The most well-founded complaint against AI companies is the mismatch between loud promises and real results. This statement was made by Anthropic CEO Dario Amodei, linking society's negative perception of the technology to a deep crisis of trust that has been building for decades.
The occasion for the discussion was a post by investor Gavin Baker, who accused Amodei of overly pessimistic rhetoric and suggested he adopt a more positive tone. According to Baker, the head of Anthropic "lost the argument" on AI regulation and should more actively promote the benefits of the technology, especially ahead of the company's entry into the public market.
The essence of the conflict: marketing versus reality
Amodei responded with a series of posts on X, emphasizing that he usually avoids social media, but here the issue is a matter of principle. He noted that he pays equal attention to both the risks and benefits of AI, recalling his essay "Machines of Loving Grace," dedicated to the technology's potential in medicine and economics. However, the main thesis of his response was an acknowledgment of a systemic problem.
"At the core is a fundamental crisis of trust. People don't trust companies, governments, or the tech industry, suspecting that we are looking for new ways to deceive them," Amodei stated. At the same time, he took responsibility for this situation: "This is entirely on us. We should be criticized for unfulfilled promises, not for marketing tricks."
"Only one thing will work — actually curing cancer"
The head of Anthropic emphasized that a new advertising campaign will not change society's attitude. Claims that AI will defeat cancer have become a cliché perceived as deception. The only way to restore trust is to demonstrate measurable achievements. Amodei announced intensified work in biology and medicine, promising "incredible results" in the coming years, but without specific details.
Not everyone agreed with his position. AI researcher Yann LeCun has insisted for about a decade that the only path is open and widely accessible technology. According to him, AI, like the printing press and the internet, amplifies human intelligence, expanding access to knowledge. "We need diverse AI for the same reason we need a diverse press," he wrote, also supporting Mark Zuckerberg's position that the idea of dangerous AI requiring centralized power is problematic.
This approach diverges from Anthropic's policy, which does not call for a complete ban on open models but insists on stricter safety requirements as their capabilities grow. In turn, OpenAI GTM specialist Angel Brodin pointed to pharmaceuticals as an example of an industry with real achievements that nonetheless suffers from low public trust. In her view, attitudes toward AI will depend not only on results but also on pricing, transparency, and the distribution of economic benefits.
My analysis: Amodei's statement is a rare case of honest self-criticism in an industry overheated by hype. However, betting on "real results" in medicine is a long-term game that may not align with the expectations of the market and investors. Until AI solves specific, tangible problems for ordinary people, the crisis of trust will only worsen — and here, proponents of openness are right: transparency and accessibility could become a faster and more effective tool for restoring reputation than promises of future breakthroughs.