The Large Language Model Race with Pete Huang, Founder of The Neuron

Google said, we've got a new large language model. That's awesome. And then OpenAI went, Hold my beer and watch this.

That was crazy. That was absolutely crazy.

Right?

And look, I think the whole thing around the one person business that can generate 10 million a year, 20 million a year, a billion dollars, whatever sort of fancy thing you want to put on, it is possible.

Hey, welcome to the Next Wave podcast. My name is Matt Wolf. I'm here with my co host, Nathan Lands. And today we've got an amazing guest for you. Someone that loves to nerd out around AI and keeps his finger on the pulse of AI just as much as we do. Today we've got the founder of one of the most popular AI newsletters, the Neuron. On the show today, we're talking with Pete Huang. Thank you so much for joining us, Pete, how you doing?

How's it going? Everyone good? It's good to be here.

Can you give us the quick rundown of what the Neuron is, what type of reader you have, what your goal with each issue is?

So the Neuron is a daily AI newsletter. It's filtered through the lens of knowledge workers, non technical audience, small business owners. And we give you updates every day, Monday through Friday. It's a very, very quick read. We do two stories and a bunch of links just showing what's going on. And what we try to do is filter through the lens of like, really the average person. There's a lot of, you know, talk around the technicals of AI, but if we zoom way back and make sense of how does this actually manifest in, in your work, in your business and how you make changes in how you do work, what does it actually look like?

Yeah, that's amazing. I mean, congrats on the success of that newsletter. That is huge. I have to ask you one question because it's the question that people ask me the most. So I actually want to turn it around on somebody else. Is there anything in the AI world that you've kind of heard of, it's in the pipeline that maybe you're aware of or you've seen already that has you really excited right now.

Oh, it's the agents for sure. Agents for sure. Because if anything, because my, my lens is on what running a business looks like in the future. And today right now, a lot of business owners, enterprises, small business, everyone included, they're a little bit confused by AI because they're like, okay, well, the press and the media, they're promising me these transformative things. I should be able to trans like run a completely different business because of AI today. And that's not actually happening because the reality of AI, at least in the ChatGPT of the world as it is today, is it's automating what we call like the bits and pieces of work, right? It is. I'm a marketing coordinator and I can save 10 to 30 minutes of my day every day because of this small little thing that I now created a prompt in Claude or ChatGPT to do and then now it's giving me a first draft and I have to spend that time doing it. That is not as transformative as it is for that one particular person. For the business owner, it's not like holding wholesale transformative, right? Agents, I think have a much bigger promise along that front because you are now going to unlock end to end automation for an entire process, right? So if you have an agent chain that can go from creative concept to a brief to a proposal, to all the iterations to all the back and forth, like really all those things and handle all those pieces of work. Now you're talking about for a large corporation, for example, a set of AI agents could take on even 10%, 20%. That's huge, right? That is a huge amount of work. And those people can now do other things. And now you're looking at real multiples of creativity, not this 10 to 20% thing. You know, on a per person basis you're talking about entire sort of like chunks of work at a time. I think the whole thing around the one person business that can generate 10 million a year or 20 million a year, a billion dollars, whatever sort of fancy thing you want to put on it is possible, right? But you need agents to get there. I don't think you sitting in front of ChatGPT every day is going to do it. That is again giving you 50% leverage on your time. We're really looking for a hundred X, right? And I think agents represent that. A hundred X.

Using only 20% of your business data is like dating someone who only texts emojis. First of all, that's annoying and second, you're missing a lot of context. But that's how most businesses operate today, using only 20% of their data. Unless you have HubSpot, where all the emails, call logs and chat messages turn into insights to grow your business. Because all that data makes all the difference. Learn more@HubSpot.com.

So when, when you say an agent, let me just kind of clarify a. So I know that we're, we're talking the same language. But B, the listeners know what we're talking about. When you say an agent, you're talking about, you give it, you go into like a chat window, like a chat GPT, you give it a task that you want it to complete, but instead of just giving you a single response back, it will keep on working on that task and iterating and iterating and possibly even using external tools and APIs until it reaches the end goal that you initially prompted it with. That's exactly right.

And so I think to me the easiest way to explain it is there are three parts exactly. All the things that you just said, Matt, today, right now in ChatGPT, it's like very directed, which is like, I want you to write a blog post about this. And it's like a very specific thing. Like it's very self contained, there's no extra research involved. But an agent has the ability to plan, it has the ability to reason, it has the ability to use other tools. And so instead of saying write a blog post, you might give it something a little bit more abstract and say, come up with, come up with and execute the marketing strategy for my business. And it'll go through everything around. Okay, like I'm going to research what does a marketing strategy mean? What does it mean to have a good marketing strategy? What does it mean to execute it? What are the sort of best practices for your line of work? What are all the pieces and sort of like compile everything and as it's sort of reflecting on its own work, it's reasoning through the problem, it'll then go into execution phase and start to use tools. Right? Which is okay, Now I know that I need to create an Instagram post, so let me go draft the thing and then go call the Instagram API and upload the post every week, week, every two days, whatever it is. And so now all these things that were normally done by a human, going from chatgpt to clicking over to instagram.com, whatever it is, is now replaced by code, right? It's, it's, it's code. And, and these large language models that have just like developed enough capability, right? And, and so this is sort of the prospect of GPT 5, llama, 3, et cetera, to do all these things. And so this is getting very close, honestly, to what it feels like to have an AI employee.

Right.

Like in some cases, I know previous jobs that I've had where the only value add that I've brought to the table was the fact that I could think because I have a human brain and I was clicking on software all day. Like I was just moving bits and pieces of information from one software to another. Like an agent could totally do that. You know what I mean? Like an agent can go and build a Salesforce dashboard for you, it can go run SQL for you, it can do all these things. So there's like this whole new wave of capabilities and startups and products that are all agentic in this way that'll come hopefully in the next couple of years. Again, my lens is on just like the future of business, right? Like, I think that's going to really fundamentally change how businesses are run.

Yeah, it feels like probably GPT5 will be the first time where agents kind of start working. Because like, you know, I wrote this big Twitter thread back, you know, like a year ago about baby AGI, right. It went really viral. I went on TV and talked about it and it was really promising. But like, you know, all of those, you know, you, you'd boot them up and get them working on something and it would just end up spinning in circles because it wouldn't be able to complete the task. It would basically get lost, you know. And so I think as soon as you get where they can actually complete tasks and come back and show you the results and, or even go on and do the next logical step, that's going to be such a huge unlock for businesses and entrepreneurs. And the rumor is that with GPT five people have seen demos and it can not that it's perfect. I'm sure it's going to like take many, many years to get amazing, but apparently, yeah, you can tell it to do some task and it just goes off and does them and come back and reports to you.

Yeah, yeah. And, and Nathan, that getting lost thing I think is the unlock, right? Because right now, even if it's 2% off, 3% off, you take point like 97% times. 97% times. 97% times however many steps you want to, to complete. And like at some point there's like a guarantee that it will not complete it correctly. If you have enough steps, right, It'll. That number will converge to zero. And so the, the real part is like how do you get that 97% correct to as high as possible? Like 99.9999% or even a hundred percent, theoretically, I have no idea. Right. But that's the biggest gap. There are a couple of platforms out today, but that's just how it is, right? Like it, it is just not there. Yet. And so once you get past three, four steps, it just starts to crash every single time.

I think everyone needs to be thinking about how they're going to use agents when they get good because it's going to create so much competition that if you're not using them, you're going to be so far behind. If you think of like the Peter levels of the world who just like crank out like, oh, I made a new startup and now it's making $20,000 a month. And like, and then next month he does another one, right? He just keeps doing this somehow. Imagine when he has agents where he can literally, it's like, here's my idea, I have a social following so I can kind of jump start it. I don't have to worry about like, no one's going to try it. At least a thousand people will try anything I do. The agent goes and makes the landing page an hour after I thought about it on the beach and, and now I'm testing the idea. It didn't work. Okay, well tonight we'll try some more ideas. It's just like, that's going to be amazing.

Totally.

The week that we're recording this kind of the biggest thing going on in the AI world is all of these new large language models dropping. We got Gemini 1.5 now became available publicly. We've got GPT4. They just released a new April 9th version that they claim is a major update, but didn't really tell anybody what they updated about it, just that it's a major update. We got Mixtral 8x7B that just got released, which I believe is open source. The 8x7B was open source, so I believe this one is as well. And then we also got the announcement From Meta that Llama 3 is most likely coming in May.

So.

So lots of large language model news going on right now. What is your take on large language model news? Do you see like one of these companies, like being a clear winner? Do you think there's place for all of these large language models to all coexist? Like, what's your take on it?

It's a race. I mean it's all a race to build the best models, right? And so of course you have your GPT4. There's that upgrade that happened this week. Everyone's sort of already talked about GPT5 though, you know what I mean? It's like when, when both Sam Altman and Satya Nadella are Talking openly about GPT5, like it's coming, like everyone knows it's in the works and it's just a matter of time, I think rumors. If I have this correct, guys, it's like summer, I believe that's GPT5, is that right?

Yeah.

And so it's just like all these models are great. I mean like Claude 3 I think was, is amazing and I, I still think Claude 10 today, chatbot wise, is the best option. So that was a big upgrade over GPT4. We had Gemini 1.5, public preview availability for all developers. A lot of advancements there that they claim. Gemini 1.5 beats GPT4. You have all these, this sort of second group of the mixtures, et cetera, that are trying to make their mark and find a position in the market. You have llama 3 with meta. But all of this is sort of looming with the sort of shadow of GPT5 coming, you know what I mean? Like, everyone's comparing it against GPT4, but GPT4 has been out for a year already. And that was like the old GPT4, you know what I mean? And so you factor in like this week's upgrade with some math capabilities and then you just have to ask yourself like, okay, if everyone's just now catching up to GPT4, are they just going to get blown out of the water by GPT5, you know what I mean?

Yeah, yeah, yeah. I've been thinking too that like, that GPT5 is probably way ahead of everything else we're seeing. And some evaluations came out for GPT4, the new upgrade, maybe a few hours ago, and it's showing that actually it's outperforming Claude again in coding. And it's kind of weird too because they released it and it's so vague and like you said, yeah, GPT4 has been out for a year. They're still releasing updates to GPT4. They're not calling it 4.1 or 2. It's kind of confusing and I kind of wonder if they're doing that on purpose just to like, you know, they're so far ahead, it's like, oh yeah, compared to the old model. And we'll slightly like, you know, Claude gets ahead. We're always just barely better than the best model until we release the real thing, we're actually focused on which the thing they've been really focused on is GPT 5, not 4.

When you look at it from a product, marketing and PR lens, OpenAI knows that they're had, right? Like, let's just be, let's put it out there, right? Like they're they're pretty much ahead. And if you've seen their behavior over the last year and a half, every single time Google has launched something, they're always ready. And in fact, I remember, if you remember last spring, every time Google made an announcement, Google or OpenAI made an announcement that was like 10 times better. And everyone was just like, it created this atmosphere of just like, what is Google doing? You know what I mean? And Google's trying, they're like trying to release new things or prepping their launches, their PMs, their PMMs, like going nuts. Trying to craft a campaign and a message that works and then opens like, boop. Let me just go like drop something that beats you really quickly and then like, oh, here's another one from the backlog that just like beats you. Right? And you know, so this one is ready. And to your point, Nathan, like, this thing that dropped on the April 9 update for GPT4 wasn't actually a real launch. You know what I mean? From a product marketing perspective was not a real. There was no blog post, there was no specificity into like what actually improved other than like, I think it's better than math, you know, I mean like, it got a little bit better.

Yeah.

When there were somebody who said like major update or major upgrade and there was other was like upgrade or update.

It's like it was like that was.

That was different messaging from OpenAI.

Totally, totally. And so it's just like they, they had to have been sitting on this for so long that they're kind of like, okay, we know Google Cloud next is coming. We know they're going to do something. Google Gemini 1.5. Yeah, let's just be ready and just like even a half baked release, an upgrade of whatever, like wouldn't even need to be specific about it. We'll just drop it and then people, you know, then we'll be talking about it on this podcast.

Right.

Versus Gemini 1.5. Right. So it's like it's just a masterclass, I think, in OpenAI flexing on the rest of the market and knowing very clearly what position they're in.

Yeah, I mean, Paul Graham always said that like Sam Altman really shined in terms of strategic thinking and planning. And so I think a lot of people are not realizing that like he's like one of the masters at being strategic and planning things out. I don't think a lot of the people think like, oh, they're just kind of floundering right now because GPT5 is not out. I was like, no, I think this is like part of the plan. And then when it comes out, it's going to be so much better than everything else. And I think Claude and other ones are going to. Oh, wow. They're like way behind. That's my personal belief.

Even going back to Pete's point about just sort of OpenAI always sort of having something queued up to be ready. Wasn't it the same week that Gemini 1.5 was announced that OpenAI dropped the Sora video? I'm pretty sure that was like the same week Google said, we've got a new large language model. That's awesome. And then OpenAI went, Hold my beer and watch this and then put out the Sorrow videos.

Yeah, yeah, that was crazy. That was absolutely crazy.

Right?

And look to Google's credit, and maybe this is where we start talking about Gemini a little bit. But like, that release when Google dropped Gemini 1.5, that was a pretty big deal. Like, there's a lot going on with Gemini 1.5 that matters a ton. And it wasn't just like this, let me catch up to GPT4 thing. It was like a pretty landmark statement. But it made for a crazy Newsweek, right, Because you had Gemini 1.5 and everyone's just like, oh my gosh, now videos a real thing with Sora. And like before, like those clips weren't like, the video generators weren't actually doing a thing. And then now it's like a real sort of thing to watch out for. I remember that it was, it was absolutely wild. But I'm curious, Matt, if, if, if we step back here and kind of just go back to Google Cloud next and all the sort of announcements there, Gemini is definitely, I think, one of the headliners coming out. 1. Can you just help me level set the Gemini conversation? Just like, what is it? Like what, what is the big deal about it over something like GPT4? And let's talk about like what's coming with, with this model and all the other rest over the next next few months.

So one of the biggest things about the Gemini 1.5 model when they announced it was that it was going to have up to a 10 million token context window. But I think the release version that was going to be made available was going to have up to a million. But then the version that we actually have now I think is like 248,000 or something like that. So the version that we have available to us right now inside of like the Gemini platform, I think is like a 258,000 somewhere in that range. I don't know the exact number off top of my head, but it's somewhere in that range. So it's the largest context window we have available. You know, supposedly it's much better at coding. I actually haven't tested it myself for coding yet, but the context window I think was really, really that big leap. And in the article that they shared, they did the needle in a haystack test, right, where they had a, you know, 700,000 word article, they put a single sentence that was unrelated to the rest of the article somewhere inside of it and then asked questions about it. And the needle in a haystack test found the answer like 99.9% of the time. So it did really well on the needle in a haystack test for like the, like the retrieval, augmented generate generation tests. And it also has that huge context window. I'd say those are probably the biggest factors.

Well, I saw a tweet saying that you could upload files up to 2 gigabytes. And I don't know if I've actually, I haven't confirmed that, but like, that sounds incredible. Like if you can actually like upload incredibly large documents to this thing and have it evaluate them or summarize them or, you know, be able ask questions of that document, that's, that's huge. I'm not sure how big ChatGPT is.

But I feel like, yeah, I don't know the upload number on ChatGPT, but I know Bill of all also shared something either yesterday or today around Gemini 1.5, where he uploaded entire like Interview and had Gemini create the timestamps for it. So it actually listened to the whole interview, made timestamps and he said they were super accurate. I've tried doing that with ChatGPT and A. You have to sort of export the transcript first. Right. If you're using Gemini, apparently you can just upload the video and it'll just pull the data from the video chatgpt, you have to grab the transcripts, copy, paste them into ChatGPT, tell it to write you timestamps. It will actually get the sort of notes for the timestamps, right? But it'll have the timing way off. Like the actual like time that it wants you to jump to will not correlate at all. So that's another big leap is supposedly it's getting better at actually, you know, the multimodality. Multimodality so far has mostly meant we can upload text, we can upload images, we can Talk to it. But all it's doing behind the scenes is transcribing it. So it's still just kind of text, right?

Yeah, there's video understanding. So. So just to put some numbers out there, right? So with Gemini, commonly cited, although it sounds like the release is a little smaller than this, but the 1 million token context window, like one of the comments that we get a lot is like, okay, tokens, like, why did you invent like a new unit to count things by? So token is like three fourths of a word, basically, it's like three or four characters. And if you, if you have a million tokens, then you should assume that's like 700,000 words in like one lengthy conversation. With, with Gemini, you can feed 700,000 words into Gemini. Right? That also roughly equates to one hour of video, about 10 hours of audio in terms of understanding there. So to your point to, to Bill of All's experiment, right, it's like, man, you could just toss an hour video in there and it'll have full understanding. And this is where that needle in the haystack test gets really important. So stepping back for a second, the Neil and Haystack test, as Matt was describing is, yeah, you just bury some piece of information just like randomly in this huge amount of text. Like somewhere in 700,000 words, you just drop in that random thing and just see if it can actually detect it. The reason this is important is when Anthropic first released Claude, like an upgraded version of Claude, they had this thing where like, oh, like we can do like 128,000 tokens, which at that point was like, pretty significant, right? Like compared to ChatGPT, which is in the five digits. And then someone ran a test that, this kneeling haystack test. And then it turns out that like, the more information you were giving Claude, it would get worse and worse at finding the needle in the haystack. Um, and there was something, they, they, they found something where it was just wasn't performing correctly according to spec. And so this is sort of like a, A, a test to show. It's like not only do we have the ability for you to just like submit 1 million tokens worth of stuff, but, but it actually works, right? It's like it's not going to drop it. It's not going to progressively get worse. It's like this sort of like a uptime guarantee or like a quality guarantee of your, of your, of your token context window. So there's so much here, right? Like, I think with one hour, like 10 hours of audio. There is so much you can do with that that, you know, we were trying to come up with use cases or like products that could exist. It'll be things like, okay, now there's like some speaking coaches, AI powered apps, they'll do speaking, training and coaching. It's like as you're on a call or like you can upload a video and then it'll analyze the transcript and then pick out the ums and the us and all the repeat words and all these things and tell you like, you know, you should be using, you should be speaking with more authority. You're, you're, you're hedging too much or you're using ums and us too much. Now imagine it's actually analyzing the video of you talking so in like a public speaking setting and giving you feedback on your stage presence. Right. Because actually you can see the video now and like what you're actually doing in the video. So it's like these types of things that I think are really interesting and I'm just so excited about all the apps that are getting built now that again, developers have wide access to this, public access to this. You can imagine that a lot of builders who are like, oh man, like there's this thing that I've been wanting to build and like it requires video access and now we have video with Gemini and so we get an example, kind of get to experiment with that. So I'm just super excited about all those ideas that are going to come out well.

And also there's Hume as well. H U M E where it actually understands the context. I don't know if it actually reads your face right now, but it listens to your voice to try to understand whether you're sad, happy, angry. Right. It tries to understand the emotion in your voice and supposedly down the line it'll actually look at your face and try to, you know, pick up micro cues in your face to figure out the emotions as well. So that even takes it to another level. You were talking about like being on stage and it giving you feedback about your stage presence. Imagine if it can tell when you're more nervous, when you're more excited, when you're more happy and even give you feedback on that stuff on stage.

Yeah, yeah, right.

Like that's, that's coming, I'm sure.

Yeah.

Nathan, is there anything you're excited about with, be it Gemini or Llama 3 or any of these models that are coming out, anything that you're particularly looking.

Forward to on the open source side, you know, Elon Musk is saying that GROK is going to be as good as GPT4 or better very soon. So to have that kind of open source version where you can kind of ask it whatever and not worry about how it's filtering. You know, I don't know if you saw some of the stuff with, with Claude, how, how they're filtering what you say to it, like where they actually, they don't actually tell Claude, they don't tell the LLM what you're actually saying. Even they pre filter what you type. And so they don't, they don't even give your actual message to Claude.

So basically you write it a prompt, it goes and rewrites the prompt for you in a way that it thinks is probably going to get you a better response, but also in some ways may potentially censor you. Right?

It is like making sure that it's, you're saying things appropriately or whatever.

Right?

It is doing that. So it's like, and it's not even showing the LLM that that's like their solution is like to literally change your own words. I think it is good to have a kind of an open source alternative where like, okay, if OpenAI and all those go way too hard on that side, that there's an open source alternative where you can ask it things and not have to worry about that.

We've talked about all these large language models from Google, from OpenAI, from Anthropic, from all of these places. And Grok didn't come up once until just now when Nathan mentioned it. Yet Grok has an open source model that anybody can use if they want. I mean, I don't know if most people could run it on their computers with the size of it, but anybody can access it however they want. And supposedly he's about to drop a model that is as good as GPT4 that is going to be widely available to users of Twitter. So it's just kind of interesting to me how much GROK gets sort of left out of the conversation when it comes to large language models.

Well, it's just not as capable as far as I understand it, right. Like in open source you have the meta stuff, with Llama you have Mixtrol that's doing really well and, and then you have like the, the likes of Grok, et cetera. But then when you look at the surveys, for example, I, I think it was Andreessen Horowitz that did a survey of enterprise buyers and practitioners for generative AI. Basically what they were saying is that the, the, the three or four players that, that any sort of practitioner is considering using model wise is OpenAI because they have, the leading model is Metalama because it's open source and, and it is just like the only sort of option basically. Like why, why would you give up anything on the capability front when you just need your best, the singular best open source model and then you have Google and Claude or anthropic models similarly because of sort of the latest capability reasons. Right. And so from that perspective, it's sort of like why would you consider even, even Mistral stuff, even Grok stuff if it's not going to hold up to meta llama? Like I think that's frankly the answer. I don't use Grok because I just don't think it's as good. And until it's good, I see no reason to actually switch over.

I know that like the fact that Elon Musk is behind the project and actually really pushing it, I think long term he is going to catch up and surpass a lot of people. I do believe that.

One other interesting thing about Grok, right, is that it does have access to Twitter. So it's trying to use like all the conversations on Twitter as part of the context. But we also saw last week that that can be an issue, right, because last week there was a issue where Grok reported some news that wasn't even real news. They just pulled from like memes on Twitter, thinking it was real news and then fed it back to somebody as if it was real. So you know, maybe feeding all of the information off of Twitter isn't that great of an idea after all. But that is one of the sort of benefits of GROK as well is that Twitter.

That's fascinating. I never even thought about that. So you can like make up memes and then Grok thinks it's real, then actually gets spread to real news.

It reminds me of, I live in San Francisco and there's a lot of these like shenanigans that happen with self driving cars. So like people, people will take like a traffic cone and like put it on the sensor and prevent it from moving. Yeah. A more violent one unfortunately is I think it was like two months ago people literally blew up a self driving car.

Chinatown, right?

In Chinatown here. Yeah, yeah, in Chinatown. And it kind of feels like this, you know, I mean like I could totally see if Grok ends up being very good, widely used and it's very well publicized that pulls from Twitter, that there is constant instant like meme campaigns to throw it off, you know what I mean? And just to get it to say whatever. So I don't know, like, never underestimate the Internet, you know, I mean like there are going to be shenanigans abound with, with this type of stuff.

Why do you think there's such an aversion a, a hatred towards the self driving cars? Is it like people that are Uber drivers and Lyft drivers feeling threatened? Is it people that are just anti AI? Is there like any feeling of why that's happening?

My feeling is if you take the same person and put them in any other city that's not San Francisco and has this exposure to attack whatsoever, the arrival of self driving cars is like magic, you know what I mean? Like when, when I first got access to to Waymo, for example, in its beta and I hopped in, everyone immediately around me was like, oh my gosh, what is that? Like, that is so cool. They were taking pictures and everything. And I think there's just something about how much experimentation there is in San Francisco that for some reason the newfangled thing, like having a newfangled thing is just baseline. And in some respects you get desensitized to just how cool it can be. And as a result, I think you just get tired of just how much change there is. Or like you feel this narrative of like big tech kind of invading you because you just have to deal with this change all the time. And it happened when there were these deliveries, delivery robots that I think it was doordash. I'm not too sure it was these food delivery companies that were piloting, you know, sending these little rovers around to go pick up your food and then you wouldn't need a person anymore. You can just like have this whole thing. And then all of a sudden like, you know, pedestrian safety became an issue, right? It's like, oh, what happens if someone trips and like falls over one of these things? And these are like good concerns. I think these are like the right questions to ask. But just like this overall feeling of just like the first instinct you have when you receive something new and innovative. I think you get reminded a little bit of all the other times where it's failed or it's been annoying or been a nuisance. The scooter wars were a very big example, like line scooters, bird scooters, those invaded San Francisco, polluted the sidewalks. I think that was a little bit too much by the end of it. But you kind of just get reminded of that kind of era A little bit. I think that's, that's where that comes.

From when you're there. Like, it's so expensive to live there in San Francisco, and the people who are in tech are making crazy amounts of money and the people who are not, they're barely scraping by in San Francisco on average. And so when I was there, like, like when I first moved there, it wasn't that bad, but it seemed like every year there was more and more hate towards people in tech in general. And I think a big, a big part of it is like that difference of income and things like that. You'd even see it like in like tech journalists, like a tech journalist started not liking people in tech and it was become this really odd dynamic there where it's like you got tech entrepreneurs and people writing about them who don't like them. It's like this is very odd.

I feel like a lot of journalists seem to hate like tech AI specifically. If you see anything in like a big media platform these days, it usually is taking a more negative slant towards the AI tech than it is the positive slant. Right. So like people like us who are creating content, that's more on that positive spin of where all of this is going. Yeah, we seem to be having our heyday right now where people are paying attention to what we're talking about and, and they don't trust the big media anymore. So, I mean, I think there is a path to be optimistic and still get views and still get clicks. I just think the media is so used to this, like, old way of doing it of like, drama sells, you.

Know, I do think with some of these conversations, as much as optimist as we are, I do think it's important to be sympathetic to some of these points of criticism. For example, this week, along with all the LLM stuff that's been going on, music is like on top of everyone's minds, right? So, I mean, today specifically, there's the Udio release. There's like another one coming out of YC called Sonato. I think it was last week you had 200 of the top artists sign a letter saying, hey, look, like, we generally think AI is good for creative work and, and it can be built in a way that supports creative work. However, we like strongly, strongly, strongly encourage that tech companies, investors, etc. The entire ecosystem working on AI do not build AI that replaces or diminishes humans or their compensation that are in the industry right now. The exact achievement of that, like, how to actually strike that balance is like very Very hard, in my opinion. Like, there's. This is going to be coming and going, like, with all of this.

Right?

But if you were an artist on that petition, signing that letter and your latest understanding was Suno or one of these other tools, I frankly, I don't even think it was Suno, frankly. Like, a lot of these musicians or a lot of people outside of our bubble aren't even exposed to this. And it's like only truly the early tinkerers on, on your Discord servers, et cetera, that are really playing around with this. Like, you can't be happy seeing the Sonato from today or Udio today, which are even steps ahead of Suno, right? You can't. I don't think that that makes it happy, even. Even if from Silicon Valley's perspective, it does not. Like, I think everyone here is sort of aligned on this general idea of like, okay, well, like the intent here is to create tools that promote creativity, that democratize access, that democratize output. That all sounds great, right? But if you are in the industry and you're watching all these developments happening, I think it's very fair to assume that like, that person is just like, they're coming for us. You know what I mean? Like, that, that's sort of the mindset and it's like, like if this is happening this quickly and in a matter of a couple months, the models have gotten this far, then what's next? Like, the only thing I see left remaining is for this to fully replace me or, or, or outright replace my job. That can't feel good. Right? And I'm very sympathetic to that because to. I feel like to, to brush that conversation, that narrative under the rug, I think is missing the point. I think that we're gonna have a much harder time convincing people to one, use these tools and then two, convince them that the proliferation of these tools is net good for society if we don't at least acknowledge that sort of thinking. Right?

Yeah.

So I think it's. It's important to be nuanced in these things. It's just there. There are a lot of. When I notice the animation communities, the gaming communities, the music communities that are all deeply creative people, and there is a lot of anti AI sentiment. Right. I think, like, all three of us have probably been subject to some, some amount of like, pseudo cancellation on Twitter or YouTube comments or whatever it is as a result of this. Matt, maybe, I don't know, you've been, You've been pretty. Your comment section always looks pretty positive.

You're not seeing it all, then I.

I do think, like, you know, like, we have to acknowledge it. Right? And it's not just a matter of like, oh, these people like that are just going to get left behind. It's like, no, like, I think these are, like, pretty legitimate concerns. And we just have to figure out a way economically, legally and technologically that we can sort of just have everyone come along. It's not gonna be perfect. Right. But I feel like there's gotta be a solution there. I'm an optimist about that. At least that there is sort of a way forward that that kind of fits everyone in the future.

That's fair. Like, I think this will, you know, there will be some jobs that will dramatically change. Some will disappear, new ones will appear. Most jobs will change in dramatic ways. And there will be some aspect of like, okay, how do you make sure the artists get compensated? And. And it's tricky. I think there is going to be a big legal battle, right? Because, like, I'm sure OpenAI has trained on so much data that maybe you would say they shouldn't have been able to do. And I don't think they would have done that unless they had talked to lots of lawyers and decided that, okay, this is actually a gray area because when you're training on this data, since this is such a new thing, you can say that this is basically the same way that a musician would learn from listening to another musician. Did that person steal that person's work? You know, it's like, you know, and so I think that's really the argument they. They plan to make, and they're just not wanting to publicly state that yet. But I do believe that is the argument, that legal argument is coming because it sounds like they've actually scraped YouTube's videos possibly, and that's how they've, you know, built up their AI video.

Yeah, I mean, I, I take that same approach. I try to be very empathetic, sympathetic to those concerns. I try to take them all into account when I am sharing the news. I always try to kind of talk about both sides of the equation. You know, I really do feel for a lot of these people, like, when it comes to music, I don't think it should be okay to just go and create a song in the style of Drake that sounds like the beats Drake would make, sounds like his style, sounds like his voice. And you just took him out of the loop completely. I don't think that's cool. Honestly, I don't. I don't think people should be allowed to do that. But saying that there is a very, very fine line. What if somebody has a voice that sounds very similar to Drake? Are they not allowed to make music sound similar to Drake? Like, there's some weird, like, nuances, gray areas there that are really, really hard to deal with. But I think at the end of the day, a lot of it comes down to the sort of monetary aspect of it, right? If somebody is like, who dedicated their life's work to becoming the best musician that they can possibly be, the best game developer they can possibly be, the best artist they can possibly be, and now people can create stuff at the level that they were able to create it. Well, how do they sort of still make money? How do they still have a livelihood? Like, it all comes back to the monetization. And I think something needs to happen in the AI world where there is some sort of, like, Spotify style split. It is insanely complicated to figure out. And I do not envy the people that have to figure that out. I just get to be here on the sidelines, armchair quarterbacking. But, you know, I do think there needs to be some sort of way to compensate the people who maybe the data was trained on. I think if I was a musician right now, and I, you know, you can see my background. I've got guitars, I've. I've played in bands before, I've recorded, I've gone on tour. I've done the whole music thing in the past. But if I still played music as often as I used to, I would actually be really excited about what's going on because I can be like, man, if my voice is popular, I can do the Grimes thing and license it and have a million people going out there and creating music on my behalf, and I can somehow earn from those people using my voice and using my style of music. To me, that gets me excited. But I also understand the flip side of that coin.

I think that has to be coming because, I mean, will. I am invested in Udio. That was announced like two hours ago. And so, you know, alongside Andreessen Horowitz and a bunch of other great investors. And I don't think he would be investing if they didn't have some kind of plan for revenue share with artists. I could be wrong, but I think. I think the fact that he's involved shows that there's probably. They're probably not just me saying F you to all the artists. I think there's going to be some kind of plan to compensate people.

If you look at Udio, and you look at Suno, you can't go into either one of those apps and say, generate a song for me that sounds like Katy Perry. It'll just say, sorry, can't do that. That other one, Sonato. I actually did a live stream and I was playing with it and just out of curiosity, on the live stream, I was like, generate a song that sounds like Blink 182. And it's like, here you go. We use the lyric, the voice style of Mark Hoppus and the guitar style of Tom DeLonge to make this sound like a Blink 182 song. So Sonata actually does let you plug in a band's name and generate a song of like, whoever you want. I have a feeling that feature will probably get phased out.

I don't know. I think they need to nail that because that's like, consumer wise, that's where you'd actually get traction. Like if you can't say artist name, you're not, you're not actually get like a major, you know, mass adoption of these tools, I don't think.

But consumer wise, that's where it's at. Legal wise and not wanting to get sued up, asked by every music industry person in the world.

Yeah.

I don't know if that's the approach you want to take. I don't know.

Yeah.

What is your guys's take on Udio and Suno? How do they sort of compare to each other? I've been playing around with a couple of them. I think Udio is very clearly the best one. But what have your experimentations led you to?

Personally, I'm just having fun with them. Like, I love them. I think they're both so much fun. I've made probably like 20 potential jingles for future tools. Right. Like write me a jingle about why Future Tools is the place to go to learn about AI and the latest tools. And it makes like really fun songs and I work them into my YouTube videos. For me, it's just kind of like a fun toy. I don't really see a whole lot of practical applications for it yet, but I do think they're really, really fun to play with. Yeah, right. I think Udio probably overall is better for the most part, but I did give it a prompt, like generate some like, electronic dance music for me. And Suno did that style of music better than, than Urio did. So I think, you know, genre dependent. But I think as a whole, Udio is probably better.

I think the, the overall direction that I'm excited in is one for, for you to refine what you just generated. Right. And so instead of having this be like a one shot sort of, it just generates and that's pretty much all you get. And you just have to keep on trying over and over again to go in and change little aspects of it. Maybe there's like individual parts of it that you can, you can tease out the little, the individual stems and edit them. And I think that is generally the direction that I also think is helpful to turn this into more of like a creative tool, more than an outright sort of prompt to full song type of thing. Because Matt, to your point, like, I am a little bit questioning the use case beyond doing your, your jingles for YouTube videos. I've also done this for, for jingle purposes. And like jingles aside, like the jingles are not a big market. You know what I mean? Like, they're, they're not gonna like, like make the, all these companies enough revenue to compete over the long term. However, if you're to build more tooling around it, which is like, I think this is maybe Sonato's approach here, which is can we allow a little bit more directed iterations over time. That is interesting, right? Like that, that now is opening up a new canvas for, for new artists to explore and creatives to really get in there and master the individual pieces. And that can introduce a new way of creating music. So generally excited about it. But yeah, I'm very curious to see once we've solved that sort of like it can create a 30 second clip of like pretty coherent, pretty good music end to end. Then it's like, then what, right? Like then, then what level do you go to? Which direction do you head in?

I saw a Twitter post X post earlier today of somebody saying like, I'll be impressed with AI music when an AI song tops the charts or something like that. Is that like the goal of some of these tools? Are they trying to like get you to make songs that's gonna be top of the charts? Like, I don't know, as a human myself, yeah, maybe you guys can relate. But as a human myself, I enjoy listening to music created by other humans. Like there is something about listening to like a Led Zeppelin song and knowing that like Jimmy Page played those notes on his guitar and that was a very complex thing that he played that makes it that much more pleasing on the ears. Just knowing the human element was attached to it. I don't feel that sense of accomplishment when I listen to an AI song that I generated. Right. I don't really like, I Might go share it on Twitter because I think it's funny, but I'm not going like this is the next, you know, dance club hit or anything. I feel like the AI music right now, what it's best for, like a real world use case is actually take the lyrics out of it completely and use it as like a sound bed for your. Your videos or for your podcast or things like that. But once you add the lyrics into it and you're trying to make your own, like, pop songs with singing, I actually don't know the real use case for it right now.

Was it musically that started off as and then turn into Tick tock? Yeah. And so, like, you know, I think you could have some kind of new music app where like, yeah, if somebody could just generate any kind of song as part of some kind of social app and then. And then that thing could go viral, you know, and that could become a top chart thing. Like. Like some. Some songs from TikTok have, like, become very popular because they were big on TikTok. Right. I don't know.

Once again, I do think it's sort of genre dependent.

What I get excited for is, do you guys remember that clip of Maggie Rogers when she was still in school and Pharrell came and listened? It was like a music production class of some sort of. And it was like she blew up. Like, that's exactly how she blew up was off of a video recording of Pharrell listening to her music that she produced all by herself. She's saying just everything was hers. Right. And I think Pharrell's commentary was just like, I have no notes because this is so authentically you.

Yeah.

Like I can't come in and. And sort of give you any sort of advice or criticism or editing off of this. And what I think about is if you took Maggie Rogers then and gave her a set of new, like these AI powered tools, how much more powerful of that experience would that experience have been? Right. And it's not about to say like, oh, Maggie, you only give us the lyrics and you sing and sort of AI will do everything around you and sort of. It'll one click, sort of like spin you into different genres. There are probably a lot of smaller technical things that she probably did not do on the first pass. Didn't know how or just didn't see it creatively or whatever it is because she's young and. And sort of early on in her music career. Getting you that first draft of it sparks the creativity, if anything. Right. And you can. You can have her be what she was in that moment, which was the vibe curator of the song.

Right.

Like she brought the spirit of what was the message that she wanted to send, the, the emotional state that she wanted to conjure. But the rest of the technical pieces, maybe there's help that we can give the next Maggie Rogers, right? That would be cool. To your point, around the Pop Factory songs, you know what I mean? Like, these are all the same set of musical artists that are writing songs for each other, right? They will literally sit in a room and be like, okay, today we're writing for Shawn Mendes and like, you know, eight of them will get together and just sort of all throw songs together. One of them will hit and it'll become Shawn Mendes's next song.

Right.

There's something about that that feels a little bit different perhaps. And, and you know, it is a much more mechanical process, I suppose. But when I think about the essence, like the spirit of like what we're trying to do with these creative tools, or at least the, the companies doing them, I would hope is to find that next student, right. And to give them, you know, that much easier of a time to express the spirit and the concept of what they're doing. That feels very powerful to me.

I do actually have like a couple like sort of last questions I want to ask. One of them is, do you have any advice for business owners in like the current landscape of AI with the existing suite of tools that are. What do you tell business owners who maybe are hesitant to get into using AI? Where, where would you send them first?

The num. If I had one line. One line. It's to give your entire business access to one of the paid plans of ChatGPT and Claude and just let him figure it out. Honestly, it's $20 per user per month for most businesses. This is more than affordable, right? Like just one month, 30 days. And be serious about it, right? Like give everyone access and tell them, I want you to figure out where you can be more effective. Arm the people with, with a mandate to go figure it out because people will again, given the nature of the product. Right? Again, we're, we're focusing the. A lot, A lot of the benefit comes from the bits and pieces of all these different processes. Who knows them better than your employees that are dealing with these processes every single day, right? This is not a top down thing where we're saying, okay, we are now going to apply AI in like our marketing department. It's like, no, there are all these like little things that there are enough early adopters at your company that will take the time and experiment and go figure out if you let them and give them the tools and the, the support from leadership to do it right. Host the hackathon, encourage sharing, reward the behavior. Right. So, so typically a lot of the, the patterns that I've seen is there's just like one person in this one department that spent their off hours, one or two hours every now and then playing around with a prompt or like a zapier integration or something like that. And all of a sudden it's saving the entire department a hundred hours a week because of how many people they scaled the entire operation to maybe again something as simple as everyone here. I sent the, I sent the prompt over email copy and paste the prompt from the email, go to chat GPT or Claude or whatever it is and then use this when you're doing this thing right. That's already impactful and it had nothing to do with hiring an outside consultant or some other vendor or whatever it is. It was all just harnessing the inherent sort of interest and attention there is around AI and grasping onto those early adopters who like to figure it out and giving them the environment for them to do it for you. Right? So that's the main thing. You'll be surprised at how many businesses have not taken this. Like I've talked to a few enterprises that like from the outside looking in, they really should have had some movement on this in the last year or so and they just haven't touched it at all. Not one bit. Because they're, they're, they're just like, okay, this is cool, we get it. But like ah, like we're fine. You know what I mean? It's like no, take it seriously, right? Like go through the exercise, get, get it in people's hands and it'll come, It'll come. So if again that's, that's the one thing, there's a whole bunch of other ways you could do it but like this is the most dead simple one. You already have the people and the resources, just giving them the environment.

Thank you so much Pete for, for hanging out with us today.

That's awesome.

You are non mayor Pete over on X. The newsletter is the neuron. Where should people go to go learn more from you and to, you know, get inside of your ecosystem?

Yes, find me on Twitter non Bear Pete, go to theneurondaily.com for our newsletter and then find me Pete Huang on LinkedIn.

Thanks again for joining us. This has been such a fun conversation.

Thanks, guys, for having me. This is so much fun.