For the best viewing experience, download the Substack app ⬇️
Friends,
Earlier this week, we shared an excerpt from Reid Hoffman’s new book, Superagency: What Could Possibly Go Right with Our AI Future. Today, I thought we’d offer up something new: a ten-minute highlight reel of the conversation I had with Reid this week on The Next Big idea. Have a listen to the full audio of that chat, available on Apple Podcasts and Spotify, or, if you’re short on time, check out the video (above). Let us know in the comments if this kind of condensed video summary is something you’d like to see us do more often.
Reid, as you may know, is one of a limited number of people with real visibility into the inner workings of Silicon Valley. He co-founded LinkedIn, sold it to Microsoft for $26 billion, and currently sits on the Microsoft board. He was among the early backers of OpenAI, co-founded Inflection AI with Mustafa Suleyman, and has been called the most connected man in Silicon Valley.
Whereas many see AI as a threat to human autonomy and purpose, Reid sees it as the opposite — the most powerful tool available to broaden human agency and improve our shared future.
Reid and I discuss, among other topics:
AI’s promise as an “always-on” team of advisors for everyone
The positive impact of increasing humanity's EQ and IQ
The role of AI in creativity
The dawn of “agentic AI”
The confluence of AI, robots, and renewable energy
My concern about the risks: less human interaction, totalitarianism, and the problems of alignment and containment
Why Reid, who supported Kamala Harris, is happy to see the close relationship between Silicon Valley and the White House
Why it’s essential that we all engage with AI and think through the future we want
It’s a fascinating and timely conversation. Just after we spoke, news about the performance of DeepSeek, the Chinese AI upstart, sent tech stocks spiraling. I asked Reid for his take, and he followed up with a voice memo saying:
“Well, it means that the acceleration of the proliferation of AI will be even faster because you'll have high quality models on many more devices, operating independently and probably outside of governance circumstances. People want to tell the story that the big models don't matter, but the big models will still be like — if you could do something as a small model, you can do it better as a big model. And so the big models are actually, in fact, very important. If you have a 10% or 20% better coder, better lawyer, better doctor, that's actually worth a lot of money In the training, because you amortize it across all of its use, potentially across billions of people over a year. So that's why it's so valuable to even spend tens of billions of dollars for 20% better if you get there. The other part of it is almost for sure, all of the training of these smaller models will be greatly enabled, accelerated and enhanced by larger models.”
Reid’s perspective dovetails with Anthropic CEO Dario Amodei's recent post, which explains why AI requires both massive investment in computing scale, and software innovation like that demonstrated by DeepSeek. There is nothing particularly surprising about DeepSeek’s progress, according to Dario, except that it’s a Chinese company rather than an American one driving innovation forward.
I have found Reid Hoffman and Dario Amodei to be the two most sensible voices inside the industry on the topic of our AI future. They both see US leadership in the AI revolution as critical to the future of democracy. My conversation with Yuval Noah Harari really drove home the stakes of what might come to pass if China takes the lead in this race.
I was interested to learn that though Reid did not support Trump in this election, he is pleased to see the current administration working closely with tech leaders. Politics aside, we need government and industry collaborating.
The DeepSeek curveball supports Reid’s refrain: “the future is sooner — and stranger — than you think.”
Share this post