Dario Amodei, CEO and co-founder of Anthropic
Insights on AI Progress, Governance, and the Future
Last week, on a chilly day in Menlo Park, we had the pleasure of hosting a long-awaited dinner with Dario Amodei, the CEO and co-founder of Anthropic as part of Generative AI series at SV Icons.
Anthropic is an artificial intelligence startup and public-benefit corporation that aims to develop safe and beneficial artificial intelligence for the long-term benefit of humanity.
Dario shared with us his thoughts on AI progress, governance, and the future. Here are some of them:
There will be exponential progress in AI due to scaling laws. As models get bigger and training data grows, performance improves dramatically. Anthropic has seen huge leaps from scaling Claude. While new architectures may emerge, scaling existing models could achieve superhuman abilities in some domains in 2-5 years.
Dario advocates for “guardrails” and regulation before models become too powerful. He sees a role for governments to moderate progress, due to a “prisoner’s dilemma” where companies individually have incentive to move fast. Dario also talked about his continuous cooperation with the regulators, and how international cooperation will be critical to avoid an AI arms race.
Dario expects AI to be widely used for disintermediated and intermediated applications. The interfaces may evolve to include video and voice. He also believes that while jobs will change, the exact impact is hard to predict. Dario suggests that skills requiring human presence and touch will be hardest to disrupt.
There are great opportunities in AI for “picks and shovels” companies providing data, chips, infrastructure and services. Anthropic aims to benefit humanity through partnerships and as a public benefit corporation. Public Benefit Corporation status of Anthropic means that it has a corporate purpose that goes beyond maximizing shareholder profits.
Overall, Dario sees AI testing society's ideologies and values. But at the same time, he believes that technology is shaped by human choices. With care and cooperation, he says, we can build AI that amplifies the best in us and builds a better future for all.
Despite being at the forefront of Generative AI, even Dario finds it challenging to accurately predict what the next three years will bring in terms of developments and advancements.
We live in an exciting time!
Organizer and host: Aizada Marat & Chyngyz Dzhumanazarov
Contributor: David Vernal