AI + a16z
a16z
1 Listener
All episodes
Best episodes
Top 10 AI + a16z Episodes
Goodpods has curated a list of the 10 best AI + a16z episodes, ranked by the number of listens and likes each episode have garnered from our listeners. If you are listening to AI + a16z for the first time, there's no better place to start than with one of these standout episodes. If you are a fan of the show, vote for your favorite AI + a16z episode by adding your comments to the episode page.
06/07/24 • 37 min
In this episode, Ideogram CEO Mohammad Norouzi joins a16z General Partner Jennifer Li, as well as Derrick Harris, to share his story of growing up in Iran, helping build influential text-to-image models at Google, and ultimately cofounding and running Ideogram. He also breaks down the differences between transformer models and diffusion models, as well as the transition from researcher to startup CEO.
Here's an excerpt where Mohammad discusses the reaction to the original transformer architecture paper, "Attention Is All You Need," within Google's AI team:
"I think [lead author Asish Vaswani] knew right after the paper was submitted that this is a very important piece of the technology. And he was telling me in the hallway how it works and how much improvement it gives to translation. Translation was a testbed for the transformer paper at the time, and it helped in two ways. One is the speed of training and the other is the quality of translation.
"To be fair, I don't think anybody had a very crystal clear idea of how big this would become. And I guess the interesting thing is, now, it's the founding architecture for computer vision, too, not only for language. And then we also went far beyond language translation as a task, and we are talking about general-purpose assistants and the idea of building general-purpose intelligent machines. And it's really humbling to see how big of a role the transformer is playing into this."
Learn more:
Investing in Ideogram
Denoising Diffusion Probabilistic Models
Follow everyone on X:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
1 Listener
06/14/24 • 43 min
In this episode, Inngest cofounder and CEO Tony Holdstock-Brown joins a16z partner Yoko Li, as well as Derrick Harris, to discuss the reality and complexity of running AI agents and other multistep AI workflows in production. Tony also why developer tools for generative AI — and their founders — might look very similar to previous generations of these products, and where there are opportunities for improvement.
Here's a sample of the discussion, where Tony shares some advice for engineers looking to build for AI:
"We almost have two parallel tracks right now as, as engineers. We've got the CPU track in which we're all like, 'Oh yeah, CPU-bound, big O notation. What are we doing on the application-level side?' And then we've got the GPU side, in which people are doing like crazy things in order to make numbers faster, in order to make differentiation better and smoother, in order to do gradient descent in a nicer and more powerful way. The two disciplines right now are working together, but are also very, very, very different from an engineering point of view.
"This is one interesting part to think about for like new engineers, people that are just thinking about what to do if they want to go into the engineering field overall. Do you want to be on the side using AI, in which you take all of these models, do all of this stuff, build the application-level stuff, and chain things together to build products? Or do you want to be on the math side of things, in which you do really low-level things in order to make compilers work better, so that your AI things can run faster and more efficiently? Both are engineering, just completely different applications of it."
Learn more:
The Modern Transactional Stack
Follow everyone on X:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
1 Listener
07/05/24 • 33 min
In this archive episode from 2015, a16z's Sonal Chokshi, Frank Chen, and Steven Sinofsky discuss DeepMind's breakthrough AlphaGo system, which mastered the ancient Chinese game Go and introduced the public to reinforcement learning.
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
1 Listener
05/17/24 • 40 min
a16z partners Guido Appenzeller and Matt Bornstein join Derrick Harris to discuss the state of the generative AI market, about 18 months after it really kicked into high gear with the release of ChatGPT — everything from the emergence of powerful open source LLMs to the excitement around AI-generated music.
If there's one major lesson to learn, it's that although we've made some very impressive technological strides and companies are generating meaningful revenue, this is still a a very fluid space. As Matt puts it during the discussion:
"For nearly all AI applications and most model providers, growth is kind of a sawtooth pattern, meaning when there's a big new amazing thing announced, you see very fast growth. And when it's been a while since the last release, growth kind of can flatten off. And you can imagine retention can be all over the place, too . . .
"I think every time we're in a flat period, people start to think, 'Oh, it's mature now, the, the gold rush is over. What happens next?' But then a new spike almost always comes, or at least has over the last 18 months or so. So a lot of this depends on your time horizon, and I think we're still in this period of, like, if you think growth has slowed, wait a month and see it change."
Follow everyone on X:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
1 Listener
07/03/24 • 65 min
In this episode of the AI + a16z podcast, Luma Chief Scientist Jiaming Song joins a16z General Partner Anjney MIdha to discuss Jiaming's esteemed career in video models, culminating thus far in Luma's recently released Dream Machine 3D model that shows abilities to reason about the world across a variety of aspects. Jiaming covers the history of image and video models, shares his vision for the future of multimodal models, and explains why he thinks Dream Machine demonstrates its emergent reasoning capabilities. In short: Because it was trained on a volume of high-quality video data that, if measured in relation to language data, would amount to hundreds of trillions of tokens.
Here's a sample of the discussion, where Jiaming explains the "bitter lesson" as applied to training generative models, and in the process sums up a big component of why Dream Machine can do what it does by using context-rich video data:
"For a lot of the problems related to artificial intelligence, it is often more productive in the long run to use methods that are simpler but use more compute, [rather] than trying to develop priors, and then trying to leverage the priors so that you can use less compute.
"Cases in this question first happened in language, where people were initially working on language understanding, trying to use grammar or semantic parsing, these kinds of techniques. But eventually these tasks began to be replaced by large language models. And a similar case is happening in the vision domain, as well . . . and now people have been using deep learning features for almost all the tasks. This is a clear demonstration of how using more compute and having less priors is good.
"But how does it work with language? Language by itself is also a human construct. Of course, it is a very good and highly compressed kind of knowledge, but it's definitely a lot less data than what humans take in day to day from the real world . . .
"[And] it is a vastly smaller data set size than visual signals. And we are already almost exhausting the . . . high-quality language sources that we have in the world. The speed at which humans can produce language is definitely not enough to keep up with the demands of the scaling laws. So even if we have a world where we can scale up the compute infrastructure for that, we don't really have the infrastructure to scale up the data efforts . . .
"Even though people would argue that the emergence of large language models is already evidence of the scaling law . . . against the rule-based methods in language understanding, we are arguing that language by itself is also a prior in the face of more of the richer data signal that is happening in the physical world."
Learn more:
Follow everyone on X:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
04/26/24 • 36 min
Pinecone Founder and CEO Edo Liberty joins a16z's Satish Talluri and Derrick Harris to discuss the promises, challenges, and opportunities for vector databases and retrieval augmented generation (RAG). He also shares insights and highlights from a decades-long career in machine learning, which includes stints running research teams at both Yahoo and Amazon Web Services.
Because he's been at this a long time, and despite its utility, Edo understands that RAG — like most of today's popular AI concepts — is still very much a progress:
"I think RAG today is where transformers were in 2017. It's clunky and weird and hard to get right. And it has a lot of sharp edges, but it already does something amazing. Sometimes, most of the time, the very early adopters and the very advanced users are already picking it up and running with it and lovingly deal with all the sharp edges ...
"Making progress on RAG, making progress on information retrieval, and making progress on making AI more knowledgeable and less hallucinatory and more dependable, is a complete greenfield today. There's an infinite amount of innovation that will have to go into it."
More about Pinecone and RAG:
Retrieval Augmented Generation (RAG)
Emerging Architectures for LLM Applications
Follow everyone on X:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
Remaking the UI for AI
AI + a16z
04/19/24 • 38 min
a16z General Partner Anjney Midha joins the podcast to discuss what's happening with hardware for artificial intelligence. Nvidia might have cornered the market on training workloads for now, but he believes there's a big opportunity at the inference layer — especially for wearable or similar devices that can become a natural part of our everyday interactions.
Here's one small passage that speaks to his larger thesis on where we're heading:
"I think why we're seeing so many developers flock to Ollama is because there is a lot of demand from consumers to interact with language models in private ways. And that means that they're going to have to figure out how to get the models to run locally without ever leaving without ever the user's context, and data leaving the user's device. And that's going to result, I think, in a renaissance of new kinds of chips that are capable of handling massive workloads of inference on device.
"We are yet to see those unlocked, but the good news is that open source models are phenomenal at unlocking efficiency. The open source language model ecosystem is just so ravenous."
More from Anjney:
The Quest for AGI: Q*, Self-Play, and Synthetic Data
Making the Most of Open Source AI
Safety in Numbers: Keeping AI Open
Follow everyone on X:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
05/24/24 • 65 min
For this holiday weekend (in the United States) episode, we've stitched together two archived episodes from the a16z Podcast, both featuring General Partner Anjney Midha. In the first half, from December, he speaks with Mistral cofounder and CEO Arthur Mensch about the importance of open foundation models, as well as Mistral's approach to building them. In the second half (at 34:40), from February, he speaks with Stanford's Stefano Ermon about the state of the art in video models, including how OpenAI's Sora might work under the hood.
Here's a sample of what Arthur had to say about the debate over how to regulate AI models:
"I think the battle is for the neutrality of the technology. Like a technology, by a sense, is something neutral. You can use it for bad purposes. You can use it for good purposes. If you look at what an LLM does, it's not really different from a programming language. . . .
"So we should regulate the function, the mathematics behind it. But, really, you never use a large language model itself. You always use it in an application, in a way, with a user interface. And so, that's the one thing you want to regulate. And what it means is that companies like us, like foundational model companies, will obviously make the model as controllable as possible so that the applications on top of it can be compliant, can be safe. We'll also build the tools that allow you to measure the compliance and the safety of the application, because that's super useful for the application makers. It's actually needed.
"But there's no point in regulating something that is neutral in itself, that is just a mathematical tool. I think that's the one thing that we've been hammering a lot, which is good, but there's still a lot of effort in making this strong distinction, which is super important to understand what's going on."
Follow everyone on X:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
ARCHIVE: GPT-3 Hype
AI + a16z
05/01/24 • 33 min
In this episode, though, we’re traveling back in time to distant — in AI years, at least — past of 2020. Because amid all the news over the past 18 or so months, it’s easy to forget that generative AI — and LLMs, in particular — have been around for a while. OpenAI released its GPT-2 paper in late 2018, which excited the AI research community, and in 2020 made GPT-3 (as well as other capabilities) publicly available for the first time via its API. This episode dates back to that point in time (it was published in July 2020), when GPT-3 piqued the interest of the broader developer community and people really started testing what was possible.
And although it doesn’t predict the precambrian explosion of multimodal models, regulatory and copyright debate, and entrepreneurial activity that would hit a couple of years later — and who could have? — it does set the table for some of the bigger — and still unanswered — questions about what tools like LLMs actually mean from a business perspective. And, perhaps more importantly, what they ultimately mean for how we define intelligence.
So set your wayback machine to the seemingly long ago summer of 2020 and enjoy a16z’s Sonal Chokshi and Frank Chen discussing the advent of commercially available LLMs.
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
05/15/24 • 22 min
In this bonus episode, recorded live at our San Francisco office, security-startup founders Dean De Beer (Command Zero), Kevin Tian (Doppel), and Travis McPeak (Resourcely) share their thoughts on generative AI, as well as their experiences building with LLMs and dealing with LLM-based threats.
Here's a sample of what Dean had to say about the myriad considerations when choosing, and operating, a large language model:
"The more advanced your use case is, the more requirements you have, the more data you attach to it, the more complex your prompts — ll this is going to change your inference time.
"I liken this to perceived waiting time for an elevator. There's data scientists at places like Otis that actually work on that problem. You know, no one wants to wait 45 seconds for an elevator, but taking the stairs will take them half an hour if they're going to the top floor of . . . something. Same thing here: If I can generate an outcome in 90 seconds, it's still too long from the user's perspective, even if them building out and figuring out the data and building that report [would have] took them four hours . . . two days."
Follow everyone:
Check out everything a16z is doing with artificial intelligence here, including articles, projects, and more podcasts.
Show more best episodes
Show more best episodes
FAQ
How many episodes does AI + a16z have?
AI + a16z currently has 24 episodes available.
What topics does AI + a16z cover?
The podcast is about Venture Capital, Entrepreneurship, Startups, Podcasts, Technology, Business, Artificial Intelligence and Machine Learning.
What is the most popular episode on AI + a16z?
The episode title 'The Future of Image Models Is Multimodal' is the most popular.
What is the average episode length on AI + a16z?
The average episode length on AI + a16z is 41 minutes.
How often are episodes of AI + a16z released?
Episodes of AI + a16z are typically released every 7 days.
When was the first episode of AI + a16z?
The first episode of AI + a16z was released on Apr 8, 2024.
Show more FAQ
Show more FAQ