It isn’t common for Dario Amodei to engage in Twitter disputes. When he does, however, he responds with written essays rather than tweets, and the recent two-part follow-up essay in which he addresses VC Gavin Baker’s claims that Amodei is too doom-laden with his messaging in public forums is an example of why Anthropic’s CEO can either be considered one of the most honest voices in the industry or the most calculated one. The debate began when Baker criticized Amodei for being too doom-laden with his messaging publicly.
Also read: AI companies have not delivered on big promises, says Anthropic CEO
“Saying that AI will cure cancer is more a cliche than it is inspiring, and most people think it is deceptive,” he wrote. “The thing that will work is actually curing cancer.” That’s not something a CEO would want to admit in public. Every AI company on earth, Anthropic included, has spent the better part of three years promising a future of abundance and disease-free longevity. Amodei is now saying, in effect, that his own industry’s favourite talking point has become white noise. People have heard “AI will change everything” so many times it’s stopped meaning anything.
Things get very personal when Amodei starts discussing the Hepatitis C situation. The man lost his father right before the drug called sofosbuvir, which cures the disease in 95 percent of cases, was released. That is not spin – it is a person talking about his own motivations. And it is a story that makes the rest of his message much more compelling than it would have been in an average press release.
Also read: Fireflies.ai Email Assistant launched: Auto drafts, follow-ups and mor
But the real core of his thread is his explanation for the lack of trust towards AI companies and the reasons behind it. The man seems pretty blunt in his statements considering the fact that he manages a billion-dollar laboratory. According to Amodei, the problem with people not trusting AI companies has nothing to do with him or with any other company leaders warning about the potential risks of using such technology (as he puts it, “interview clips tend to be disproportionately negative, as that gets clicks”). In Amodei’s opinion, the problem with distrust is “fundamentally a crisis of trust”. People do not trust any kind of company, government, or industry as a whole.
I will take half of this argument and am quite dubious of the rest of it. Sure, institutional trust has been declining before even the existence of ChatGPT but let us be clear on the fact that it makes perfect sense for a CEO of an AI firm to frame the conversation in such a way since it helps him absolve his companies of any responsibility for the situation. It was bound to happen anyway. Quite convenient, isn’t it?
And here is what Amodei misses completely – a general decline of trust in institutions does not necessarily mean a decrease in distrust in a very specific and potentially dangerous technology created by some specific firms who were never elected to anything. People are not seeing parallels between banks and pharma and some advanced system that can shape up jobs and information structures at an unbelievable speed. This is not some old baggage. This is a very specific worry.
Amodei’s absolutely correct in one final note, however – the criticism that should really be made about Anthropic and the industry in general is not about the tone or messaging. The problem is that none of them have managed to deliver the promised stuff yet. No breakthrough cures for diseases, no abundance, just lots of essays telling about what a beautiful future awaits us all. This is the hallmark. Amodei may write any number of reflective X threads, but trust is built in responses to VCs not through replying, but through delivery.
Also read: Researchers are extracting AI reasoning traces from Claude, GPT and Gemini: Here’s how