Dario Amodei acknowledges that AI companies have failed to fulfill their commitments.

      Dario Amodei noted on X that a primary criticism of AI companies, including Anthropic, is their failure to fulfill promises, asserting that only true breakthroughs will earn public trust. This statement comes as Anthropic is set to list, with backers anticipating a valuation of $2 trillion.

      Amodei acknowledged the criticisms directed at his industry. “The most accurate critique of AI companies, including Anthropic, is that we haven’t yet fulfilled our major promises to improve the world,” the CEO said on X Saturday.

      His response is more direct than the acknowledgment. “At this stage, claiming that AI will cure cancer has become more cliché than inspiring, and many view it as misleading,” he stated. “What will genuinely work is actually providing a cure for cancer.”

      The timing of his remarks adds significance. Anthropic is weeks away from a listing that supporters expect to value the firm at $2 trillion in October, a valuation heavily reliant on the anticipated capabilities of the technology rather than past achievements.

      Data supports his assertion. Pew found that half of American adults express more concern than excitement about AI in their daily lives, up from 37% in 2021 when the question was first posed, compared to just 10% who feel more excitement.

      However, he does not believe his warnings are to blame. “I do not think my messaging has been overly negative,” he commented, referencing his 2024 essay, Machines of Loving Grace, which suggests that AI could positively transform the world.

      His rationale is broader and challenging to verify. He argued that the public's distrust of AI companies is a reflection of longstanding skepticism towards corporations in general, noting that “AI is merely the latest version of this.”

      Others are not convinced that breakthroughs alone are sufficient. An employee from OpenAI responded that the pharmaceutical industry continuously makes advancements yet remains one of the least trusted sectors.

      She stated that people will evaluate AI companies based on more than just outcomes, mentioning factors like pricing, access, lobbying, transparency, how benefits are distributed, and who holds the power.

      Yann LeCun presented the perspective that Anthropic is least inclined to embrace, arguing that progress hinges on AI being “widely available, shared, and open.” This stance is one that Anthropic opposes concerning its frontier models, while Meta advocates for the opposite in Washington.

      Amodei concluded with a timeline, expressing that Anthropic anticipates “incredible results in the coming years and some early signs in the months ahead,” which suggests the listing is projected to occur before these indications materialize.