think the question on that, I know we're going to probably talk about this
in another space or later on in this space, but...
Er du bekymmer om hva du ser, nesten hver AI-episod vi gjør, er det noe slik snakk om noe slik legislasjon, noe slik reglering, og det er ingenting som har vært konkret eller lagt ned, så hva er dine tanker på det?
Yeah, well, I think this is a classic example of where people don't really know. I'd say even like industry experts have a tough time parsing through all of the, I mean, this is a highly technical field in so many ways. We have some great experts here. But, you know, I think like Congress has a hard time figuring out what the internet is, right, and how their mobile phones work. Sorry to be...
You know, I think actually maybe even they would agree with me on that.
Like, it's just difficult stuff.
The AI stuff is going to be even more complicated, right?
This is why I think the EU legislation, the EU AI Act...
is almost like jumping the gun, right?
It's like putting legislation
upon the internet
when we're in the days of ICQ, right?
In the early 90s, right?
We don't even know what this thing is,
let alone how to legislate it.
So I think there's a little bit of worry,
but it's good to see
that they'll get some experts on board.
So hopefully the discussion will be fruitful.
We'll see where things go.
So in other news, you know, of course, Chinese President Xi Jinping and Bill Gates did, you know, discuss the rise of AI.
You know, I think the two big contenders for AI innovators in the world is going to be the U.S., particularly the Bay Area, as well as China.
To be frank, those are the two big AI players.
For a while here in the Bay Area, we thought China would overtake us by dance, TikTok, facial recognition.
But thanks to open AI and stuff, we in the U.S. have...
I see Josh thumbs downing there.
I think we in the US have quite the lead actually now.
So that's good to see.
No other news, Biden is going to meet some AI experts in San Francisco, and a bunch of other things just happening around the world, like Southeast Asia setting guardrails on AI with new governance code.
A lot of people trying to control AI, but do we even know what it is?
So, you know, I think that's an open question.
Actually, Josh, before we dive into the main meat, I mean, did you have thoughts about geography and who's going to be winning the geography war when it comes to AI innovation?
Sure.
Ja, takk. Jeg tror, nei, du har bare gjort en bra punkt.
USA har definitivt overkommet noe av denne tekniske landskapet
med OpenAI droppning chat-GPT 3.5 specifikt.
Så jeg tror vi kommer til å se mye mer avvansning
fra hjemstaten her i USA.
Jeg vet at dette er en global tank, men jeg ville bare harpe på det.
Ja, definitivt, TikTok var en interessant punkt
for kinaisk infrastruktur og alle data-server som vi var å swappere,
men...
Men ja, takk for det.
Ja, jeg tror vi har fått leaden for den momenten,
vi som er USA,
men vi får se om den leaden last,
for det er mye konkurrens i verden.
All right, så
vi går og dyver inn i det første.
Så vi pratar om, så Grammy's
bare utslog de nye reglerne,
og jeg vil dyver inn i noen ting.
there's the how so what is it i mean yeah you actually go ahead continue but what are these
new rules like what is what's so important about them like i assume they would just not allow any
form of ai or they would make it either for me just without even looking into the issue
significantly any any time you go into these type of situations either you allow it
completely or you don't allow it because when once you start putting some kind of
like semi parameters onto it,
it's going to be hard to ensure that those parameters are met.
But anyway, that's my initial thoughts,
but I'd just love to hear from you.
To the contrary, these rules state that the minimum amount of human contribution that's required is 20%.
So this is...
This is perhaps one of the main things that has come out of this.
So, you know, it's interesting because one of the things is that only humans can win the awards.
It's actually written there, right?
But only 20% is required.
So 80% of the contribution can be from AI.
So take that what you will.
I think the big question, Svink, is that's a great outline, but the percentages are difficult, right? And actually, there's a different article that came out recently about people using AI in their work, but not telling people they're using AI. I think even like the Airbnb co-founder was talking about this.
So like,
When we're in a world where people are just using chat GPT in the background,
like, how do you know that X percent was written and helped written by an AI versus not, right?
So Sphinx or others, I'd love for you to kind of outline.
How can you say, okay, 20% is not?
Yeah, that's a really good question.
I mean, I'd like Brian as well to jump in on that, if that's okay,
because that's the question I've got.
Because as soon as you allow some kind of window,
or basically a movement on allowing certain things to happen,
then it becomes like, who can game the system the best?
So why not just allow it completely?
But Brian, I'd love to hear your thoughts.
Well, thanks Mario. This is exactly the problem. You know, and it's going to be a moving goalpost on what is the definition of AI as we move forward, right? I mean, I started working with AI in the late 70s, early 80s, and those were just basically expert systems.
You know, and we can have sort of a hazy notion of what AI is,
but you're having regulators and you're having, you know, committees saying,
well, this is the defining amount that you can use, things like that.
And let's take it an iteration backwards.
When Google Search came about,
I don't know if people had so much of a concern of whether or not somebody looked through an encyclopedia
or in fact looked at Google search, or had an assistant help them write a transcript from a Google search,
something along those lines.
I get it. I mean, these things need to be discussed,
but typically what happens is a lot of hand-waving goes on,
and it's like, wow, AI, we just have to be really concerned.
And I absolutely promise you, the AI that we think is AI today
will look absolutely ridiculous five years from now.
And we'll look like, why were we concerned about those things? But yes, as far as artistic things, I, all right, let's look at any of the most popular top 10 songs on Spotify over the last five years.
How much of that was mechanically produced inside a computer?
How much of autotune was allowing a vocalist or even a drummer to quantitize,
or a guitar player to be on pitch?
And what percentage of that should have been allowed to win a Grammy?
I don't know.
And so it's an interesting slippery slope that we have to all explore.
Ja, jeg tror at vi tror at det er bare verkrykninger,
det er som, ja, en gitarre er en verkrykning,
en piano er en verkrykning,
så tror du, Brian eller andre,
tror du at AI er bare en verkrykning i framtiden,
eller tror du at det er noe annet enn en gitarre?
Ja, la meg finne på min pointe.
Det er definitivt en verkrykning.
Første spørsmålet er,
At what level are you just phoning it in, right?
If, for example, I can type a phrase,
and I get this immaculate picture
that a photographer would have taken maybe hours to set up,
and it's done by me just typing a phrase,
is that me really contributing to that?
At that level, probably not, unless...
We can argue that the prompt used to create that image wasn't a sentence, but it was an entire story.
So now we're looking at a different iteration.
What is the input versus what is the output?
And I think as time moves on, those are the goalposts that we're going to start looking at,
is how much did somebody have to put in to constructing what came out?
And obviously, for all of us, we're going to judge it from our gut.
If somebody just types a two-word statement into a prompt,
and something grand comes out, and they take some kind of credit for it,
we know that that's probably not really a great deal of work.
But, you know, I work with people right now, like with Claude 100k.
We have a 100k token window, and I've seen some prompts that are literally mini stories.
And what comes out of it is something absolutely phenomenal.
And as a context window in token amounts, and these are the input...
You know, limits, to make it simple, for us prompting AI.
As those get more sophisticated and complex, then we're going to start looking at that and saying, wow.
And I would finally say this.
Look at what some really good mid-journey prompts look like.
They are incredible at setting a scene and a stage and weaving a storyline.
And so...
You know, let's not diminish that aspect of it, because linguistically, philosophically, you know, just storytelling, writing, you know, using Hero's Journey, the whole thing, this is required for some incredible outcomes.
That's great, Brian. You know, and it's interesting, I see a lot of reactions from the audience, both on the up and on the down.
So, John, I want to go to you. I know you had your hand up earlier and wanted to opine on this.
Do you think, Brian talks about how it's about effort.
It seems like it's about effort. It takes effort to learn how to use a piano and be a master at it.
Is prompt engineering not effort?
But yeah, John, please brave in.
So this is really an unfolding discussion that I find very interesting.
So let's go to the Grammy press release.
Was that a validation of AI or was that a validation of humanity as the core creative construct here?
I kind of read it both ways, but when you look at the 80-20 rule,
It seems that this is a validation of technology
as a core element of the creative process.
To me, that's the big news.
The fact that someone can walk up on stage and get an award
is really a function of the entertainment and television construct.
So I would be very interested in Brian and chiming in on that.
But to me, that's what I'm writing about.
And before you answer that, Brian, as well,
another point just to what you were saying,
and then you can maybe...
respond to both.
So what you're saying is
The way it comes across to me isn't the quality of the music or the artistry that's been involved, but it's the quantity.
So how is the prompt being relayed? Is it a sentence? Is it a very good paragraph?
So aren't you basically deciding who is the best prompter as opposed to who is the best musician, who is the best artist?
Brian, I'm sorry, anyone, anyone.
Yeah, go ahead, Brian.
So, when rap music became popular, and I lived through this,
we dealt with sampling.
I'm a musician, and I spent an inordinately long hours
alone in my room playing guitar, and then, you know,
From a musician standpoint, when I heard that somebody can sample three or four of my notes and essentially build an entire corpus of work on that, or a drumbeat.
Look up 808 drumbeats, you know, it's been everywhere. That guy never got compensated, right?
Whether you love or hate the music, it's irrelevant. It's that somebody actually...
was physically in front of a drum set to create that sound.
And one can argue this true within some of the,
you know, Facebook just put out an incredible music AI system,
and I've been working with it quite diligently,
and it can come up with, you give it a melody,
and it will give you,
tusen av variasjoner av den melodien.
Det gjør utslippning av en musikk,
mykkelig som utslippning av visuellt.
Og så, ja,
entertainment kommer dramatisk å forandre,
men dette har vært på en ark.
Og så,
det er på en stund at alle sier
«Åh,
når vi ser kunstnere som blir krasjt».
And yes, artists are being crushed.
Have no doubt of where I am on this.
I'm just also a realist about the Pandora's box.
In a perfect world, I would love to see live music musicians everywhere
and just applying their art.
I'd love to see young people who actually have a mission
just to be a musician for the rest of their life.
But now what we've come to is that the standards have changed dramatically.
And I could say again,
Vi sette oss selv for dette av å akceptere, kjøpe og spela musikk som primært var samplade musikk.
Vi går på live-serier, og så finner vi ut at de fleste av det ikke er egentlig lagt på av den utgivaren.
De utgiver på stedet, og det er valuert.
Jeg er mer enig i å ta min penge på å bli utvandret.
Utevandring er ikke musikalskap, ikke sant?
Så når en går på stedet eller rekorder en album, er vi teoretisk utvandret.
Men er vi utvandret?
I 1970-80s, du hadt den poseur, eller? Var den guitaristen på den guitaren eller var det en hundre person?
I dag, har folk ingenting med dette. De hører en lyd, som er utvalgt som en overflødning.
All vi får er mest korus, og ingenting vers.
And we get a repetitive melody and it sounds pretty much like the last song that came up on Spotify, not by accident, because AI actually chose that about three years ago.
Brian, I think it's such an interesting point.
I think it's a question around when is a tool a tool and when is a tool something else, right?
Exactly.
We brought up Gino.
Even more importantly, just let me finish up this point.
Yeah, John and Gino.
Yeah, go for it, go for it.
How do we quantify creative work?
In other words, if a car in the Indianapolis 500 wins the race, is that 80-20?
I mean, how much proportionately do you really give to the driver versus the engine and the mechanics and the technology?
So I think this, you know, I try to quantify this in a way by saying horsepower.
Each driver has access to the same car that they choose, right?
I have my hand up, sorry, a quick point to John's point.
Hi everyone, Team Human here. Kidding.
So, you know, it's a derivative of derivative of derivative, right?
So if you look at the source, the source is mostly, in this case, even in music,
is humans.
And it's, if you look at chat GPTI,
consider it as a derivative,
and maybe perhaps a derivative of a derivative.
But if you look at the source, it definitely comes from a human.
Now people can say, oh well, it can reason and it can talk,
and I mean, right now it cannot reason.
The day it starts reasoning about things and gets the deeper meaning,
by deeper meaning I mean the white space that is not written between the words,
like, read the tea leaves, then I think we'll have a real sort of...
argument around this
I like a lot
what you said, but the question is around
like, well, once the in-betweens
right, like, you know, in between
Ernest Hemingway's work
there's so much in that sparse prose, right
but once
an AI can actually intuit itself into that
and create the space in between,
that's actually an interesting philosophical question.
Gino, I want to go to you.
I know you had some reactions when you were in the audience.
Welcome to the space.
So please go for it.
What's up, guys?
Let's hear what you have to say.
How are you?
Mario, been a minute.
How you doing, buddy?
Mario's away from the space at the moment,
but yeah.
Okay.
So I'm a two-time Grammy winner,
seven-time nominee.
I am also...
en stor proponent av AI, AI musikk.
Jeg har en generativ AI musikkprojekt.
Jeg elskar det Grammys gjorde,
fordi jeg tror det har det i to måter.
Til en annen måte,
det anerkommer teknologi som en tool,
og jeg tror det er viktig,
fordi mange av mine foreldre
rejører ideen om AI,
og er bevægte av ideen om AI.
Det funneste om AI musikk for meg,
er alle mine kollega-tekkbroer som er utrolig og som er sånn
Åh, sykt! Nå kan jeg ta på musikk
og...
Nei.
Du kan ikke.
og jeg tror...
det er en så viktig distinsjon å forstå at
de er utrolig nye verktyg
men de fleste har ingen smak
de vet ikke hvordan å ta på musikk
de som tar på musikk, og har vært utdraget i tider
de vet ikke hvordan å ta på god musikk
Og jeg tror at det som AI kommer til å utforske de neste få årene,
er at om du ikke kan ta en hit,
AI kan hjelpe deg, og det kan gi deg foundationen og proppsen,
men det er ikke så mye forskjellighet i det AI er
som noen har sagt tidigare,
som samplepakke, eller bare å gå på splitter og finne dope sounder,
eller å bruke Arcade,
eller alle disse verktygene som produksjonene har allerede brukt.
And really all it will do is help you hyper focus on what you're trying to find and give you just incredible outputs, you know, and that's really, I think, what we're going to see in terms of...
how, you know, like,
are you basically saying that it requires taste?
That's what you're saying, right? Like, today
you need taste to create these, right? I mean, look at
that Jay-Z song we used to, you know,
we used to advertise the space, right? I mean, that sounds like
Jay-Z, right? I mean, it was probably
somebody who knew music who created that, but
you know, how many years will it be before
or anyone can do that.
Well, so that comparison,
that example is,
you know,
that's a songwriter
that made that.
I think a lot of people think
that, you know,
the Drake and Weeknd song
that went viral
that was AI,
an AI vocal rap,
you know,
people think that someone
just plugged in like,
Give me a Weekend Drake hit song that sounds like, you know, started from the bottom.
When in reality, that's...
And I know who did it.
I know who Ghostwriter is, so I'm obviously a little, you know, which I won't say on stage.
But, because I started as a Ghostwriter.
But anyway, those are songwriters who are making these.
And these are demos.
So when we pitch records, a lot of people, again, don't know how...
pitching and songwriting works.
When you pitch a record, like any record I've wrote
for like Ariana,
or Natty Peluso,
or, you know, Sweetie, a rapper,
a singer,
you're writing a full
demo. These are full demos
with my voice
or a demo vocalist's voice on.
They're full songs that are being pitched.
99% of the music that you hear
on radio is written by a songwriter.
And the only difference between
you know, these vocal AI models,
is that essentially a songwriter
can just put the artist on the song.
So that creates a whole bunch of other legal issues
with right of publicity and all types of things
that are going to be a battlefield for years to come.
Well, probably...
Vi kommer nokt til de legale problemene, for det er sitt eget sted i og med seg.
Jeg ser mye 100% og mye thumbs up.
Hvem her tror at i noen år, disse verktygene er pågående så snabbt?
Vi er alle i publikum så entusast av,
enthrault av hvor snabbt chat-GPT har vokst de siste månedene.
We are well outside of Gino.
Yeah, I wouldn't necessarily I agree with you know, and absolutely amazed by his work.
And I gotta say this, the tools are going to get it's going to get more simplistic.
I mean, right now, Facebook's music gen is open for anybody.
Does it produce hit recordings? No.
And Gino is absolutely right.
There's a zeitgeist of what's taking place in society
that normally dictates what is going to create a hit.
There is timeless music. It's extraordinary.
It's extraordinarily difficult to write timeless music. The older you get, the more likely you are to resonate with timeless music. But on top of that, whistles are getting more sophisticated.
Hey Brian, is anyone altering?
Ryan, det ser ut som at din kontakt er litt svart, men vi skal bestå det og komme tilbake til deg.
Jeg vet at vi har fått noen folk som har vært på vann, så jeg prøver å få noen nye folk, men Illustrata, jeg tror at du er ny i denne området.
Jeg vil bare kommentere på det du har spurt om Gino, det Gino sa, for jeg tror Gino er rett om situasjonen i dag.
Men jeg tror EYC, det du sa, er korrekt.
The technology is advancing so rapidly that I don't think that in the future there's going to be such a gap.
I think that this gap is going to be lessened, right?
So that it will be, it sounds crazy to say that it's kind of like who has...
It does sound crazy, but I do think as we advance forward, it's not going to seem as crazy,
because look at the advances that we've made in such little time,
that when I hear EYC say this, it's kind of like, you know, who has great taste?
To me, that doesn't sound crazy, because I can just see in a year or two or five,
What technologies will be available
and how that will be.
And I'm not saying that that's the way it should be,
but I can just see it going in that direction.
I do think...
Thanks, I think you're...
Paula, one sec, Gino, before you jump in,
we do have some...
ok, vi har nois major,
breaking news for those who are fans of Mario Spaces.
Så, the major guest vi har vært tala om,
vi har vært teisning om for så länge,
men den major guest er i fakt
former prime minister Imran Khan.
Det har bare vært annerledes,
det er en promo-video
for anyone som har vært seende
ting som skjer i Pakistan.
Jeg har faktisk en venn der som er
former MP og former medlem av
of his party.
Wow, this is huge, folks.
So, yeah, big congratulations.
I know it's a bit of a non sequitur,
but, yeah, huge congrats
to all the folks involved in this.
This is pretty incredible.
Yeah, I was just going to say...
You know, and yeah, let me do a little clap emoji for that here.
Hold on, let me clap emoji that.
I don't want to just step over that.
Look, yeah, I think, you know, yes,
the technology is going to get incredible.
I think what we've already seen happen in the past six, seven years
with the growth of TikTok and AI algorithms
are the skill gap has significantly shrunk.
And the, you know, as volume and output has 10x'd,
og musik har blitt fast fashion,
det har absolutt vært en nærvare,
og det har allerede skjedd.
Og som teknologi fortsetter å skala,
vi kommer til å se det fortsette å skala med det.
Og jeg tror at AI vil definitivt
bringe musik til middelklasset,
My fear with AI music and these tools is it's going to just continue to 10x output,
and there's just so much goddamn music out there.
It's overwhelming, and it's continued to cheapen the value of music.
And I feel like it's only going one direction, right, Gino?
Correct.
In terms of output.
So I think...
And I don't disagree with that, Gino.
I don't disagree with that.
Strangelip and Sphinx, I think, let's get Illustrata, but I think there's so many, yeah, a lot of people are trying to jump in on this, but let's get Illustrata, I know she's been trying to get in for a little while.
Thanks.
Yeah, so I think I kind of liken it to when YouTube was created
and it opened it up, opened up creating, making videos, sharing videos
to so many people that, you know, didn't necessarily need to be affiliated
with a TV station or a movie studio or something.
And that just opened up the ability for so many more people to be creative.
But at the same time, it doesn't mean that everyone who creates a YouTube video
is going to get millions of followers
or it's even going to be good quality anything.
I think AI is very similar
in that you need to know how to use the tools
to create something of quality.
All right, thanks for that.
Ok, so strangely, why don't you jump in, because I know you disagreed with some things some people said.
Yeah, so, ok.
And we'll go to Shakar after that.
Yeah, sure, sure. So, you know, people are talking about exponentiality and scaling.
I mean, music, you know, Suli had raised this point, I think, in previous rooms, like the qualia aspect, right?
The quality, the subjective aspect.
So there is a very less element of scaling here when it comes to what the audience like, what appears fresh to them.
AI is not going to come up with the next Jimi Hendrix or Beatles or some kind of new genre unless somebody is actively...
involvd, som en viss måte av mennesket,
for as far as I know, for now,
because there is a lot of,
you know, what I was trying to get at is,
The unsaid words between the lines.
I mean, instructions and computers depend upon everything mentioned explicitly, right?
But there are certain things that cannot be written anywhere.
And I can go into several examples, but I won't for the sake of time.
So that's my point.
Strangely, by the way...
These are all good points, and we're going to stick to this for a little bit, but we're going to shift gears.
It's still related to Gen I and music, but we're going to get into the how soon.
So, you know, speakers prep up for that.
But I want to go to perhaps Shakar, Xavier.
Maybe you can give us some last thoughts on it.
So, Shakar, go for it.
Ja, jeg tror at vi ser hver industri som har blitt påverkad av AI, vi har hatt den samme narrativen over og over igjen.
Det er en tool, det kommer til å hjelpe meg å gjøre mitt jobb bedre, det kommer kun til å bli litt bedre.
Å, vi bør aldri ranka det i den samme kategorien som mennesker.
Å, bare en skam, vi har brukt teknologi i år, og det er kanskje så bra, eller i det minste hjelpsomt,
for å få oss til det neste nivået.
Jeg tror at den trenden vi kommer til å se, 10x, særlig i musikkspåret, fordi det er absolutt så høyt innkjøp på investeringer for å kunne rapidt ta noen kontent som har vunnet,
sier de siste 100 årene av hitmusikk.
Being able to train an AI model that is really good at figuring out the relationship between those things,
and then actually creating a perspective that allows us to replicate and test as we replicate,
that gets us closer to, I guess, kind of the Netflix effect, right?
Where we're able to use data to effectively anticipate how successful something we've produced is going to be.
Det er ingenting mer å si om det, uten at mennesker er virkelig dårlige i å konseptualisere eksponential utstyr.
Vi har også sett dette i skrivingsspelet, hvor vi inntilfølge at AI er en co-pilot.
Faktisk er at AI tar mer og mer av de tasken, særlig med GPT-4 og kvade-releasninger,
for å kunne automatisere kontentkreasjonen.
Lenten av å ta det kort,
It really is challenging for me to see that not happen.
And arguably, you know, to everyone else's perspective here,
maybe taste is actually a differentiator as to what makes you successful in music long run.
You know, I worked at Pixar, and they used to say that part of John Lasseter,
who's, you know, basically the head creative in the beginning,
there are of course more now,
his primary, like, superpower in the world is taste.
So just something to...
å tenke på i tradisjonal animation,
og jeg tror det er CG animation.
Ok, så vi har fått mye hånd,
men vi må forberede oss
å holde det i flott.
Og jeg vil skifte på det på hvordan.
Jeg har gjort noe med gen-A,
med musikk og kunst,
og de er utrolig utrolig.
Noen av de GitHub-repos
er utrolig intens.
Who here can actually tell us how are these songs being created?
Like literally from a, it doesn't have to get too deep technically,
but let's get a little bit beyond layer one, right?
Everyone's talking about it.
There's always news reports, but no one's talking about
what GitHub repos are being used, right?
What's the processes being used?
How is that Jay-Z song that we talked about?
Who here wants to go through?
It looks like Brian has his hands up.
You know, who wants to talk about the actual process of it?
Maybe Brian, you can kick us off,
and then we'll go to the other hands on here.
Okay.
Thank you. Alright, so, Githubs and other Hugging Face, other types of repos out there, there is new models that come up basically every 12 to 15 minutes.
Så det er egentlig behov av AI å forsøke å trakka på hva denne AI-utviklingen gjør.
I tanke på konsumtionsnivå, gode produkter som er ute der, er Musikgen, tror jeg, den varepunkten for disse verktyg.
I think it's going to be probably the point where people are going to look back and say,
this is where a kid in their bedroom was able to start sampling 40,000 different melodies on their local computer
and came up with the ultimate killer hit song.
That's it.
That's certainly going to happen.
But like we were saying before, it's going to take creativity,
it's going to take taste, it's going to take
feeling the pulse of the zeitgeist at that particular moment.
All those things align, and maybe a great backstory,
and we have a hit.
So, MusicGen is one of the latest, as far as I'm concerned,
the potential of being quite great.
If you put it on a local computer with the proper processing power,
you're going to get some success.
pretty interesting melodies. Now, that's actual musical sound, so that we have to sort of separate a whole lot of this.
The output of MusicGen is not really like MIDI or anything like that, it's actually generating a complex series of synthesized sounds
that sound pretty close to the instruments that, you know, you would...
find familiar if you're prompting it correctly to do that.
If you're prompting it to be highly creative,
you're going to hear sounds that you've never heard before.
Certainly, they're going to sound synthetic,
but some of them sound quite natural.
That's that part.
There's another part where you can just have an AI model,
and there's a few of them, where you can just sample
ideas about lyrics and the potentials of melodies.
and they don't actually produce tones but they'll produce
musical notation which you can gen just you know generate through any other
type of system so those tools are in existence and now
there are commercial tools that are coming up and i'll let
somebody jump in on some of those yeah actually actually you know what's
really funny i actually had chat gpt just produce like midi waves so you can
actually program this in and actually have it
for example do paco bells canon
this is just chat gpt right and give you the frequency to kick a program in so
even that's available to everybody you may be xavier gino than sphinx but xavier i know
you've had your hand up how how are people creating uh these these beats
Through GenAI, what GenAI tools are you using?
Oh, thank you so much for bringing me on.
Yeah, it's actually, just without being too complicated,
it's pretty much a simple text-to-speech,
and then you just give it a flowing algorithm,
where literally it's pretty much like Harmony.
and so that's the simplest way to put it
because if you go deep in depth
obviously you're not LLM
LLM work is
they pre-predict the next words
you're going to say based on
pre-existing data that you feed into
its parameters
the only difference between
the GIN
and again I'm hyper simplifying so please forgive me
so I don't want anybody to be offended
so it literally just simply
is just a simple text file
text to speech
but with a
harmony
it just measures up and down
if you think about it
when you sing
the only difference you are from talking
like hello how are you doing
is going hello
how are you doing
Right? There's no, there's really that little slight difference where it can almost seem like it's just the biggest, grandest thing, but the thing is perception. And that's the thing with the AIs right now, is if you notice, they're different use cases, but they're all essentially the basis of chat GBT, right? Or one language learning model.
And even now we have Orca, I think is going to compete at some point, and make their own music as well. Orca is Microsoft's new gen of AI competing with even OpenAI's GBT. And it's faster, stronger, more capable.
And I think that's essentially it.
So that's really the basis of it.
And what you can do is you can go on to GitHub repo,
download it onto your computer,
and literally test it out.
All you have to do is fine-tune them.
Literally.
There's so many open-source, gen before even music gen.
But right now, music gen is open-source,
download it onto your computer,
fine-tune the hell out of it,
make it whatever you want.
And Google has one too.
Xavier, I'm so glad that you serenaded us in our audience this morning, even if it's only briefly.
It's a welcome respite.
Didn't that sound like a Willy Wonka?
Hello, how are you doing?
We're already interpreting Xavier's music for us, even though it's short.
But yes, Gino, how do you create Gen.A.I. music and form the audience as a Grammy award winning artist?
So like you said, yeah, Music Gem's great.
I think Music LM will probably be the first great consumer product
out of the AI kitchen, Google's AI kitchen.
And as Xavier explained, it's essentially like...
You know, just a sequence to sequence modeling task that generates music like over, I think it can go over several minutes right now.
It's pretty much text to music at the moment.
I think it can transform hummed melodies and stuff like that.
The thing is, we're not really, there isn't really an existing tool for music.
professional musicians, I mean, I haven't met anybody that's actively using it yet.
I think until we see, you know, a plug-in that I can throw in on Pro Tools, I'm
we're not gonna really start seeing mass adoption from the...
Do you know, are you sure that you haven't seen,
I mean, I've seen professional musicians behind the scenes,
like even people who've, you know, won awards and stuff
using this stuff, to be frank with you.
I'm not sure that's...
When I say using it, I mean using it in their, like, work,
like pitching to labels using it.
Not like, oh, I'm just kind of messing around with this
and trying to see...
But how soon is that?
I mean, that can't be that far, right?
Very soon. I think it's very soon.
I think, you know, we'll see a plug-in pretty soon.
I know a few people working on one.
Yeah.
And not just a plug-in, a very good plug-in.
And I think, you know, for that reason, and things like Google LM, or Music LM and Music Gem,
the people that should be most threatened by this in music are the loop makers.
We talk about this a lot. There's a big economy of producers who...
get most of their co-pro credits by just creating loops and melody loops and then they'll send
those loops out in packs to you know people that put drums on like you know london on the
track or or you know south side or i see illa my boy illa's in the audience right now uh illa the
producer you know super dope producer he might be a good person to bring up too here if you're looking for
producer producer uh perspective on this but yeah like these these tools are going to continue to
to evolve um you know there's a lot of ai music and generative music in the web 3 nft space which
i'm not going to sit up here and chill but uh those are exciting ways to you know show
a different space my friends different space for sure but i think music ip and showing um
How you can feed those things into music language models.
We'll be really curious to see how that evolves in the next few months.
Sounds good. Alright, so Sphinx, let's go to you Sphinx, and then we'll go to John.
And then I want to get AGI in this as well, but Sphinx, go for it.
Yeah, I also want to point out Chantal in the audience, she's Grammy nominated as well,
and TheBomb.com. I wanted to say also, I have some friends who are musicians,
And they've enjoyed using Boomi, which I haven't used, but they said that you can create like an actual song from beginning to end, like an original song versus, you know, altering a song.
And just by setting certain filters, you...
and producing it, and then you can save it or reject it.
So I'm actually going to look into that just for fun.
So there are certain things that certain tools out there,
they're different.
Some will let you change certain parts of the song,
others will let you...
You do something from beginning to end,
produce and make the song
from totally beginning to end.
One thing I wanted to say though,
is that this does have implications for copyright.
And the Grammys used...
I'm going to tweet something in the comments
that the Grammys used the guideline
for how to copyright this stuff.
What you use,
anything that's used via AI
cannot be copyrighted.
So...
if you use, if any
lines or anything is used
part of the song,
that can't be copyrighted. If you
write it yourself, that can be copyrighted.
Thanks.
Do send me the names of the folks in the audience who said
have Grammys. I think it would be great to get their perspective.
Chantel's Grammy nominated.
I will.
John, let's go to you, and then we've got GP.
John, go for it.
My relationship here is that I've eaten graham crackers.
but I've never won a Grammy, so I have no musical basis.
But I think we are tremendously underestimating
the contribution of artificial intelligence,
and as we push to AGI, into creativity.
And so I think that we cling to the human composition
as we cling to a baby blanket,
and I think the transformation will be profound and significant.
Now, will that become asymptotic?
In other words, will the creative domain change
bli merre og merre, men aldri krosser utenomstegningen.
Det er det som jeg tenker mye om.
Men for meg er det en stor underståelse av teknologiske kapabiliteter.
Kanskje den siste kvelden, den siste noten, er det som humanitet står for i den komplekse realiteten.
Yeah, John, GP, before you jump in,
GP, yeah, John, I think that's like the fundamental question,
right? I think you teed it up, and we've been kind of bouncing
around it. It's like, is this
just another tool, like, you know, the piano,
or is this like the final platform,
right? I think it's kind of the
the deep philosophical question.
Look, let me just jump into that real quickly.
And Brian and I have talked about this.
When you look at the nature of, let's say, GPT
as part of a cognitive construct,
it's no longer a search engine.
It is an active participant in the cognitive dynamic
of wisdom and knowledge and learning.
And that's what's going on here.
I think that there is some transcendent reality
that's really freaky and it bugs people.
I had a quick counterpoint to John's. I think you mentioned AGI. We don't even know if these systems will create music, because when we are talking AGI, these...
Hey Strangelove, you got a bit of an issue with your connection, it sounds like, so yeah, maybe we'll fix that. We'll go to GP first, and then we'll come back to you. GP.
Come on. Hey, thank you, thank you, thank you. I led to the party today because my stand for free speech,
the preservation of free speech,
i Irland, rann late today.
Ja, I'm going to just quickly go back
to a real quick point about
the essence of what makes a great song.
Where does great music come from?
It comes from the heart and the soul
and life experience.
And it's my view personally,
and I've said it repeatedly,
that we are anthropomorphizing AI
and that no developer can codify love
kan kodifisere hat, angst,
kjærlighet av en annen person, kjærlighet av en barn,
observasjoner av naturen i reallivet,
opplevelsen på siden av en munt,
som du kjærligger,
eller som du ser over en vall og en vista,
kjærlighet av en egen nation,
eller kjærlighet av noe som ikke kan kodifisere,
som en annen person,
eller presens og følelse av Gud,
eller presens og følelse av gloria,
These are all emotions that regardless of the anthropomorphization or the complexity of the algorithm,
the finest music that touches other people's souls,
and that touches the thing that we cannot define by numbers,
that defines us as humans and humanity and our shared struggle,
that it will take eternity.
And by that I mean never.
For an artificial intelligence to be an essence of what it is to be a divine human in a divine universe.
Thank you.
JP, I feel like, beautifully said, and you have generated so much reactions just based on the flood of emojis on the screen.
I want to go to AGI. I know AGI John, you guys both had lots of reactions, but let's go to AGI first.
He hasn't had a chance yet.
So I want to address, so we did talk at the beginning to go in between reactions.
Så jeg vil snakke om å skapa nye lyd.
Så AI er til å skapa nye lyd.
Jeg ser alle instrumentene.
The music instruments that have been created by humanity so far,
the guitar, the piano, and so on,
as points in kind of a latent space.
And what AI allows us is to learn a latent distribution
and interpolate between every instrument.
So AI allows us to cover the full spectrum
of all kind of music possibilities
by creating new sounds that are not possible right now.
So AI allows us to create extra possibilities
New sounds, and that is kind of extremely amazing.
Because we have only a few sounds from the instrument that we have,
and maybe we can have something like a thousand more times new sounds.
And that is very exciting.
And also I want to address the fact that AI technology,
I think AI will have extremely great taste.
In fact, superhuman taste.
And I think AI will be able to compose,
to adjust those sounds
in a way that are extremely more kind of
attraktiv, eller
to vibrate with those sounds.
I think I can train an AI
to do that.
And it will vibrate with
every human being. So we could have
an AI that will learn
from what makes you vibrate.
And to optimize that,
and generate in real time
new sounds, and new
kind of symphony of music
that will be like
the sound that
The orchestration of sound that you like the most, but you always get that in real time.
So AI can get us there.
AGI?
Yes.
That's incredible.
AYC on back, can I respond quickly to that?
Sure, and yeah, we'll go to John or AGI if he wants to respond.
Yes, we can hear you. Go for it.
Strange, can you hear me? Hello? Strange, we might have to bring you down and back up.
You want me to jump in then?
Yeah, go for it.
Can you guys... Go ahead, go ahead, John.
Når jeg tenker på det, er det naturlig å ha en fysisk konstrukt.
Det er vibrasjonen av luft som går inn i håret og i hjertet.
Det er en utrolig liten dynamik.
La oss se på teknologi og AI,
og tenk på hvordan vi kan skapa elektroniske lyder
som kan transmittes direkte til hjertet,
som har ingenting med vibrasjonen av luft.
It's a whole new domain, and it makes music transcendent.
So I think we have to be careful when we look at vibrational sound as the domain of creativity.
I think it's, you know, a lot of opportunities there.
there do we believe that um agi says that uh ai will have super taste that's as first i've heard
it in that format do we believe that's true no i don't sorry i'm back uh sorry about the disruption
uh so let's talk about different genres of music yeah because can it can it come up with the better
country song uh from scratch not from derivatives right i mean yeah you can have edm or um
you know, other genres of music,
but, um,
komposisjon og alle de faktorene.
Du kan få godt nok,
men det er fortfarande en person i loopen.
Det er ikke langt til å gå bort.
Eirik hadde en punkt om AGI,
at når AGI kommer,
vi har den beste musikken aldri.
Vi vet ikke.
Om disse blir kundsjøs,
eller kundsjøsmaskiner,
noen tider i fremtiden,
Which will be, in my opinion, long time. I mean, how we are supposed to know whether they like music or not. Music is very inherent sort of activities to humans, maybe to certain animals, but yeah, I don't know.
Ok, who has a different take on all this?
I do, but I wanted to say one way to look at this is you've got to take our human emotion at the moment out of it and ask,
what is AI able to do that we cannot do?
And one of the things is that it can go through vast volumes of data and analyze things.
som vi ikke kan gjøre som mennesker.
Så en måte som jeg kunne se det som skapade,
av en AI som kom med noe som var mer av en hit
eller mer populært enn noe som en menneske kom med,
var av AIs måte å gå gjennom alle disse volumene
of material that's out there.
All of the music that's ever been produced,
whether it's been a hit or not, right?
Everything.
Everything that's been streamed.
And go through all the ratings.
Go through all the hits, all the views.
And...
bylge noe, kreere noe, baserat på det.
Så du må på en viss måte,
igjen, jeg sier ikke at dette er en god ting eller en dårlig ting,
men er det mulig? Det var spørsmålet, er det mulig?
Jeg skulle ha sagt, selvfølgelig, pga denne kapasiteten, vi har sett dette i
medisinen, vi har sett dette i radiologi,
Absolutely, I think it can be done in music.
This is a math situation.
I'd love to jump in on this, thanks, if I can.
Yeah, sure, go ahead.
And argue with you.
No, I'm just kidding.
First of all, GP, I need you to drop an audio book, GP.
I just want to listen to you talk all day.
So do that for me, thanks.
I think, like, we can't compare...
hva lenguagjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjegjeg
Not at the speed that AI could do, but songwriters and producers and creators have been trying to make TikTok records for labels for five years.
And all the music is, again, not through literal AI, but through people trying to replicate what they think a viral hit is going to sound like.
It's created this...
en vast katalog av skitig musikk med ingen substans,
og vi ser denne effekten i musikkindustrien.
Det er en utgift...
Jeg må visslå på det, for med ting vi gjør,
som musikk,
men hva om bilder, for eksempel?
Gå inn i bilder, midtjern, stabil diffusjon,
og sånt utav det.
Kan du si at vi har kommet til en punkt
Would you agree that we're at least getting to a point where we can create generative AI visual art?
you know absolutely without the humans and we're so why why wouldn't we yeah we're agreeing
apply to music because because i don't want people to think that i don't think that music
ai music is going to get good enough where most people will not be able to distinguish between
the two in some capacity again so your argument is that for the masses it's going to be like that
but the next like beethoven ninth symphony or something not even beethoven but the music
that's being like you know 85 percent of music being consumed today is cataloged
Nobody is listening to new music for longer than their attention span can...
And the reason for that is the music feels, in my opinion, it feels tingy.
It feels like metal in your mouth after a while.
There's no real gravitas to it.
And again, I don't want to sound like a fucking, back in my day, we had to buy the CD at the store.
And I'm not trying to get into that.
It's okay, we respect our elders here.
Yeah, yeah, yeah.
I don't want to age myself,
but that, you know, that feeling of music
with real investment, real emotion behind it,
and this is kind of what GP was getting at,
is that although AI can scale and make great music,
and it'll sound great,
at the end of the day, in my opinion,
and I don't, who fucking knows?
Who?
En kort punkt, labeling musik, kategorien, for anyone is very subjective.
Rap musik for somebody might be inspirational, for somebody it might not be.
So when we label these systems, what are we going to be labeling them as?
Det er en annen spørsmål, denne subjektiviteten.
Ok, så å vende tilbake litt på alt dette.
I Beethoven-tiden, Mozart-tiden, det var mye krammusikk som ble utgjort.
Vi minns bare dem fordi det var greit.
Shakespeare minns det fordi han var greit, men Christopher Marlowe,
His plays ain't that great.
I mean, they're not bad, actually,
but we remember him only because he was a contemporary of Shakespeare.
So I guess the question I ask is,
sure, there's a lot of crap being created by humans and by AI,
but will the eternal music,
like the music that we remember generations from now,
can that be created by AI eventually, if not today?
I think the answer to that has got to be yes, right?
Does anyone agree with that?
Not without humans.
I agree, and...
Jeg tror at tidsløs musik, og det som Gina sa,
og definitivt det som GP sa så poetisk,
er at tidsløs musik er basert på hvordan det dyrt påverker deg.
Nå, jeg tror at det viktigste vi må gjøre
er å kvalifisere lyssningens betydning til musik.
Og Gina har sort av togt på dette med det vi kaller fastfood musik,
er at om du ikke har en dyrt betydning til musik,
It's background noise.
And I've established about three levels of commitment.
And when you had to spend $14, $8 for an entire LP,
and then you had to savor every track and your red liner notes,
and you made some sort of commitment into, say, Pink Floyd's Dark Side of the Moon,
is a good example, you talk to any older guy,
or gal who has gone into that album when it came out,
you realize that they were going on a journey
with that performer through all of those songs.
So it was a deep, deep commitment to that music.
I would say a lot of classical music is a deep commitment.
You go to a live classical performance
and you will see some people cry and get deeply emotional
over a song that they've heard probably a thousand times
but reinterpreted.
Unfortunately, because we've created throwaway music or fast food music,
there is no deep commitment to it.
It's background noise, and you don't get to hear the catalogue
or the journey of that particular artist,
unless that artist seems to resonate with you as you're growing up,
and you sort of follow their journey,
or there's just something that's...
unusual about their music that this has drawn you in.
And this is the problem with the palette and the canvas that the artist has today.
They literally have the attention span of the hook,
and almost all songs are basically hooks at this point.
Basically TikTok. TikTok has basically changed the way we write.
Absolutely.
We write songs, right?
So if you don't have a commitment to music, right?
right if you're not really savoring it it's background noise to something that you're doing
that's fine but that's if that's music for a lot of people then absolutely ai is going to
supersede that let me tell you something i do in my lab i use human telemetry to try to predict
music that i will need to listen to based upon my galvanic skin response
my heart rate variability my breathing patterns things of that nature i can
pop this on anybody, and I can create generative music in real time
that is going to either enhance or deter you from a specific emotional state,
and it works quite well, and we can empirically show that.
That's the future of a lot of background music.
It's going to be to either pull you into an emotional state
or pull you out of an emotional state.
And so the AI is going to self-select for that.
And that's going to be a powerful thing.
It's going to be new for a lot of people.
It's going to be hard to deal with.
and musicians are going to try to think about what does that really mean,
and there's still going to be a place for musicians in that,
because timbre and melody and things like that are going to play into it.
But that's kind of what's going to happen.
Personalized music, 100%, is where we're going with a lot of this.
Can I jump in there real quick?
Yeah, actually I was going to go to you, GP,
but also Dr. V, we brought you up because you're new to the space,
so just ready after GP, but yes, GP, I was going to go direct to you
since you kicked off this set of debates right now, so yes, go for it.
Great, thanks so much. And to what Brian says, I think it's super important, and it intersects with Sphinx's comment on Gino.
devolving everything to a homogenous
shade of beige
is not going to create
uniqueness within either
original thought
or complex or relative truth.
I think Gino said, you know,
it tastes like metal in your mouth.
It feels
like it's something not quite right there.
Just when you walk into a place that's
not familiar or
is uncomfortable.
Now, derivatives,
as per what Sphinx is saying, there is the
epiphenomenon, of course, coming out of sampling
of derivatives. Let's look at the
unexpected success of Dido
when Eminem sampled that track
of hers as the background.
We may get some epiphenomenon
out of the use of
AI-generated music, where a particular
chord
or a particular set of notes
resonates, and that resonation
then allows the inspiration, as Brian
says, of somebody in IRL
who lives for music. Remember,
music is a vocation. Create
kreativitet er en vokasjon, det er noe som er essensen av noens søl, det er en kalling.
I en veldig mundan video som var laget av Henry Kissinger, Håfenløker og Schmidt tidig,
de, selv om i deres forsøk å vare alt til enes og nødvendig, og å definere alt med vestlige tenkninger,
And I think that's super important in terms of music as well.
Remember, Within You, Without You, and Love You Now,
or Love Me Now by George Harrison using the sitar,
are some of the best music ever produced by the Beatles.
The relativist cultural norms and the relativist sampling of other nations' music,
where Hoffenlacher, Schmidt and Kissinger mandate that all AI should be based on Western values,
So just to wrap up real quick,
we are
in a situation
where people's units of attention is so
low that, you know,
I write long-form tweets, some of them
20 lines. People are annoyed
that it takes 20 lines of their attention
to understand something like neuropharmacology,
neurostimulation,
but more importantly, what John mentioned,
both
Brain-computer interfaces
in the militarized and civilian context,
if you look up Dr. James Giordano,
are significant existential threats
to determining what we are,
and our perception and cognition,
and our own position within reality.
And the idea of piping
samples derivative sound
based on knowing what resonates
with a particular person
is actually a weapon, not a piece of music.
In order to manipulate mood,
as I think Brian mentioned in background,
musikk, er ikke egentlig noe som er bra for de mennene.
Det er bra for bottom lineen av reteileren som prøver
å pulle deg mot en sikkerhet i forhold til å kjøpe en sikkerhet.
Og jeg vil rappe det her.
Play O Fortuna by Karl Orff.
Play it, or listen to the Ru Orchestra do it
in the central plaza of Rotterdam,
and the tears will flow from your face,
because it's based on Karlsdorff's deep understanding
of what it is, the essence of humanity.
And with that, I have another point which I can't read,
but I think that's enough.
The essence of music is humanity,
and the essence of AI is anti-humanity.
I understand that, but...
One thing, GP.
Can I say one thing to that?
Hold on, one thing.
Yeah, sure, of course, I'll let you jump in,
but I do want to say Plato did say, right,
that he wanted to ban all the artists and the musicians
in the city-state, in his sort of perfect republic, right?
So, Sphinx, I want to go to you.
Yes, you did, and real quick, Eugene,
the reason he said...
I just want to say...
GP, jump in.
Ja, jeg er dårlig, han er åpne ut for meg.
De menneskene som blir tilbake til siste, om aldri, blir kreativene.
Kreativene i vårt samfunn er de menneskene som vil holde oss menneske.
De menneskene som har en søl som har en inegsorabelt lenk til noe som er ute der,
som ingen annen kan putte en hånd på eller ganske oppnå.
Kreativene, stande framfor en tripte,
du vet,
ehm,
by eponymous botch
are standing in front of the Mona Lisa
that quintessential
human element
that no amount of sampling
or looking at back data
or blogs or legacy data
can give you that spotlight
spark of beautiful inspiration
after a vocational calling to be a
creative, because you're wired that way.
And I'm not being anti-AI, but I'm being
this pro-human. Have faith in the humanity.
The essence of humanity.
As they called it, a speciesist,
right, in that Tucker Carlson thing. Look, Sphinx,
I want you to close it out, but before that, Dr. V
has been waiting in the wings, and then we do have to
close out the space to get ready for the next one.
So I know we got a lot of people wanting to chat, but Dr. V,
welcome to the show. What do you have
to say for us? Yeah, you...
Thank you for having me up here. Thank you, Mario, and all the hosts.
You know, the last speaker talked about the Mona Lisa. That's right in line with my point.
I just thought I'd offer the perspective of someone who can't help but to sort of sit around and write songs all day, every day, for their entire life.
And, um...
That's what you see in the Mona Lisa, is you see someone who sat around and painted every day for their entire life.
And that's a, it's a perspective, and it's an experience.
And these machines, they can produce anything, but they only have the perspective and the experience that we bestow upon them.
So, you know, you can get into the arguments about who's creating what, and we have like the legal argument about
Can you copyright it?
And you have this sort of moral...
Hey, Dr. V.
Dr. V, I really like what you're saying,
and also your profile just looks awesome,
but your audio is a bit odd.
So, speaking of audio,
so maybe if you could fix that,
we'll go to Sphinx,
and then we'll come back to you,
hopefully, if you're able to fix it.
You sound actually kind of like an AI robot,
a little bit.
So, if you can fix that up, that would be great.
Sphinx, what do you have for us?
Well,
Ja, så jeg elsker alle de tårerjørkare språkene, de er flotte, og jeg har ikke nødvendigvis å avtrykke dem.
Men spørsmålet var, er det mulig for AI å skapa...
something better, more popular,
than what a human can create.
The answer is unequivocally, yes,
because of AI's ability to go through the data.
They are doing it with radiology right now,
with cancer treatments right now.
That is...
Ein simpel ja.
Det betyr at
mennesker er tilgjorda
eller bør bli tilgjorda?
Nei.
Det betyr ingenting om Mona Lisa?
Nei.
Men frågeteertes om det var mulig. Hvordan kan du si at det ikke var mulig?
Så jeg har fred mot de som har tål på det, og de som har talat om all det om mennesket.
Det er fint, det er fantastisk, men det er ikke fråget.
Vi må leve i realitet her. Er det mulig?
AIs teknologiske mulighet til å gå gjennom stor del av data,
infinitum
is just unmatched
by humans. So
again, that's not just
songs that have
gone platinum. That's anything
that's ever been
out there. Any TikTok, any melody,
anything, and
rank them. So
is that there now?
Probably not, but can
that be there in the future?
I would say yes. I would bet,
you know, if I had to bet the house on it,
I would say probably, if I had to say
is it possible, I'd probably say
yes. And that's not advocating
for it. We're betting the species on it.
i noll mån, ikke sant?
Hold on JP, hold on en sekund.
JP hold on en sekund.
JP hold on, please.
Det er ikke att jeg advokaterer på det,
men det er bare å si at det er sånn som det går.
Og jeg har absolutt fred mot det menneskelig.
Mange av mine vennar er kunstnere og skrivere.
Så jeg er her i dette rummet,
ikke advokaterer på det,
men er det mulig?
Jeg tror det er mulig.
You see, the absence in your thesis is that you were absolutist about your statement that it will.
And you can't be absolutist because you're mixing up something very fundamental about AI.
No, I said it's possible.
No, you didn't. If you look at the recording, you said absolutely it will at the start of your...
I said it's absolutely possible, David.
Absolutt klossisk. Det er semantisk, men hvem responderer?
Definisjonet av å redusere det som er, jeg tror,
de som er kjærløse, de er ikke kjærløse.
Det heter filosofi, og det er derfor teknologene ikke er
to be left alone with this AI business,
and why it should be a multifaceted,
multidisciplinary conversation.
But you're making one serious error in your thesis,
over and over when I hear you in these spaces.
You're mixing up the difference between
information, knowledge, and wisdom,
The availability of data does not give you the oversight that you say it does with respect to the inexact social sciences.
And thumbs down or thumbs not from the people in the audience who have their vested interest in AI.
The simple fact is that you cannot, in data, infer from data, the inexact social sciences information.
that so much dominate our landscape.
Our landscape is dominated by psychology,
psychiatry, anthropology,
economics, and so on.
All of which are inexact social sciences.
Yes, they can be predictable
if you have enough legacy data,
but Sphinx data does not equal knowledge.
You bring up a good point, actually.
How do we judge us?
song? How do we judge a song?
Right? You do bring up a good point, because how do
we judge a song? Who's to say
this song is better? This song is
the best ultimate song, this is
the best song that I've ever heard, right?
So there is that element
of it, right? So is this going
to be O Fortuna? I'm guessing
no, there's only one O Fortuna
if you ask me, right? So
there is that, but you know
it's not
is it
It's possible, but the question of how do we render judgment on that, is it going to rise to the level of something like O Fortuna? Well, that's a whole different discussion. That is a whole different discussion. And do we, do we, uh...
Er vi til å få det til å stige på den nivåen,
med tanke på at det var AI?
Det er en helt annen nivå.
For meg, det var noe som var skrevet av mennesker,
rises to a different level than something that was purely AI,
even if it sounds great.
I'm sorry, but it does,
and I just want to finish it by bringing it back to the Grammys.
The one thing that I think is the flaw with this Grammys thing is,
I really think they just need to have a couple of categories for AI,
because they have a minimum requirement of 20% of human contribution.
Imagine, you have...
People there, 20%, so that means someone can have, an artist can have 20%.
Who's the arbiter?
Yeah, it's a right back.
Who's the arbiter of taste?
30% or 90%.
Who's the decider?
And the thing is, exactly, and how do you know?
But if you have certain categories where you say, okay, these are AI produced, these are AI, this is our AI category, that's the thing.
I feel like we're just getting started actually in some ways, but we do actually have to wrap up.
But Illa, the producer, literally just came up, so maybe you could take it home for us, but we do have to end after this.
But definitely more to come later. Illa, go for it.
Yeah, I jumped in because I saw the Grammys outlined some new rules about AI and
I think AI is a great tool. I use it a lot in the business world.
I've had meetings or conversations with my peers in the studio,
and we are all excited about using the voice AI.
For sometimes we'll be like, yo, I want to make a beat for Beyoncé.
And as a producer, if you're a real producer, you're a songwriter.
So you'll write the song, and you can be like, oh, I'm going to have Beyoncé sing the vocal.
And when somebody broke it down to me like that, I was like, oh, shit.
For I'll be writing songs for the future, I'm a little baby in my head, I'm this one, I'm that one.
So if I can deliver a full song to a label, that gives me an edge.
But I do have to agree with GP when it comes to the human side of things.
Most of the great sounds, when somebody comes with a new sound, it's a mistake.
It's not something that somebody can use the data and give it to the data and this, that and the third because it was a mistake.
And most of the fucking mistakes turn into whole sounds that shift the whole industry in the direction of music.
That's the great thing about music, it's not like anything else because the fucking, the people who know the least,
who have the most lived experience, they literally start to dictate the sound and the vibe and the culture.
Like, it's crazy. Like, I'm partnering with Scott Storch, one of the best music producers ever.
He is trained by ear.
And when he goes in with somebody who's classically trained, they're blown away.
And they're like, what the fuck? I didn't think about that chord leads to that chord.
And it's because it's a feeling. It's the goosebumps.
I think what you're talking about is like happy accidents, right?
Which you see a lot in art.
So we definitely want to have you on for the future.
And there's so many hands here. I would love to go to them.
But I do got to say this before we wrap.
IDC, which is Mario's company, incubates and accelerates AI and Web3 companies.
We...
partner with bcs and funds work with their portfolio companies zero sorry equity but zero
cash so if you're interested do dm mario and his team we'll get a call organized and also we're
going to be starting to do shark tank style pitches have been doing them in the crypto
and ai spaces people seem to like them so if you're startup or portfolio company you know
hit up mario for the process and also don't forget to subscribe so um with that i mean
we got a lot to go i feel like we're literally just getting started
But this is a big topic.
A reminder, every Tuesday and Thursday,
we're doing these AI Town Halls,
12.30pm Eastern Time.
Please do join.
It was a lot of fun to chat with you all today.
We clearly have more to go,
so thanks everybody for participating.
Take care, everybody.
And by the way, Imran Khan later this week.
