Loading summary
Transcript116 lines
- [00:00]
A
Foreign.
- [00:09]
B
Should we get started, Jeff?
- [00:10]
A
Sure. Sounds great.
- [00:11]
B
All right, Jeff, welcome. And again, thank you so much for being here, especially. I just got a cold. And thank you for being here.
- [00:18]
A
Yeah, I'm afraid I've lost my voice. I don't normally sound quite like this, but we'll do what we can.
- [00:24]
B
So you built MapReduce, BigTable, TensorFlow, the TPU, Gemini. We could spend a whole hour on all these things. Things are done. But what I love is that you're still making bold predictions in public. Last year, yes, last year in May 2025 at AI Ascent, you said that AI is at the level of a junior engineer. That was about a year ago. It's been. How close are we to that prediction?
- [00:58]
A
Yeah, I mean, I feel like the models have been getting a lot better at sort of agent based, longer running coding tasks and it seems pretty clear that they are now actually pretty capable. And depending on exactly your definition of junior engineer, it seems pretty spot on, I would say.
- [01:15]
B
What did you underestimate from that prediction?
- [01:21]
A
I mean, I think the, the ability to do more and more complex tasks has been growing faster than I thought. And I also think outside of coding, these agent based systems are really starting to shine in other domains and I think that's going to be an important trend in the future.
- [01:44]
B
So give us another bold prediction. What do you think is going to be the 2027 edition?
- [01:52]
A
I think you will see a lot more automation of ML systems themselves. Basically getting ML systems to improve their capabilities by running lots of experiments, breaking things down into sub problems, running those sub problems in a tight automatic experimentation loop, putting the results together and being able to then get some improvement system out from that sort of fully automated problem decomposition and automated experimentation. I think that's going to be really exciting. I think that also applies not just to ML, but also to other fields of science and engineering. Basically anything where you can have a measurable objective, I think you can actually make a lot of progress these days.
- [02:40]
B
Now let's go back to a little bit in history. Back in, way back in 2001, Google search used to run on hard drives and you and Sanjay did the math and realized that at some point the whole search index would finally fit in all of the RAM of all the computers you had running. And you made that radical realization and you basically, in few days with Sanjay shipped in production a whole new search version that worked in RAM rather than hard drive. And that was the thing that got Google to be so fast. Google searches. So history tends to remix what is the it fits the memory moment right now in 2026 that everyone in this room is still should be thinking about and designing.
- [03:37]
A
Yeah, I mean it's a little. But I think you're going to see more and more high performance and low energy inference hardware systems. Because I think everyone is now realizing that inference is the key to making these agent based systems be available to more and more people. And that latency is really important and that specialization of the hardware is a really key way you can make things that are more energy efficient and lower latency than more general purpose computational devices like say GPUs or TPUs.
- [04:16]
B
Because I think everyone here is used to waiting for responses on models.
- [04:23]
A
Waiting is no fun.
- [04:25]
B
Master of speed. So you're saying what if we don't have to wait anymore?
- [04:31]
A
Yeah, I mean, I think we'll imagine what you could do with something where the latency is, you know, 50x better.
- [04:38]
B
Interesting thought. Now, what's one assumption that perhaps 6,000 people in this room hold as already false about AI?
- [04:50]
A
Yeah, that's a good question. I mean, I think probably one thing is people don't quite realize how possible it is to have agent based systems that can run not just for an hour or two hours on a problem you care about, but for some problem domains and with highly capable models underlying them. You can get them to run for days or weeks and do really, really complicated tasks. And I think that's starting. Some people are starting to see inklings of this, but I don't think everyone has really internalized this and that's going to be really a pretty big deal.
- [05:28]
B
What's a particular task that you have run that has run for weeks? What did you tell the agents to solve?
- [05:35]
A
Yeah, I mean, I think you can tell agents to go off and implement, you know, completely new versions of software in different programming languages that might be, you know, have better safety properties or better performance properties and then they can go off and actually do that in a pretty serious way.
- [05:59]
B
That's pretty cool. Now one thing that you've been very well known for is you're really good at napkin math. Sounds funny. One of the stories about you is that back in 2013 when speech recognition started to work at Google, you did the napkin math, where if every Google user used their phone and talked to it and used the speech recognition system for just three minutes a day, you found that the system requires a Google server. You would have to double the fleet, which would be really, really expensive, just to do speech translation. Instead, you basically build a custom ship and that was the origin story of the tpu?
- [06:48]
A
Yeah, yeah, I mean, I sort of had done. You know, we were starting to see really good quality results on the sort of deep learning based speech systems, speech models we were training, but they were computationally expensive compared to the old speech system, but they halved the error rate. So that was like the equivalent of 20 years of advances in speech recognition in just a few months of like fiddling with the model and scaling it up a bit and getting better data. And so we started to get worried that if speech worked a lot better, people would use it more. And so that back of the envelope calculation was really about that. Like, well, what if people start to use speech recognition more to dictate emails or to talk to their phone or whatever. And yeah, it turned out that we realized that we needed some better solution than running on CPUs at the time. And so we came up with TPUs, which are sort of very specialized for essentially low precision dense linear algebra, which is at the heart of nearly all of the modern machine learning algorithms we use today. And if you build a specialized chip for low precision dense linear algebra and can't do anything else, that turns out to be really useful for machine learning inference, even though it can't run Chrome or Word or whatever. And so that system produced a chip a couple years later that was 30 to 80 times more energy efficient than CPUs and GPUs of the day. And also much, much lower latency, like 20 to 30x lower latency, which is incredible.
- [08:29]
B
What the foundation that TPU has become today. No way you would have predicted that TPU would be so foundational now with transformer architecture, which was invented way later, before you actually invented the tpu.
- [08:42]
A
Yeah, I mean, that's sort of why we built a general purpose linear algebra system, which is what a TPU is really, because we knew ML algorithms were still evolving and you didn't want to over specialize, but you wanted to specialize enough that you got the dramatic performance benefits of we could have very big multiplier units, we could have high speed memory, we could have high speed interconnect or later TPUs that brought many, many chips to bear on the same problem efficiently. And we've continued to scale those up and improve their performance over many, many generations now.
- [09:20]
B
Incredible napkin math. So what's good? Napkins are good. So actually, what's a good napkin math that everyone here who wants to be a future founder should run tonight to potentially build something as consequential as the tpu?
- [09:39]
A
Yeah, I mean, it's always hard to say. I think, think about what problems you see and whatever it is you're thinking about what bottlenecks you see. And are there very different ways of thinking of the solutions to some of those problems that would get you an order of magnitude or two orders of magnitude better performance or capability or whatever it is? Because sometimes if you just squint at a problem and you think about not necessarily being anchored on exactly how that problem is solved today, but how you would solve it from first principles, you can come up with really good ideas that are maybe not what other people are thinking about.
- [10:25]
B
That's a good tip. Now for everyone here who doesn't know, years ago Jeff wrote a very famous list called the latency numbers. Every engineer should know these are numbers around, for example, how long a cache miss takes disk seek a network package traveling, let's say from California to Netherlands. Lots of numbers like this about distributed systems and systems engineering. And it's been sort of taped and become the bible for a lot of distributed systems engineers.
- [11:00]
A
Okay, yeah.
- [11:02]
B
Now fast forward that list is up for an update. Give us the AI edition for now. 2026.
- [11:09]
A
Yeah, I mean, I think if you looked at what is important in AI systems these days, you would want to know things like the bandwidth between your main memory system on your accelerator to the on chip memory to the multiplier unit or whatever. You want to know how much energy does it take to do a single multiplier operation? What is the interconnect bandwidth between chips and how much does that, how many chips can you connect without bandwidth? And then if you go beyond that domain, like what is the fall off in network bandwidth when you need to talk to 10,000 chips instead of 500 or something. I think these are all really important numbers to learn and really affect how you think about solving particular kinds of problems.
- [12:06]
B
And one interesting thing that I've heard you talk about is that nowadays the unit that you measure everything is energy.
- [12:15]
A
Yeah.
- [12:16]
B
You pointed out that doing a calculation or math costs about one picojoule, but moving the data and doing data IO costs thousand times that.
- [12:26]
A
Yeah, Just bringing it in from HBM on an accelerator into the processor so it can actually compute on it.
- [12:32]
B
Yep. That gap kind of quietly decides what products are possible and how these algorithms in AI are built. So what are the kinds of problems that founders keep calling model problems, but are in fact actually energy or data IO problems?
- [12:51]
A
Yeah, I mean, I think the example you raised of 1,000x difference in bringing moving data versus actually computing on it in terms of energy is a pretty significant one. And it shapes a lot of aspects of what we do in machine learning. Because if you didn't have that thousand x difference, then you wouldn't have to do batching, but you have to do batching of many examples or maybe many tokens at once in order to amortize that data movement. So that you can not pay 1000x slowdown, but pay a 1000x divided by batch size energy cost for really low latency batching is not really very good. So I think these kinds of things and the energy behind various decisions in the computer hardware we use really affects a lot of decisions we make in building higher level systems.
- [13:52]
B
A very concrete example is just how training models is done. There's this whole concept of batching the data sets and running epochs that's basically. People perhaps may confuse that as a model problem, but it's really a systems data I O problem, right?
- [14:08]
A
Yeah, yeah. I mean, you have to assemble batches to get better efficiency in your hardware. Ideally you might do batch size one training, but it's not as good in terms of efficiency. So people use pretty large batches these days.
- [14:26]
B
Do you think it's possible for. I know you're well known for taking off on a long week or weekend and coming up with this brilliant solution. Is there such things of Jeff Dean going and working on it for a couple of weeks and getting batch size equals one training done?
- [14:43]
A
Yeah. I've been thinking more about inference actually. So I think inference is a pretty interesting problem because you do want very low latency training. You don't necessarily need incredibly low latency. And I think there's a lot of room for specializing hardware more for inference than we are today.
- [15:02]
B
What are some of those interesting things that are on inference that you're really thinking a lot about?
- [15:08]
A
I mean, just trying to minimize data movement, trying to think about incredibly low precision operations and maybe not supporting lots and lots of different kinds of precisions. If you feel like you have a good answer for what kinds of precision you need, maybe just build that into the hardware and not much else.
- [15:33]
B
Which I think it brings down to a core analogy. I heard from famous computer scientists that really the whole process of AI is a big compression problem because in order to have the data to be fully lossy and compress it and then restore it, you basically need to understand it.
- [15:54]
A
Yeah, I mean, if you truly understand the data, you should be able to compress it really well.
- [16:00]
B
And now transformer architecture is basically one of the ways that has turned out to work really well. Yeah, yeah, I would say working pretty well so far.
- [16:10]
A
Good work by my colleagues.
- [16:12]
B
Yes. Now let's zoom out a bit. AI progress used to mean just better models. You had more data train bigger models with bigger parameters. But increasingly in the last years or so, it's everything around the model, not just the model size and number of parameters or more data. It's everything around things like retrieval tools, memory agent tools. And it might kind of get consolidated into what people call context engineering.
- [16:41]
A
Right, Yeah. I mean, I think the model is really only one piece of what you're trying to do, which is build an overall system that can solve really interesting problems. And that involves a model that knows how to use various tools. It maybe knows how to retrieve relevant information, maybe has a history of other information that has retrieved for past problems. And it can put information into the context of the model. And the nice thing about that is that information is really clear to the model. Unlike the training data the model was trained on, where it's all kind of like trillions of tokens stirred together into a soup of hundreds of billions or trillions of parameters. But it's all less clear than the actual context that the model sees directly for this particular problem or use case. And then I think being able to understand what tools are available, which ones are going to help me solve the, help the model solve this next phase of the problem. How to decompose a problem into a sequence of tool calls, maybe trying multiple approaches to solve the problem and seeing which ones work and being able to evaluate that this is the whole orchestration of complex agent and multi agent systems that I think is going to be more and more important and super exciting times, I would say.
- [18:10]
B
And I think the fun thing about this particular problem domain set is actually something that everyone in this room can actually do. Because before to train a model you needed incredible amount of resources, incredible amount of access to GPUs and data. But for context engineering, everyone here could do it. You just need the API to something like Gemini and then work on your own setup for your own retrieval, your own tool calls and et cetera, et cetera. So what are some tips for everyone here? How does everyone get better at and become exceptional at context engineering?
- [18:47]
A
Yeah, I mean, I think a really good way to do it is to use these models and sort of harnesses and tools and so on to try to solve problems. And then sometimes you can actually see where the models are failing and often you can actually make the model work better. And succeed at that kind of problem by not just adjusting the model parameters, which is hard to do from the outside, but from creating better guidelines for the model writing skills for the model to know how to use different tools that would be incredibly useful for solving this particular class of problem. And I think as you do that, you end up on this kind of improving, self improving of the setup that you're trying to use to solve things. And that's a really good way to get better at understanding what additional information the model would want in order to become more capable.
- [19:46]
B
Can you give an example of some context engineering you personally have done? I don't know. Skills you wrote, tools that really made a huge difference in your workflow.
- [19:57]
A
Yeah, I mean, I guess Sanjay and I were working a few weeks ago and we often do some amount of performance improvement for very low level libraries. And we have a micro benchmark library we've written at Google where you can write micro benchmarks of how long different kinds of operations take or how long does it take to populate this data structure or whatever. And sometimes those data structures are used on millions of processes across Google. So it's actually pretty important to make sure they're high performance. And so you can write micro benchmarks, but then without an agent based system, what you usually do is you measure what the current performance is on some benchmarks you care about, you make some modifications to improve the performance you hope, then you rerun the benchmarks, see where things improved, you run maybe a broader set of benchmarks, measure the cache footprint of things. And so we wrote a skill that basically taught the model how to do most of those things in various sequences so that it could actually do self improving. Benchmark measurement, benchmark code changes, measure the performance improvement and then iterate on that. And that seemed to work pretty well for some kinds of problems. And it really just is us giving the approach we would use as people to the model in a form that it could use.
- [21:24]
B
Wow, that seems very impressive. So you're saying you have this skill that if someone got access to it, it could do perform optimizations like Jeff Dean. Seems like the world would love this and is worth infinite amount of money to someone have access to this.
- [21:40]
A
We actually published a document maybe a few months ago called Performance Hints that Sanjay and I wrote. That's like a 30 page document about various kinds of performance tricks. And some people have taken that and then given it in summarized form to various models and seen that that model can now get better at reasoning about performance issues in code.
- [22:01]
B
So you heard it all here you could actually get your own optimize your own code like Jeff Dean, if you take this paper that you publish when perform a sense.
- [22:09]
A
Yep.
- [22:10]
B
It's all free available so you should all try it. Very cool. Now we talked about agents. Everyone here is probably building one or built one at some point. And I'm sure everyone has seen your agent go off the rail at perhaps like step 30 or 40. Like agents are great for up to step, I don't know, 10 or something and then gets shaky at step 50. What do you think is the constraint today? Is it like context evaluators or just errors that compound because it's basically an open loop system.
- [22:41]
A
Yeah, I mean obviously we want agents to be able to run for very long periods of time because that's how they're going to solve more and more complicated problems. But as you observe today, they sometimes stop working after 10 interactions with tools and so on. And sometimes that's because the model is trying to do something it doesn't have a lot of experience doing. So it's been trained on a whole set of things and as soon as you get a little bit off the distribution of things it knows how to do, then like most machine learning models, its performance will suddenly start to degrade. And the farther you get off the comfort zone of what it knows how to do, the more likely it is to not work as well. So there's a bunch of things you can do. So one is give the model skills and hints that kind of tend to keep it in the sort of more brightly lit path of things it does know how to do. I think having multi agent systems where you have multiple agents trying different approaches and you can evaluate, you have maybe another model or another agent that's evaluating which ones of those seem promising is another way to kind of in some sense search the path of. Search the space of possible solutions and stick to the ones that seem most promising and discard the ones that didn't seem to work or maybe that went off the rails or whatever. And that's a very, very useful general technique is inference time compute to perform search over plausible ways of solving the problem. That can get much, much higher performance or much more reliability in long running agent flows.
- [24:30]
B
How are some ways you implemented this particular workflow for your agents internally?
- [24:35]
A
Yeah, I mean we have harnesses and then we have a whole set of skills, particularly in the internal Google development environment. We have skills so that the agents can know how to use lots of our internal tooling for coding or for code reviews or for measuring performance or fetching log files. And those are just skills that you can add to make the base model more capable, even though it hasn't necessarily been trained on exactly the way that Google internal engineers would fetch log files from our proprietary system. With the right kind of skill definition, you can actually get it to work and that improves the usefulness of the agents.
- [25:22]
B
Now let's talk about where startups can win. This section is one that I personally care a lot about because also everyone here in this room needs to decide what to build in the future of your future founder. So the thing about Google is you co design everything on the system from the processors to the products, which are the layers that someone like Google would keep building and compounding being better. And where does a two, three person team can still win?
- [25:57]
A
Yeah, I mean, I think obviously Google and our Gemini models and our hardware infrastructure are really trying to build very general models that can do almost anything. But in a lot of cases that means that we don't have a lot of attention on particular domains where perhaps a really well designed surface and maybe a model and set of skills, or maybe a specialized model that isn't in sort of the general mix of things that our models do well can actually have a significant advantage because you can build something delightful and really high accuracy, really high quality for a domain that you are really passionate about. And I think that's where the two or three people in a room building that, that they're really excited about can have an advantage. But I would also caution that the general models are definitely getting better at a broader and broader range of things. So you have to figure out, you know, is that thing you're working on, is that going to be a durable thing, or do you think the models at the forefront are going to get better at that in the next six months or 12 months, or is it something they're not going to be able to do for a couple of years or three years? And you want to weigh that as you're deciding what to work on.
- [27:22]
B
So let's dive deeper into this. So the general models, of course you're going to keep working on and keep making them all better. And how should the audience reason about what are those areas that it doesn't? I mean, how should founder think about things to pick on and work on?
- [27:41]
A
Yeah, I mean the most important thing is to pick something you're super excited about and want to build and you think would be useful in the world. Right. So if you do that, you're already way ahead than if you wake up and you're like, oh, I don't really want to do this or whatever, or you're going to build something that is actually not that useful to the world or to many people. So I think that's the number one selection criteria I try to apply for what problem should I work on next? Second, I think you want to look at what the current, more general models can do in that problem domain. You can test them with are they able to do this thing very well? And if they're completely failing, that's probably a good sign. If they're kind of able to do some of it, but not very well, that's maybe not a great sign because that's probably a sign that the capability is starting to be present in those models. And with more training data or larger scale models or whatever, it's likely to get better. So look for something where the model succeeds 0% or 1% of the time, not 20%.
- [28:55]
B
How do you find those? I mean, are those things effectively out of distribution from the training set? And what exactly is the problem shape that fits that?
- [29:04]
A
Yeah, I mean, I think sometimes it's a product that you build that might have access to particular kind of data that the underlying model might not a general model. So it might be you're building something to help users organize all their own personal information and the model won't necessarily have access to that. And so there you can have a big advantage because all of a sudden your model has visibility or your product has visibility into important data. It could be some incredibly hard problem where if you get the right training data and you can train a more specific model than a general purpose one, you can actually do that in a very affordable way. Maybe it doesn't take that much compute to train a niche model for this particular problem, but you can get something that's highly accurate, that can sometimes be a really good building block for solving a important problem that is maybe not handled very well by the general model.
- [30:05]
B
I think that's interesting. I think you're basically two paths. The first path is a little bit funny. You guys are organizing the world's information that's probably kind of well covered, but organizing your personal information, that's open, which is funny.
- [30:21]
A
Yeah.
- [30:21]
B
And then the second path, you talked about more specialized models in certain domains. Can you tell us more about what are some of these domains?
- [30:31]
A
Yeah, I mean, I think, like if you look at my colleague's work on, say, AlphaFold, that was a very specific model for protein folding and it was highly successful and was able to really handle that domain quite well. So that all of a sudden you now have this amazing tool and model that can give you answers to questions about proteins and their structure really effectively. But it's not a general model, it's a very specific one. And there are other, I think, domains where that kind of approach can work really well. Maybe a material science or chip design or things like that that will enable you to leverage the capabilities of a very accurate but niche model to do things that are hard today.
- [31:19]
B
That's a good example. So if some of you find a problem that's similar shape like AlphaFold could be a good problem to work on. Now let's assume you found a problem to work on. We're going to talk a bit about how do you become a AI native founder? How do you really become good at it? You in the past said that managing a fleet of agents, let's say 50 or 100 agents, is all about writing really good crisp design docs or specs. And how do people get good at that? What do those look like?
- [31:49]
A
Yeah, I mean, I think. You'll have a lot more success when working with your virtual agents if you can clearly specify what it is you want. And the clearer you are on what it is you want, the more the agent will have guidelines and rules, an outline of what it is trying to accomplish. Whereas if you don't specify very much stuff, the agent has to infer what it is you meant. And in many cases it might infer things that are different than what you imagined. So we've always told computer scientists from the very beginning, but it's really important to specify what the software that you're writing is trying to accomplish before then going and writing it. And so now we actually have agent based systems that can do the writing. But the importance of specifying what it is you want has actually gone up, because before you'd be handing it off to a very intelligent human who maybe has context or can ask you follow up questions. And agents can sometimes do that. But I think clear specifications is a really good idea. And to give you an example of a use of a coding agent that works extremely well is you can ask today's models to translate software from one computer language to another very effectively, because in that case you actually have a incredibly detailed specification. You have the whole software that says what the system is supposed to do. And so if you have a Python implementation of something and you want a go implementation of it, that is something that the models seem incredibly capable at doing these days because it can sort of take all the tests that are in Python, make sure they pass in the Go version, translate the test to go, compare behavioral differences between the implementations until there aren't any, and be highly effective because that spec is so clear.
- [33:52]
B
Now let's assume now every founder gets good at running hundreds of agents at the same time and all the code is written for them by the agents. What becomes the scarce skill?
- [34:04]
A
Yeah, I mean, I think it's really having incredibly good taste in what you ask your agents to work on. Right. That is the, the crux of, from my background, a research problem. A researcher can have all the tools and all the techniques, but often most of the battle is what problem are you going to spend your time on? And if you pick the problem well and you succeed in solving it, that's way better than if you delightfully execute a research investigation into a rather boring problem. And so that high level wisdom of what to work on I think is incredibly important. I think models are not necessarily going to be that good at it. So you're going to have people steering a lot of AI assisted computation in order to accomplish great things and more quickly. But that essence of what it is you want your models to do is the key thing you should focus on.
- [35:08]
B
So let's talk a bit more about taste, because it gets talked a lot about right now in this current era with agented coding. How do you exactly build taste and do that? I mean, yeah, that sounds so esoteric. How do you make it concrete?
- [35:23]
A
Yeah, I mean, it is a difficult thing. It's not like there's a measurable invective of, of taste in a lot of cases. I think some of it is from experience. Working on a lot of different problems in the past kind of teaches you about what kinds of problems might be interesting in the future or what kinds of things might be just barely possible. By cobbling together these previous approaches and then some open problems you might have to work on in order to get to something kind of magical or, you know, highly useful. Another way you can get more experience for yourself is to just write down a bunch of things you think might be important in the next 12 months and maybe you pick one of them to work on. But go back and evaluate in 12 months of these other things, which ones actually seemed important or which ones did other people in the world go out and create and which ones did not seem to do yet. That can give you a lot more samples for your own sort of taste creation capability. And that's an important skill to have,
- [36:36]
B
I think A third way we were talking earlier was doing very crazy thought experiments.
- [36:43]
A
Oh yeah, that's another good way. I mean, I think sometimes it's good to not take as a given things that most people seem to take as a given. So I was doing a crazy thought experiment with some colleagues the other day about. For 60 years, the whole silicon chip design industry, design and fabrication industry have done tremendous work to make smaller and smaller scale transistors that are very low error rate. Right. Because what the assumption that we want is that every chip we manufacture of the same design should be identical to every other chip.
- [37:30]
B
You don't want any bits to flip everything.
- [37:32]
A
No bits should flip.
- [37:33]
B
Deterministic.
- [37:33]
A
There's all kinds of things. There's all kinds of error margins built into memories. Have ECC memory these days at the macro scale, we don't make that assumption when we're building large scale distributed systems. Right. We build reliable large scale distributed file systems out of unreliable parts. Right. Like individual disks can fail, but your data should be safe. And so we have mechanisms at a higher level to enable us to have, you know, three copies of the data on three different machines and three different racks so that if any rack switch or individual machine or disk fails, you still have your data. We have Reed Solomon encoding techniques, but we don't seem to do this at a really extreme level in the sort of transistor level scale of the technology we're working on. So basically an interesting thought experiment is what would happen if you tried to build a system out of transistors that might have 20 errors per day.
- [38:39]
B
Oh my God.
- [38:39]
A
Rather than one every million years. Right. That would be a very different design point and might enable you to do really interesting things. In the fabrication side of things, you have very different kind of design methodologies because if you want to get a signal from here to there and you have these super unreliable transistors, you might have very different ways of signaling. You might send it along multiple redundant paths in order to make sure that it gets along one of them. And I think that would be a pretty interesting set of thought experiments. I'm not saying we should go do this, but that's the kind of thing where you do want to occasionally question assumptions. Now, oftentimes these thought experiments don't work out because there are very good reasons that for the last 50 years we've done this thing this way and not that way, but it's good to revisit those every so often.
- [39:32]
B
That is so wild. I mean, it's starting to rhyme a lot with Neuromorphic computing or the human brain and how nature works.
- [39:40]
A
I mean, exactly. Like signals in our brain are not especially reliable from getting one place to another. And so I think in brains, when there are really important things you need to get from one place to another, there are multiple pathways that enable you to sort of do that.
- [39:58]
B
I mean, you have such an impressive career. What is one of these crazy assumptions that you threw out of the window that actually built a consequential system in the past?
- [40:09]
A
Yeah, I mean, I guess that worked out actually. Yeah. I mean I think while TPUs is a good example, like being able to specialize hardware for a very niche problem domain, before that problem domain seemed as important as it is today, is one thought experiment. You know, I think the the origin of MapReduce is another good example. We had worked, my Sanjay and myself and a number of other colleagues had worked on various iterations of the crawling and indexing system at Google. And we'd sort of written lots of hand parallelized code with lots of checkpointing to make sure it would be robust and reliable if it was running on 100 computers or 1,000 computers. And some of those died. But that code tended to be intermixed with the actually relatively simple thing you often were trying to do. Like I just want to look at all the contents of all the web pages and then compute on the side mapping from URL to what language is this page in? It's the text of this page and it would get obscured by all this kind of other code for parallelization and reliability. And so we sort of remembered our training in functional languages and realized we could squint at those problems and develop this MapReduce abstraction that you could have above this implementation. And then below the implementation you could put all the checkpointing and reliability mechanisms into that lower level library that everything could then build on. And so that became a hugely successful way of dealing with very large scale computations at Google in a robust and reliable way. From that thought experiment of like, well, if we squint at it, could we find lots of problems that fit into this abstraction?
- [42:04]
B
That's impressive. So this thought experiment led you to create Map reduce. Awesome. Now let's go back to you talked a bit about your interest. Right now I'm working on a lot of customized hardware. So right now Alpha chip lays out chips. Now you also got Alpha Evolved that proposes solutions, evaluates them and keeps all the ones that work. Seems like you're starting to build all these system that can compound and build AI. That builds AI.
- [42:33]
A
Yeah, I mean, I think more generally there's this sort of the foundation of the scientific method of you propose an experiment, you implement what you need to run the experiment and you evaluate the experiment and then you get results from that. And I think there are more and more problems that are now possible to implement where that whole loop of running not just a few experiments, but running many, many experiments, because you're able to automate that loop and make the latency of that loop extremely low, is going to be really, really important. It's going to enable us to tackle lots of different problem domains in science and engineering and machine learning, model design itself, and also in engineering tasks like designing chips. And so if you can actually do those things in an automated way and have some orchestration framework that can take very high level objectives and break them down into sub problems, and each of the sub problems can be one of these automated loop that is exploring the best way to solve that sub problem and then a orchestration framework that can put together sub problem solutions into the overall solution for the higher level problem. That's going to be really impactful and it's really, really important. And I think it'll enable us to do accelerate machine learning progress. It'll enable us to accelerate science and enable us to accelerate engineering. And I think that's going to be amazing.
- [44:09]
B
That sounds awesome. I mean, it sounds like a lot of fields basically where you can have very good evaluators and maybe adjacent to basically things that can be formally verified. Right. Those are ripe for AI systems that can self improve.
- [44:23]
A
Yeah, I think in a lot of cases, sometimes your evaluators need to be made much faster. So as an example, my colleagues did some work maybe a decade ago on some problems in quantum chemistry where you're trying to understand the properties of a particular molecule and you can generate some molecule configuration and then you want to understand what properties it has. And so you can run a very computationally intensive density functional theory simulator, which is something that might take like a night of computation to tell you the answer for one thing. But what my colleagues did was take a bunch of output from those simulation runs, the input molecule configurations and the outputs of the extensive simulator, and then use it to train a neural approximation to the simulator. So this is now a validation device, but instead of it taking a night, they made Something that was 300,000 times faster.
- [45:26]
B
Wow.
- [45:27]
A
And nearly as accurate as running the full scale simulator. So now that completely changes how you would do science. Right. Because now you have 10 million things to screen. You could do that while you go to lunch, rather than it being a six month endeavor where you could try to scrape together enough compute to run all these simulations. And I think there's a lot of room in a lot of domains for much faster validation models, possibly learned validation models that can get you approximation to the true answer much more rapidly. And that changes how those experimental loops can be thought of and how quickly you can go around those loops.
- [46:10]
B
What are some of the spaces and problems that you're super excited that this super sped up scientific method is going to solve or achieve what particular problems or spaces?
- [46:23]
A
Yeah, I mean, I think, well, clearly machine learning itself is one. So can we have a model that is able to recursively self improve itself by running lots of experiments? And if you think about how models are improved today in large research teams, what usually happens is people think of some ideas, they run a bunch of small scale experiments, they see if those small scale experiments worked out well. If so, they take the most promising ones of those and they try them at larger scale and that gets then evaluated and then the results get integrated together into a new recipe for your model. But I think there's no real impediment to making that be a much more automated loop where the model itself decides it's going to explore. Or maybe with a nudge from some people at the very highest level, like, oh, why don't you try some new ideas around model architectures that incorporate this and then it will go run lots of experiments, see which ones work and then those will get incorporated at a much more rapid rate and effectively you want to optimize your discoveries per unit of compute input.
- [47:44]
B
Very cool.
- [47:45]
A
Yeah.
- [47:46]
B
Now going back to the room, as all of you will become at some point founders or start your careers, you will probably collect lots of rejections that will happen. It has happened to you too, Jeff. There's a story that in 2014, you, with Jeff Hinton and Oriole Vinales wrote a paper on distillation. It has to do with taking a big teacher model to train a much smaller and more efficient model that's a lot cheaper to compute less model parameters. And it has become a trick that everyone is using right now in industry. And the thing is, this paper got rejected at neurips.
- [48:36]
A
Yeah, I mean, yeah, I mean, I think I don't fault the program committee because you know, a lot of times a paper gets three reviews and someone will look at one of the reviewers will look at it and in this case they said, oh, it's unlikely to
- [48:51]
B
have significant impact, unlikely to have significant impact.
- [48:53]
A
But you know, I think, you know, when we wrote the paper we actually saw this was a super important problem because we knew making cheaper, highly capable models from larger scale models was something we desperately wanted to do because we wanted to serve models to more and more people in many different domains like speech or vision. But sometimes the reviewer maybe didn't have that experience because maybe they're not thinking about large scale AI services and are thinking about is this a fundamental advance so it gets rejected every so often that's fine, we put it on archive, people read it, people use it, it's all good. And we do use it in making our Flash models for example from our larger scale Pro model. That's partly why our Flash models for example in Gemini are so capable relative to their size and speed.
- [49:47]
B
They're some of the best in the benchmark for their model size class, which is impressive. And I think part of the lesson is that even if you get rejected, keep going.
- [49:56]
A
Yeah, that's the lesson I would distill from that.
- [50:02]
B
Now I think the fun thing is that you basically join when you, when you join Google as a 20 person startup back in 1999. Now if you were to take the young Jeff Dean from way back then to tele transport him to now, today in this era with your skills, I'm
- [50:20]
A
feeling so vigorous and young now, what would you do?
- [50:26]
B
Do you join a frontier lab, start a company? I don't know, what would you do? Jeff Dean today 25 year old Jeff
- [50:36]
A
Dean yeah, I mean it's always hard to say and it's a very personal choice of what it is you want to spend your time on. To me, some of the most important questions are going to work on something you really care about, will you're working on that and if you're able to make progress on it with a bunch of colleagues you like working with, if you're able to make collectively solve it or make progress on it, will that make a difference in the world in some positive way? Right, like will you suddenly be able to do something and offer that service to. Maybe it's a very niche thing, but it will tremendously help biochemists or something or maybe it's a broader thing, it'll help programmers or it will help all consumers on the Internet or other things. What you should strive to do is to have impact in the world that is positive and to work with people you enjoy working with and to work hard and do your best. So in terms of say the particular trade off you offered joining a Frontier Lab versus say starting a company with just one or two or three of you and your close friends. I think those are different experiences. Right. In a large established organization you have some structure, you have lots and lots of amazing colleagues who know lots of things you don't. You, you have lots of interesting problems that you can work on and you already have a platform for impact by your work influencing lots and lots of people in the world already. And then as a very small startup, you have to have something you're passionate about and there's a lot of risk in taking on working on that particular problem in a way that you're going to succeed and you're going to grow an endeavor in order to do that. But that can also be incredibly rewarding, I would imagine. So I think it's really up to personal taste. But at the very least, regardless of what path you take, ask yourself, if I work on this problem and the best possible outcome happens, will the world be a lot better in some way or will the world go, yeah, that's kind of cool, but whatever, that's not the kind of thing you should spend your time on.
- [53:11]
B
Now let's talk a bit about more about that second path of working with people that you really like. In a small team, you've been able to be an incredible mentor and manager to many, many engineers and you've been able to build huge systems. What are some of the lessons for everyone here on how to get the most on how to work with smart people or find smart people?
- [53:37]
A
Yeah, I mean you always want to find people who have really good skills in some area that's needed in a team you're trying to form, whether that's inside a company or starting a company. But you also want to find people that are people you delight being in round right, because you're going to spend a lot of time around people working on really hard problems and you want people who are low ego that are team players that have complementary skills to your own perhaps. I always find working in a small team where people know things that I don't know and where maybe I have some skills that other people don't have as much of, it's super fun because you're collectively building something or working on something that none of you could maybe do individually. But in the process of working on that, you actually gain a lot of new knowledge and new skills for yourself and so do they. And you kind of want to view your engineering or research career as you have an amazing tool belt of techniques and you always want to be Adding new tools to that tool belt because you never know when you might come across a problem where you need these four specialized tools rather than these three. And adding more tools makes it more likely that the problems you encounter in the future will be solvable by you.
- [55:07]
B
Now, one last thing. I'm pretty sure someone in this room or multiple people will eventually build something as consequential as you've done with MapReduce, TPU, distillation, et cetera, et cetera. What problem do you hope they would be working on?
- [55:27]
A
Oh yeah. I mean, I think there's a lot of interesting problems in the world and I'll just rattle off a few. This is not exhaustive because the world is a very big place and full of problems. I'm particularly excited about new approaches to hardware. We've had thought experiment, there was kind of a indication of that or much more efficient inference hardware. I think there are radically different kinds of algorithms for machine learning that might be much more data efficient than the approaches we're using today. If you think about our large scale models today, they probably see 1,000 times as much data as a human does by the age of 18. Get the human by the age of 18 is better in a lot of things and on par with those frontier models that have seen way more data. So could you come up with much more data efficient systems that can learn continuously learn from their own actions? Continual learning is a really interesting thing. I think multi agent interactions is an interesting thing. I think creating ways of having better discourse among people in the world could be interesting. Are there ways to have much more civil conversations and helping people meet other people all over the world that they should know based on their interests? These are kind of interesting things. I think there's lots of cool things in the world and we should all go and strive to make even cooler things occur.
- [56:59]
B
That sounds wonderful. Thank you so much. Jeff Dane, that's all we have today.
- [57:03]
A
Thank you. Appreciate it. Thank you all.
- [57:06]
B
Thank you all.