The prediction and its supporters
According to a report by IT之家, Anthropic CEO Dario Amodei stated that cancer, Alzheimer's, heart disease and mental disorders could become "history" in less than ten years, provided artificial intelligence is used correctly. The prediction is based on the idea that models can substantially accelerate biological research, drug development and medical discoveries.
Amodei is not alone. The article cites Elon Musk and Google's AI chief Demis Hassabis among supporters of similarly optimistic expectations. Hassabis spoke of the possibility of compressing into a few weeks a development path that currently takes years, and of a decade in which diseases could be cured; in support he cites AlphaFold, the system that has predicted more than 200 million protein structures. The same article points out, however, that this result does not demonstrate the ability to develop therapies against complex diseases in ten years.
The significance of the issue goes beyond the statements: if AI tools really reduced pharmaceutical research times, the impact on industry, health systems and patients would be considerable. The point is how much of this scenario has already been verified.
The pharmaceutical industry's skepticism
The most critical voices come from those who develop drugs. Jeff Leiden, executive chairman of Vertex Pharmaceuticals, called these predictions baseless, recounting that his company put about fifty artificial intelligence companies to the test by asking them to select therapeutic candidates from 10,000 compounds: none, according to his account, was able to independently produce a genuinely usable drug.
The limit highlighted by industry players is structural. AI can speed up the initial molecule screening phase, but it does not eliminate the subsequent stages: laboratory tests, animal models, human clinical trials and regulatory approvals. It is these steps that occupy most of the development cycle, which the article generally puts at 10 to 12 years. Finding a promising molecule in a computational model is only the first step.
What is missing and what remains to be verified
The reservations are explicit on the scientific level as well. Cognitive scientist Gary Marcus and cardiologist and digital health expert Eric Topol, cited in the article, believe that current models do not deeply understand biological mechanisms and causal relationships, and therefore cannot yet reliably guide truly innovative therapies. The text adds a contextual fact: as of 2026, no AI-designed drug is in large-scale clinical use.
Also weighing on the discussion is medicine's historical memory: the article recalls the "war on cancer" launched in the United States in the 1970s as an example of very ambitious promises not kept within the announced timeframe. Experts warn that such radical predictions risk fueling public expectations beyond actual capabilities.
In summary, the available facts show an AI with real potential to make research faster, but they offer no proof that it can shorten clinical trial times or that it is close to eradicating complex diseases. Amodei's prediction remains a statement, not a verified result: whether it is well-founded can only be measured with approved therapies and clinical data.
