“At this point, saying that AI will cure cancer is more a cliché than it is inspiring, and most people think it is deceptive. … The thing that will work is actually curing cancer.” – Dario Amodei, co-founder and CEO of Anthropic
For an industry built partly on extraordinary promises, Dario Amodei has proposed a straightforward test: Start delivering.
The Anthropic CEO made the comment on August 15 during an online exchange with tech investor Gavin Baker about the growing backlash against artificial intelligence. As TechCrunch reported, Baker had argued on X that Amodei and other AI leaders, by repeatedly warning about the technology’s risks, have helped create some of the distrust and regulatory pressure the industry now faces.
Amodei pushed back on that characterization. He said his messaging has not been disproportionately negative and pointed to his 2024 essay Machines of Loving Grace, an expansive argument for the potential benefits of advanced AI in medicine, science, economic development, and other fields. He argued instead that AI has become caught up in a broader “crisis of trust,” with many people skeptical of corporations, government, and the technology industry generally.
That is where Amodei’s line about cancer came in. “At this point, saying that AI will cure cancer is more a cliché than it is inspiring,” he wrote. “The most accurate criticism of AI companies, including Anthropic, is that we haven’t yet delivered on our big promises to benefit the world.” Then he added: “The thing that will work is actually curing cancer.”
The remark is striking because it shifts the conversation away from messaging and toward results. AI companies have spent the past several years making enormous claims about what the technology may eventually do: accelerate scientific discovery, revolutionize medicine, transform education, dramatically increase productivity, and help solve problems that have resisted human ingenuity for decades. Amodei has been among those making the case that sufficiently powerful AI could produce unusually rapid advances across several of those areas.
In Machines of Loving Grace, Amodei argued that powerful AI could produce what he called a “compressed 21st century” in biology and health, potentially accomplishing in five to 10 years advances that might otherwise take 50 to 100 years. He discussed the possibility of dramatically improving the prevention and treatment of cancer, infectious disease, genetic disorders, and other major health problems.
Those promises once sounded almost impossibly futuristic, but by now the claims are also becoming overly familiar. Meanwhile, the public’s experience of AI is much more mixed and immediate. People see useful chatbots and impressive new tools, but they also see deepfakes, workplace disruption, copyright battles, enormous data centers, growing electricity and water demands, and companies pouring enormous sums into an AI infrastructure boom whose ultimate economic payoff remains uncertain.
That gap between promise and experience may be one reason Amodei’s comment lands so well. When ChatGPT arrived on November 30, 2022, the technology itself felt like real proof that something very significant had happened. Nearly four years later, increasingly ambitious claims about what AI will someday accomplish naturally invite a harder question: What has it actually accomplished so far?
Amodei seems to recognize that the burden of proof is changing. In the same post, he said that Anthropic is increasing its efforts in biology and medicine and expects major results in the years ahead, along with what he called “some early glimmers in the coming months.” If the company accomplishes something genuinely important, he said, “the whole world will hear about it.”
That focus on biology is not merely rhetorical. Anthropic has been building out its work in the field, including the acquisition of biotech startup Coefficient Bio and the appointment of Novartis CEO Vas Narasimhan to Anthropic’s board. Amodei has said the goal is not for Anthropic itself to become a drug company, but to use AI tools throughout the pharmaceutical pipeline, from generating hypotheses about disease to identifying potential drugs and improving clinical trials.
Of course, even major scientific breakthroughs may not solve AI’s trust problem by themselves. Business Insider reported that Angel Brodin, an applied AI architect at OpenAI, responded to Amodei by pointing to the pharmaceutical industry. As Brodin put it, companies can produce medicines that save lives and still remain deeply distrusted. People will judge AI companies not only by the products they create, but also by such issues as pricing, access, lobbying, transparency, how the economic gains are distributed, and who ultimately holds power.
That is an important counterpoint. Trust is more complicated than technological achievement. Still, Amodei’s larger point deserves attention because it captures a transition that may be underway across the AI industry. For several years, companies have asked the public to take seriously their predictions about what AI could become. Some of those predictions involve extraordinary prosperity. Others involve extraordinary danger. Nearly all assume that AI will rank among the most defining technologies of this century.
At some point, though, the promises themselves become less persuasive. The questions shift from what AI might do to what it has done, who has benefited, and whether the results justify the disruption that has come with it.
If AI companies want the public to believe they can change the world, eventually they have to show that they can.
