The artificial intelligence industry is experiencing a “Cambrian explosion” of applications, said Sequoia Capital Partner Sonya Huang.
(Bloomberg) — The artificial intelligence industry is experiencing a “Cambrian explosion” of applications, said Sequoia Capital Partner Sonya Huang.
That’s because unlike the overhyped startups of cycles past, AI will represent a fundamental change in the way much of technology works, she said. “Whenever there is a platform shift, it opens the door for founders to build new and daring legendary companies,” Huang told Bloomberg Television on Wednesday. “And that’s exactly the gold rush that we’re seeing right now in the startup ecosystem.”
At Sequoia, one of Silicon Valley’s most prestigious firms, Huang specializes in artificial intelligence startups across North America and Europe. “Right now, the story of these companies is incredibly promising,” she said, citing a startup called Ironclad Inc. that’s used OpenAI’s GPT models to make it “possible for us automate away the creation and the analysis of legal contracts.” She’s also seen companies that aim to automate video game and film creation using the same technology.
In September, weeks before ChatGPT made large language models a household name, Huang wrote a blog post outlining the possibilities. Think AI-generated text that’s better than professional writers, video games created almost as fast as their creators can dream them up, and higher quality images than can now be created by professional artists. Huang expects all of this by 2030. Her report was co-authored by GPT-3 (naturally).
Even as AI bulls tout the technology’s potential, the nascent industry is already generating controversy. ChatGPT maker OpenAI said on Thursday it would work to reduce biases after a spate of inappropriate reactions from its tools. It’s raised the question of how enduring the current excitement over AI will be. While prior tech hype cycles have proved to be a “head-fake,” Huang doesn’t expect a wipeout of the current wave of AI companies: “This time it’s not vaporware.”
Here’s what else Huang had to say (edited for length and clarity):
Bloomberg: A lot’s happened since September, do you stand by the thesis you outlined? Or have things accelerated?Sonya Huang: We think this is a watershed moment for technology. If we think about previous platform shifts like going from desktop to mobile, going from on-premise to SaaS [software as a service], we think that this one is even larger in magnitude.
I want to just take a moment to underscore the significance of what is happening right now in technology. We now have models that are capable of writing, of creating, of conversing at human or even superhuman levels of intelligence. That’s something that computer scientists have been dreaming of since the 1950s and it’s finally arrived. And I think the impacts are going to be profound. Every industry is getting reinvented by AI.
You call it a gold rush. It has been called a frenzy. There are publicly traded AI stocks that are up triple digits year to date. Is this overhyped or is the hype justified?
I think there’s hype and there’s substance, and we try to separate the two. Hype comes and goes. It kind of creates this crazy animal spirit. You can’t control it. And so for our part, we try to ignore what happens with hype and focus on the substance. And when it comes to substance, it is just stunning to see some of the results in generative AI.
You’ve called it a “Large Model Moore’s Law.” Where in the next decade does it take us? What will we be able to do?
The pace of progress and R&D on the large models has just been breathtaking. And I think it’s only going to accelerate. You see all the announcements that Google and Microsoft have been making, there is an arms race happening right now as it comes to training and investing in these large models, both the algorithms and the compute and the data. And I think that benefits the entire ecosystem. All these applications rely on great foundational models. And as the foundational models get larger and larger, their emerging capabilities just — are blowing everybody away.
How do these companies make money? Will they ever monetize to a level where they change the game economically speaking?
OpenAI deserves a ton of credit for taking what happened with transformers and large language models and taking it out of the research labs of these big tech companies and turning it into a real arms race.
They have the results: GPT3, GPT3.5, ChatGPT, Dall-E. These results are stunning. To your question, I actually think that the business model is fairly straightforward. It is intelligence as an API, just as we’re in an API economy. Just as you buy Stripe over API, you buy Twilio over API. A lot of the companies we’re working with need to rely on these foundational models in order to run their goods and services. And a lot of them are looking to pay OpenAI for that. So the same way you’d pay for an API called Stripe, you pay for an API called OpenAI.
How do you feel about regulation and the focus on ethics that we’re getting to around artificial intelligence?
I think all of these questions are very real, they’re very thorny. But some of them are getting solved as we speak. So, for example, hallucinations, which is the tendency of these models to make up things on the fly: That’s getting solved by these foundational model companies. I wouldn’t be surprised if in the next six to 12 months we have models that are actually capable of truthfulness. There are other questions over ethics, over copyright. These are sort of hairier problems that require all the stakeholders together on the table and agree on what’s fair.
More stories like this are available on bloomberg.com
©2023 Bloomberg L.P.