- Episode AI notes
- Investing in early machine learning companies that were ahead of their time can lead to success with proper timing and patience, as seen in investments like Base 10 and Inflection.
- The evolution of AI is characterized by waves of human capital, transitioning from foundational models to infrastructure experts to B2B app developers, with future waves focusing on consumer apps and enterprise adoption.
- The trend of funding for large AI models peaked in 2023, leading to potential shifts in venture funding towards more unique and innovative companies in 2024.
- Software democratizes capabilities in the AI industry and beyond, with a focus on collaborative workflows catering to individual contributors within large enterprises.
- The modern AI stack presents significant opportunities for enterprise growth, enabling new applications and workflows, particularly beneficial for small businesses targeting SMBs.
- AI automation can lead to significant growth for small businesses with minimal employees and presents growth opportunities for startups catering to SMBs by providing AI tools or tailored products and services.
- Automating data analytics processes through natural language queries faces resistance in some segments, with differing levels of interest in automation based on organizational structures and incentives.
- Founders in the AI industry are dedicated but should maintain realistic expectations of AI’s impact on reducing engineering team sizes in businesses.
- AI is expected to impact certain verticals in SMBs, displacing jobs and expanding markets, although the current impact on areas like code generation may be overstated.
Efficient communication with executives involves brief calls and texts, emphasizing the importance of simplifying communication and the rising demand for communication automation. Time 0:00:00
Early investments in machine learning and compounding effects Summary: The speaker reflects on their experience at Greylog, where they invested in last-generation machine learning companies that were ahead of their time. They mention the long journey of some successful founders, highlighting the importance of timing in the success of innovations. Discussing their investments in Base 10 and Inflection, they note that these bets have started to pay off recently. Despite challenges in pushing the boundaries of language models, they believe that the growing interest, excitement, and funding in the ecosystem will lead to compounding effects and further innovation in the field.
Speaker 2
And so I think part of my experience was at Greylog, we invested in, I was there for a decade, we invested in some last generation machine learning companies that were a little too early And some of the most interesting founders today, they’ve been working on it for a long time actually, right? Like so you both know Lucas from weights and biases and like he tried once and he was too early and nothing worked. And suddenly like you feel it when everything begins to work. And so for me, like I invested in base 10 in about three years ago now, three and a half years ago, and like that really started working this year. We were investors in inflection, it was pretty well named company, right? But it was one of the first like, I guess aggressive bets in foundation model companies besides opening eye. And it’s not, I think you can make some arguments about it getting harder to push the state of the art in large language models from a just cost of scale perspective or obvious ness of where The next set of data is going to come from. But the compounding effect of like going from a very small group of people paying attention to a huge ecosystem being really excited about the innovation funding it, I think it compounds From here. I think we’re pretty early. What do you think a lot?Waves of Human Capital in AI Evolution Summary: The growth in AI is unfolding in waves of human capital - the initial wave was led by those who developed foundational AI models, followed by a wave of infrastructure experts laying the groundwork, and now transitioning towards B2B app developers. The future waves will involve consumer apps and eventually enterprise adoption, each wave staggered in time and marked by varying levels of technical competency and focus.
Speaker 1
Yeah, I think there’s a lot of room for growth and I think chat. It was kind of a starting gun for most people, including, you know, there’s basically no real enterprise adoption so far. So that’s going to be a big wave. In general, I think about it as five ways of human capital that have come through and will come through the first wave was just like AI needed. The others who are working on LLMs and one of the foundation models wanted to do apps. And so that’s why you had no one start character. He was one of the authors in the original French former paper that developed. Folks were Argan from perplexing, et cetera. These were all people working at Google or OpenAI basically or in some cases, Facebook. The second wave was a mix of, I’d call it like nerds and I include myself in that category, right? It’s like the hardcore dev and infra people and some of them started the companies earlier. Like Sarah mentioned around base 10, but some things were also more recent together or some of the other companies that are really now providing part of the stack for these models. And we had an infra wave and that’s still happening. And there’s tooling like what Ray Truss is doing now and others. The third wave, I think, is going to be B2B apps. And I think a lot of the people who, you know, heard about chat GPT 13 months ago or whenever it was probably quit their jobs six months ago. And then they took a couple of months to figure shit out and now they’re starting things. And so I think like we’ll see an app wave on the B2B side. And I think consumer is probably a little bit behind that. And then I think the fifth wave will be enterprise in terms of actual adoption. And so I just viewed as subsequent waves of people who are staggered in time relative to a mix of technical competency, focus, product versus engineering, thinking, et cetera.Opportunity for large models and venture funding Summary: The opportunity for large models like foundation models may be closing for startups, as the trend seemed to peak in 2023 with significant rounds of funding for companies like OpenAI and Anthropic. The focus for venture funding in 2024 could shift away from large models to more unique and innovative companies. The venture capital landscape often follows precedent, with more funding pouring in once a trend is established, but the most intriguing companies are often the first of their kind, making the direction of future funding unpredictable.
Speaker 3
I have so many directions. I want to take this in. You’re both saying the number of things I wanted to touch on. Just to go back a little bit before I sort of break off from some of those questions. Sarah, you talked about, you know, it’s becoming somewhat predictable about where the next foundation model opportunities will be based on the data sets. I think that’s, I think that’s how I understand what you, what you said earlier. Do you both feel like the opportunity for large models at this point is kind of baked? Like it felt like 2023. There were all these huge rounds, whether it be, you know, anthropic or inflection or, you know, these huge open AI rounds. Um, is that opportunity like is the window for that opportunity pretty much closed for startups at this point? And if so, and maybe you just answered this a lot, like where do we, where do we see most of the dollars going in 2024?
Speaker 2
It’s really interesting because so much of venture, like there’s so much capital that comes behind having any kind of precedent. Um, and, but some of the most interesting companies are really like first of their kind companies. And so, um, I think this is really unpredictable. And the reason I bring up like venture following precedent is there were not a lot of people trying to fund foundation model companies as a category until it became very broadly known, Um, how rapidly attach EBT revenue was growing and the excitement around it, right? And then I’d say, um, uh, you know, a lot of nighDemocratizing Capabilities through Building Software Summary: The significant revenue growth in consumer companies like Midjourning suggests that the opportunity for AI companies might lie in the consumer sector. The traditional notion of an enterprise company powering AI for the industry may face challenges. Building software is crucial as it democratizes capabilities, not limited to AI but also seen in the evolution of SAS playbooks. Companies like Sigma have succeeded by adapting and creating new workflows that are collaborative, appealing to individual contributors within enterprises.
Speaker 3
And I would argue it’s also midjourning, right? Which is doing, you know, reported hundreds of millions in revenue. Also, I would say that’s a consumer company. Like what does that tell us about the opportunity for AI companies? Maybe it actually is in consumer right now. And and also is is the notion that there’s going to be an enterprise company that powers AI for the rest of the industry, maybe as the way that people thought about open AI, maybe a lot more Challenging. How do you both think about that?
Speaker 2
Alen, why have about like why, you know, building more software is so interesting and important is that it democratizes capabilities. And that is not just AI, right? Like you said, like there’s been a playbook for SAS. The SAS playbook has changed, right? I got to live through a few generations of it at Greylock and like I think a lot and I are also both investors in Sigma. Sigma was a sort of first of its kind company, but it was also part of a generation of companies that grew by the now like coveted playbook. If you can call that, that’s pretty unique to figure out for each company of getting individual contributors inside a large enterprise to adopt a new workflow. That is collaborative. And I think if it’s that or it’s Canva or if it’s any other productivity company, like defined really broadly, that could be different types of writing.1min Snip Key takeaways: • Software democratizes capabilities, not just AI. • Playbook for SAS has changed over the generations. • Getting individual contributors inside large enterprises to adopt new collaborative workflows is key for growth. • Productivity companies encompass various areas like writing, graphic design, video production, application building, and small business enablement.
Speaker 2
Alen, why have about like why, you know, building more software is so interesting and important is that it democratizes capabilities. And that is not just AI, right? Like you said, like there’s been a playbook for SAS. The SAS playbook has changed, right? I got to live through a few generations of it at Greylock and like I think a lot and I are also both investors in Sigma. Sigma was a sort of first of its kind company, but it was also part of a generation of companies that grew by the now like coveted playbook. If you can call that, that’s pretty unique to figure out for each company of getting individual contributors inside a large enterprise to adopt a new workflow. That is collaborative. And I think if it’s that or it’s Canva or if it’s any other productivity company, like defined really broadly, that could be different types of writing. Graphic and user experience design, like video production, that could be Pika or something like Hagen application building and engineering, different types of small business enablement. I think that’s going to be a really big category.The Evolution of AI Technology in Enterprises Summary: There is a significant opportunity for tools to introspect, manipulate, and scale AI models due to the complexity of the technology. Large enterprises consider AI technology a crucial primitive, requiring the development of a new stack around it. The modern AI stack presents a more compelling opportunity than the modern data stack as it powers numerous new applications. While enterprise opportunities exist in services like free play and brain trust, workflow-oriented business user applications might lag behind in adoption.
Speaker 2
And so I think there is a huge opportunity now for tools. Like there’s this analogy that people use that’s like alien technology left here, magic. Normally that type of analogy is noise the heck out of me because it’s like very imprecise. But in, in this case, it’s actually like somewhat apt because we need, we have this technology that we cannot explain very well. We can explain mechanically, but for any given input or, or, you know, what data is driving a particular generation, we have a lot of work to do to be able to introspect and manipulate And scale these models. Like there’s going to be an entirely new stack around this new primitive that large enterprises really think is very important. And so, um, I know lights, the doesn’t end in a master here, but there’s been so much talk from investors over a five plus year period about the modern data stack. And I think the modern AI sticks actually much more interesting because feeding all these net new applications. Um, and so I, I think that’s probably the nearest term like, uh, enterprise opportunity, um, see if services like free play and brain trust and such. But, uh, I think the, um, more workflow oriented business user applications are going to be a step behind.
Speaker 3
Let’s get a little bit into this. Um, you know, you talked about prosumer. You talked about SMB a lot.The Value of the Modern AI Stack for Enterprise Opportunities Summary: The modern AI stack presents a significant new opportunity for large enterprises as it supports new applications and workflows. Investors have been focusing on the modern data stack for a while, but the modern AI stack is more compelling for feeding new applications. This stack creates opportunities for enterprise services such as free play and brain trust, and could lead to significant growth for small businesses and startups targeting SMBs by viewing AI as a valuable resource that can automate tasks previously done by people.
Speaker 2
Like there’s going to be an entirely new stack around this new primitive that large enterprises really think is very important. And so, um, I know lights, the doesn’t end in a master here, but there’s been so much talk from investors over a five plus year period about the modern data stack. And I think the modern AI sticks actually much more interesting because feeding all these net new applications. Um, and so I, I think that’s probably the nearest term like, uh, enterprise opportunity, um, see if services like free play and brain trust and such. But, uh, I think the, um, more workflow oriented business user applications are going to be a step behind.
Speaker 3
Let’s get a little bit into this. Um, you know, you talked about prosumer. You talked about SMB a lot. I think it’s really smart that you’re, you’re looking at payroll by given industry. That makes me think you’re, you’re kind of viewing AI almost as a resource, like you would capital or, or people. Um, and if you imagine that AI basically automates or does what, what people previously could, um, and you think about this prosumer SMB opportunity, you could see these small businesses Getting very, very big and very, very successful, uh, without having to employ that many people. You could also see, uh, new startups that are building businesses and products specifically for SMBs.AI Empowering Small Businesses vs. Tools for Small Businesses Summary: AI automation can lead small businesses to grow significantly with minimal employees, while new startups can cater to SMBs either by empowering them with AI tools or by providing products and services tailored for their needs. The potential lies in both scenarios, with opportunities for small companies utilizing AI to achieve significant growth as well as for tools enabling small businesses to expand. However, the speaker cautions against overestimating the immediate impact of AI, suggesting that widespread transformation through AI adoption may still be years away.
Speaker 3
Um, and if you imagine that AI basically automates or does what, what people previously could, um, and you think about this prosumer SMB opportunity, you could see these small businesses Getting very, very big and very, very successful, uh, without having to employ that many people. You could also see, uh, new startups that are building businesses and products specifically for SMBs. Which of those two do you think there’s greater opportunity? Small companies that are leveraging AI to do really big things or tools and products and services that are enabling small companies to do really big things.
Speaker 1
You know, if I had to guess and I’m probably going to guess wrong, um, I wouldn’t be surprised if a lot of mid journeys revenue is actually like medium companies or, you know, there’ll Be some small business just where it’s used as like a creative tool inside. But I, I do think there’s more, it’s, it’s kind of more of a B2B tool. And I don’t mean like big enterprise by that. I just mean, I don’t think that I wouldn’t, I wouldn’t be, I wouldn’t be surprised if there’s a creator class that does a lot of stuff on it. But then like much or most of the revenue is like the person who needs to put simple images together for a slide in a company of whatever size. This is sort of the prosumer comment. Um, so I just kind of don’t want to overstate some of these things. I think the whole AI is going to make every company one person and we’re going to rebuild every app using AI instantly and all those stuff. I think that kind of stuff is many years away.Challenges of Automation in Data Analytics for Different Organizational Segments Summary: Automating data analytics processes through natural language queries presents challenges as some analysts resist automation due to complexity concerns. While smaller and larger customers show interest, there is a segment resistant to the idea. The appeal of automation may vary between analysts and their bosses depending on the organizational structure and incentives. Small and medium-sized businesses, for instance, may be more receptive to automation due to their focus on core business activities rather than auxiliary functions like marketing or asset creation.
Speaker 2
Yeah. I, um, I’m an investor in this company called SEEK. And they’re trying to do automation of a bunch of the data analytics job. Uh, and so, you know, how do you ask you a question in natural language of some structure, data source in your company and be snowflake or warehouse or database and get the right answer Back, which is, um, not just a natural language, the SQL problem, but a much more complex one. And it’s really interesting, like where they’re getting the most traction. Uh, there’s some uptake amongst smaller and larger customers, but there is definitely a segment of the, um, let’s say like analyst audience that does not like this. Right. They’re resistant to the idea of automation and they’re. And like, you know, over a very small training period, we’re like, Oh, we can do 70% of this 80% of it, but that may not be that compelling to them. It might be very compelling to their boss. Right. And so I think a part of it is also where in the organization, you are selling to what the incentives are within that organization. Um, and SMBs are interesting because like they often don’t want to do a bunch of the other functions, right? They love some part of their business. Um, and not necessarily like the marketing or the asset creation or anything else.Dedication of AI Founders and Realistic Expectations for AI in Business Summary: AI founders are extremely dedicated and have been working in the field for years, driving their dream projects nonstop. Despite AI’s potential to accelerate business processes, the idea that AI can massively reduce engineering team sizes is unrealistic. Small and medium-sized businesses (SMBs) focus on essential tools like payroll, healthcare, tax services, and marketing software, rather than investing heavily in AI technologies.
Speaker 1
And I don’t think it’s an AI wave thing. I think it’s just the AI founders that I’ve seen and have been working with or just more hardcore in general. They’ve driven, they’re really smart. In some cases, they’ve been working in this area for years and years and years. And this is their dream come true. And they work nonstop. And I feel like that’s just a return to Silicon Valley as it used to be versus like something that new.
Speaker 3
But it does feel like AI can contribute, right? And accelerate this and like bring the ratio. But, but when people say, oh, yeah.
Speaker 1
You know, co-pilot, like you have co-pilot or chat, GPT, um, makes 10 engineering person team, one engineer. You’re like, no, it doesn’t. Yeah. No, not. What are you talking about? Right? If you are actually tried the thing, like, what do you talk? And you hear this on my guests all the time, right? Right? Yeah. You know, pun bits kind of talking stuff like this. Here’s like, that’s just completely false. Unless maybe your engineers are so bad that that’s true. I don’t know. You know what I mean? Like, you know, I think separate from that, there’s the SMB question. And in general, um, at people who run SMBs, it’s the true SMB, like a five person company, they’re really busy and they’re only going to buy like three or four things that they absolutely Need to run their business. They’re going to have payroll. They’re going to have healthcare or other benefits. They’re going to pay taxes, which is why Intuit exists, right? It’s like rippling. It’s Intuit. It’s gusto. It’s, um, they need HubSpot because they need to market or they need some sort of some form of serum.Impact of AI on SMBs and Job Displacement Summary: Small and medium businesses (SMBs) typically buy only a few essential tools like payroll, healthcare, taxes, and marketing, making them a difficult market for most products. The leverage on human capital in the SMB sector is currently overstated, but as AI advances, there may be some AI-centric solutions for SMBs. Certain verticals, like mid-journey jobs, are expected to be impacted by AI, displacing jobs faster than anticipated and expanding the market. However, the impact of AI on other areas like code generation is currently overstated.
Speaker 1
You know, co-pilot, like you have co-pilot or chat, GPT, um, makes 10 engineering person team, one engineer. You’re like, no, it doesn’t. Yeah. No, not. What are you talking about? Right? If you are actually tried the thing, like, what do you talk? And you hear this on my guests all the time, right? Right? Yeah. You know, pun bits kind of talking stuff like this. Here’s like, that’s just completely false. Unless maybe your engineers are so bad that that’s true. I don’t know. You know what I mean? Like, you know, I think separate from that, there’s the SMB question. And in general, um, at people who run SMBs, it’s the true SMB, like a five person company, they’re really busy and they’re only going to buy like three or four things that they absolutely Need to run their business. They’re going to have payroll. They’re going to have healthcare or other benefits. They’re going to pay taxes, which is why Intuit exists, right? It’s like rippling. It’s Intuit. It’s gusto. It’s, um, they need HubSpot because they need to market or they need some sort of some form of serum. There aren’t that many things that SMB buy and therefore SMB tends to be an awful market for most things, right? And it’s possible there may be some AI, SMB centric things, but I think the current leverage on human capital is a bit overstated, although I think that’s, um, gonna expand as we hit different Levels of GPD 5678, whatever. Um, but also I think there are some verticals where it’s really going to eat away at the vertical faster than we all anticipate. And mid journeys, a great example of that. We’re actually thinking of displacing people faster than you would have guessed for certain types of jobs as well as expanding the market. And I think there’s going to be a handful of those things that hit with each successive wave of capability to their predefined models and transform our base models. But I think today it’s kind of overstated for other things like code gen.
Speaker 3
Changing it up a little bit. We’ve talked about mid journey, a little bit open AI. We both, uh, we all mentioned Pika a little bit.1min Snip
Speaker 2
I think there’s existence proof of this already, right? Like, man, there’s a lot of aggregate revenue in the AI girlfriend apps, right? And so I think, like, can chat be a compelling interface for an application? World says, yes. Is that going to lead to an operating system? I can see both, I can see both sides of this argument, right? I think it is unlikely to be chat only. I think it, like, if we see a new one, it’ll be multi-modal. The bull case argument is, like, you know, this is the most natural interface. Everybody understands how to use it. We didn’t understand intent. We could not have our computers understand intent before. And there is a, there will be a killer general consumer application, not like everything, but something that will be the first app. And maybe it’s something productivity related with, or search related with different UX and it requires new hardware management and new resources. And then, like, you know, we go from that to a platform. That’s kind of the bull case.1min Snip
Speaker 1
And, you know, there’s no reason to assume that you’d want to not do that with a machine, although there’s other things that you do with a machine as well. So, you know, again, in the extreme, you probably assume that a lot of these, these interactions online, collapsing to agents, like on your behalf. And so eventually you’re not really that involved in many interactions, right? It’s sort of the extreme of a couple years from now or 10 years from now, whenever it is where you have agents representing corporations and governments and people and different, you Know, ways. And then you end up with more like programmatic interactions across agents versus you having to interact that much with a never face. So I think these things will evolve and it really comes down a little underline technology capabilities. And if you remember in the 90s, people came out with the first like PDAs, the personal whatever devices that was the proton smartphones. Yeah, they kept trying to do handwriting. Yeah. Because I thought people would write everything, right? And of course now we just type everything, right? But people thought, Oh, no, people don’t like typing, which everybody was doing. And instead we need to write everything. And they came up with a special language called graffiti where you learn to write the hell astray and the way that the machines weren’t smart enough. I understand how to write it and see how they can do. And that was really dumb in some sense. It was lauded as this brilliant breakthrough and how you think about human machine interface. But of course that isn’t used at all now.1min Snip Summary: Interacting with executives often involves brief calls rather than detailed emails. Both calls and texts have their use cases, and simplifying communication is key. Automation for communication is in demand, as seen in applications for AI accelerator programs.
Speaker 1
Almost everything that we do is we call people for certain things. And, you know, it’s interesting if you’re interacting with an executive, you know, their minions will write these really long emails and then the exact we’ll call and say like three Sentences, right? Because they don’t want to respond in detail and it’s way easier to talk about certain things. And then similarly, there are sometimes where like text makes a lot of sense. And so I just think it’s kind of like the midwitt meme. People make all the stuff super complicated and like often it’s just boils down to like what’s a dumb thing people already do for specific use cases. And that’s probably what they’ll do in the future.
Speaker 2
And we make humans do many things, including call people where it’s not a particularly intelligent communication, right? And so I think calls and chat, if you can break it down into the use cases, it’s actually pretty interesting as a AI enabled capability. We are about to launch applications for our next batch of our accelerator for AI companies in bed. And, you know, we’ve gone people doing early applications and I think we’ve seen like three different companies in this area, all of who have traction. Like in the last two weeks. And so I think there’s demand for automation of some of these types of communication.1min Snip
Speaker 1
But I think you really need a brand new capability. Separas is the thing I do this other thing better. And so in the short run, I’m a little bit, I think it’s really cool experiments and I’m excited to see where people take it. But I think a lot of the early iteration will actually end up being on the devices themselves. Yeah. And stand alone devices are going to be less performant unless they serve a very specific physical purpose, like what Andrew Rall or some SARA or Square or some of the other books did.
Speaker 3
I think that makes a lot of sense. And, you know, I think, I think it especially makes sense given that the OS is in the existing players, they keep a lot of these capabilities kind of walk down, right? Like it’s not easy to do things with Siri on iOS, right? It’s not super easy to tap into basically have the AirPods on 24 seven or something like that.
Speaker 1
I don’t know if you remember, there’s actually an error or people said that there would be AirPod companies. Totally. All these business who’ve been built around AirPods, right? So it’s kind of fast.
Speaker 3
It was hard to do.
Speaker 1
It’s hard to do. And also the question is, what’s the real functionality gain and what do you you know, generating off of it and all the rest? And so I just think like some of these things are also sometimes memetic. Yeah. Sometimes memetic startups ended being very large. You know, like when Instagram started, there was like two thousand different photo app uploading companies and they really nailed it. But then there’s often memetic companies where nothing comes out of it. They’re memetic waves.1min Snip
Speaker 2
Like there’s some application or is it because someone wants to train the model with the data from the device, right? Cause the writer is not an answer. Like it’s not a reason anybody’s going to buy the device. Um, and, and so like if you’re doing something new, like, do you actually need new hardware? Maybe. Right. Um, and obviously if you give people like new hardware, um, capabilities, then you’re going to get new applications. If you already have the distribution, but we’re just talking chicken and egg on the killer application. Yeah, I can imagine, um, just to like make explicit, I think what you were implying, like battery management, different sensors, different permissions management, um, contextual Intelligence that, uh, enables these better experiences is the glasses. Yeah, it could be glasses. It could be always listening. Um, it could be anything that is like passive image and audio output or just data from your different web services and like being able to, um, like reach that in some new or existing form Factor. But I think that’s the, I think that’s the biggest argument for it. Um, and the question is just like, can you do that in an application? Where does the compute happen? Like, well, the ecosystem’s permitted. It’s complicated today.1min Snip Key takeaways: • Robotics and self-driving cars are key areas of interest with advancements in deep learning and hardware. • Web three is being seen as having potential in AI and content authenticity. • Companies like Pika and mid journey are using AI to generate media based on original content.
Speaker 1
I’ll throw out three other examples to each other examples. I think, um, robotics is one of those areas. It’s becoming increasingly interesting. There’s been a series of research papers that have come out even over the last like four to six weeks that I think are kind of fascinating there in terms of moving more and more into like More standard deep learning world for the stuff. Um, and then the other area obviously self driving. Um, and you know, that’s just a form of, uh, again, robotics and hardware and everything else. And so I do think there are these, uh, big areas that will be increasingly tractable as we apply some aspects of. On-dition models to, to some of these domains in a physical world and have specialised hardware for them, but it isn’t that, you know, uh, consumer hardware device that seems to be on Trend right now. Yeah.
Speaker 3
Let’s talk about, um, everyone’s favorite topic, web three. Uh, there have been a number of people coming out recently and companies being started and launched that, um, it’s basically saying the one of the killer applications of web three, Uh, is actually going to have a lot to do with AI and content and content authenticity. Um, I think Fred Wilson even wrote something about this. Uh, Scott Belsky has talked a little bit about this. Um, if we have companies and models from companies like Pika and mid journey and, you know, name your, your product that generates some form of media and it’s all trained on some other Form of original media.1min Snip
Speaker 2
Um, but, uh, I think like from a technical perspective, um, it, it seems obvious that like as one of the solutions that is possible, I think in the short term, you’re just going to get declarations From companies that have rights agreements and give, um, creators in, um, commitments, uh, use this as a way to differentiate their offerings.
Speaker 3
What do you think a lot?
Speaker 1
Yeah, I’ve long thought that, um, I’ll be some blockchain resident, form of identity that’ll be used, uh, potentially for both the ascertainment of, um, origin of content, but also Specifically for credentials for agents. And so if an agent is representing you, how do you know that it’s accurate, actually representing a specific individual and can it partially reveal data or aspects of that person cryptographically In ways that are secure? Um, so, you know, that may be your healthcare data. That may be aspects of who you are or other things. So yeah, it’s an area I’ve been interested in for a while. Um, we actually had a ILIA, the cup under New York, who also was the last author on the transformer paper on our podcast. And I also did a fireside with him, um, uh, in Canada. Uh, and, uh, you know, I think a lot of these concepts are really interesting areas, but seem like beta evolve over time. It’s a big question of timeframe.1min Snip
Speaker 1
Uh, and, uh, you know, I think a lot of these concepts are really interesting areas, but seem like beta evolve over time. It’s a big question of timeframe. I think the technology is there. Out for the identity, not necessarily for the agents, right? Um, from a provenance perspective, the content stuff, I think it’s a little bit complicated because say, for example, you have an enormous amount of derivative art, uh, on the internet Already for Van Gogh. Yeah. Say that you actually removed the original van Gogh’s from your corpus of data that you train on. You’d still have enough signal on in the style of Van Gogh that it doesn’t matter.
Speaker 3
That’s right.
Speaker 1
And those works are all in the clear from a copyright perspective, right? We’re from a, uh, usage perspective. And so I also think there’s things like that that people aren’t really discussion about in terms of provenance and, you know, what actually exists in what is a training set because it’s Possibly you could remove all original art for a woman artist and it won’t matter.
Speaker 3
Right. Because it’s just already out there, right? It’s like, it’s already been implied enough such that somebody else is correct in a legal way, right?
Speaker 1
Like you are a lot of the youth and art and you are allowed to do things for style of Van Gogh by hand. And so I just think a lot of these issues are going to be, um, perhaps they may play out a little bit differently from what people think. That’s obviously very different from, hey, you know, what’s happening within New York Times article or whatever, and being copied for beta or not.
