Data

“The challenge for the analysts was they used to spend, in fact, about 30, 40% of their time just discovering data and playing around with data before they could get to analysis.”

“And actually a lot of that industry has now Built these gimmick style benchmarks as a mechanism to sell their data. And so that, that's become kind of their go-to-market as well.”

“We should zoom out and recognize the biggest problem right now in robotics is actually data. One day it'll be chips, but for now it's data.”

“The data is really the hardest part of model training. Because you need to source it. It's so dirty. Nobody wants to do that shit. Nobody wants to hire all these people to generate data and then, you know, turn that into basically data plus GPUs equals model. …”

“We were sort of at this moment where, yes, we can actually significantly raise the input, both in terms of data, compute spend. To actually meaningfully accelerate this progress, that's when we thought it actually makes sense.”

“We should be thinking about spending as much on the environments as on the compute and the data, because the environments literally shape the, what intelligence can emerge.”

“Most of the information to run, you know, DoorDash to be a great service you know, are things that we use for ourselves. And the reason why we use them for ourselves is because it's not just that simple like, oh, we just give away information, and then somehow…”

“We have a huge repositories of data at the department that's sitting there siloed unused.”

“If you scale up models, you give them more data, more compute. Again, there are a few modifications like RL, but not really very much. It's pretty close to pure scaling. You find that, you know, when you do that, you find, you know, incredible increases in per…”

“Data is a prerequisite. It is not the answer.”

“We could reach a point when, for instance, currency will be AI tokens, or even time on servers, or data. That AIs will trade with other AIs for tokens or for data, not for dollars. And the human currencies, like dollars and euros and yens, they will become irr…”

“Now, if you have the data and you have enough GPUs, you can solve damn near anything.”

“The GPUs are somewhat zero-sum, right? So the data isn't.”

“I think the SaaS platforms that will win with AI are the ones that have data across platforms. You know data is really the true winner.”

“What you really want is you want to diagnose with data and treat with design.”

“You have LLMs, which have generalized pretty, you know, like in fascinating ways, but they've kind of also asymptoted in that you know, we have run out of data for the most part.”

“But the algorithms I think are not the bottleneck in terms of making those models more useful. I do think a lot of it is still getting good quality data and then making good quality synthetic data From your good quality real data.”

“Something I've said before and I'll say again is, is that data is the most under-invested in area of research relative to its impact, and I don't think it's even close.”

“If you talk to the most talented AI researchers and you ask them, what's the secret to your success, they'll largely tell you that they look at the data.”

“And I think, by the way, people are going to buy businesses just for their data.”

“I think the difference is that like for reinforcement fine tuning, you're collecting data that's going to be useful As the models improve as well. So if we come out with, like, future models that are even more capable, you could still fine tune them on your da…”

“So my hot take is the most defensible things 10 years from now will be your brand, workflows that require high trust and data.”

“One analogy a security professional gave me years ago was you should think of data as sort of like toxic waste. You have it, you might need it, but you want to keep it contained, and if it starts leaking out, it's very hard to kind of clean that up. And again,…”

“Their mode is probably not the user experience, but because it's very easy to copy. Anyone can study the product and copy. Their mode is data.”

“The big revenue generator is the software. Okay. It's not the device. The device is purchased or licensed, but it's really the software and the data. Okay. That is the true value.”

“What hotels do after you leave a room, they literally go in, they count how many towels you use, What portion of the meal that you've eaten, they, how much of the soap that you've used, everything, and then they store it, and that's how they've been collecting…”

“There may be like natural limits to certain types of intelligence if there's not more data to iterate on.”

“Remember when it was just data? You can get access to our data. Well, data is nothing. It's the way your employees are using that data to make value out of it. Guess what models do and good models are going to do it even better.”

“Data is more like a compass than a GPS. Right? Like if you look at data as a way of like giving you the answer, you're always wrong. You're always wrong or you're slow. Wrong or slow or sometimes both. Because mostly data doesn't give you the answer. It just t…”

“If you're looking at a piece of data and the result tells you something that your intuition tells you is like insanely wrong, like that you're probably not right. First believe your intuition and go and prove yourself right. Like don't just take it at first gl…”

“In data that basically never, and like enterprise software broadly, organic growth almost never happens. There's too many points of friction, like the buying processes and salespeople and all of these different things inject friction.”

“I honestly think right now at where we are in AI development today We are more bottlenecked by data than we are compute.”

“Data is a compass, not a map. Right? The idea that we ultimately need to determine what mountain to climb, right, with the intuition we have, with our understanding, and with our empathy and yet data is ultimately what helps us climb the mountain we choose, an…”

“All the data is basically pointing one direction, which is a Trump victory.”

“It's very hard to accurately interpret data, and it's very difficult to not lie to yourself with data.”

“In which data and AI were at the center, and the work that was being done by these organizations was being done by the machines and not by the humans. The humans designed the machines, the humans monitored the machines, the humans updated the machines, but the…”

“Certainly the last five years, but arguably longer, the story of AI is really a story largely about the deployment of just enormous amounts of compute and enormous amounts of data, and not that many new ideas, candidly.”

“It's like you need data for every scenario where you want these models to reason well in, You just need to overwhelm them with data in all those scenarios, and you're gonna get models that can reason really well.”

“My vision is that in the future the efficiency is very important because we are running out of data and compute. So we have to either use the data much better and use the compute much better.”

“People often think that they get scale economies through data. And I'd say that that's possible, but it's rare. And the reason it's rare is not because there aren't scale economies in data, but rather that the range of scale that the existing competitors have…”

“I don't think we know yet that, you know, we'll run out of data any, or that there's some like limiting factor here.”

“And so that means you kind of can't do like a prediction. There's no Moore's law here where you can say, well, it'll get to that power of compute level at this, but at this point set aside the fact that we actually don't have enough data to give it 10 or a hun…”

“Getting into the data is not, is something I think a lot of people who have expertise but fail in leadership, that's one of the primary flaws they have.”

“In healthcare especially, the limiting factor in building AIs is data. If you don't have access to a really unique proprietary type of data, you don't have edge.”

“Data does not tell you what to do. Like that's why growth people who are creative, who thinks about unique ways to drive exponential growth are always going to be win and are always going to be much better than anybody else that is out there.”

“I do believe they can generalize. I think they generalize relatively narrowly, or at least, you know, as long as you stay close, you get a good manifold of information. When you start to go really far afield from your data, because the dimensions are so large,…”

“Cause I mean, you know, at the end of the day, like, you know, these models are more or less commodities. It's what feeds them. That's important.”

“Most people know what they want to do and then they use data to justify what they were going to do anyways. And you could always find some data to back up your case if you're certain.”

“Outcomes, business insights, they're a function of compute in close proximity to data. Like, compute in close proximity to data is where you're going to get the results that you're really looking for.”

“If you could figure out how to pour enough compute and pour enough data into a neural network, there's really no limit to how, how good it could get.”

“My general sense is that these sort of like linguistic oriented strategies really don't work that well. It's kind of like by feeding more data in and sort of like Working on outcomes, you can figure these things out much better.”

“Now always start with words. Do not start with data. You always start with words.”

“I've been advocating that it's, at some point, it's gonna be NILD. And the D stands for data.”

“I believe in relationships more than I believe in full on data. And while data is very, very useful I kind of feel like The data is only there to help you strengthen the relationships that you actually make. And that could be relationships within your employee…”

“Going to natural language is like, it's like the last mile here. And that is, is an enormous thing. I mean, the effect on demand will be just enormous because every mortal, if you're semi literate, maybe not even literate, you can just talk, you know, you can …”

“They weren't writing about themselves like we're the baddies and we really want to oppress the crap out of people. They wrote about themselves like we're going to do a solid for society and we're going to make society better with data.”

“It's really for the first half of the life of artificial intelligence, people thought it had nothing to do with data whatsoever.”

“If you have a business that's focused on innovation at the core, that's not actually a business. It becomes a business when that innovation becomes product, becomes distribution, when it has an advantage on data and other things. Those are real moats.”

“Between this kind of activation usage, aha moment to this conversion moment, Actually, there is a big gap and what is the gap? How you can understand the gap, how you can guide people along that journey is basically you need to have a very good grasp of data. …”

“I've seen the power firsthand from working at Facebook of what data can do to help us make better products, especially for people at scale, to help us reduce the bias in our intuitions and how we think about what is the way that we should prioritize.”