Trump forgot the word "Alzheimer's": will AI start screening politicians for mental fitness?

Irony can be a little too on the nose. Speaking to supporters, Donald Trump once again set out to stress his cognitive sharpness, but tripped over the most important word. He literally could not recall the name of Alzheimer's disease while describing how well he had passed tests showing he does not have it. For the media, it is one more occasion for jokes. For the AI industry, it is a serious case study in how health-monitoring technology is changing our idea of public life.

Let's be honest: we have entered an era of "gerontocracy," in which key decisions are made by people whose age is well past the point when cognitive functions begin to fade. Where we once relied on official statements from personal physicians, who always report everything "in perfect order," we now have an independent arbiter: large language models and speech-analysis algorithms. For several years now, researchers have been running recordings of public figures' speeches through neural networks to track changes in their mental state.

How does it work in practice? AI analyzes so-called "semantic density" and vocabulary diversity. When the brain starts losing neural connections, a person subconsciously simplifies sentence structure, uses more fillers (the "um," "well," "what's it called") and replaces specific nouns with generic terms. Trump, forgetting the word "Alzheimer's," displayed a classic symptom of anomia, in which a person understands the concept but cannot retrieve its name from memory. For a neural network this is a clear signal, and it registers it instantly, unlike a loyal electorate.

Interestingly, similar methods of analysis have already been applied to Ronald Reagan's late speeches. Retrospective analysis with modern algorithms showed that signs of the disease were noticeable in his speech years before doctors made an official diagnosis. Today we do not have to wait decades. We can feed debate transcripts to models such as GPT-4 or specialized medical AI in real time and get a chart of cognitive decline. That puts a heavy ethical question to society: do we have the right to demand a "digital diagnosis" from those who run the state?

The problem is that AI has no empathy and no political preferences. It does not care whom you voted for. It simply sees that the frequency of repeated words has risen by 15% and that logical connectives have become shorter. This creates a new reality in which a politician can no longer hide behind makeup or clever editing. Every live appearance becomes a voluntary lab test. In the coming years we will see tools that overlay a "sanity index" directly on top of a live broadcast.

Of course, critics will say this is an invasion of privacy and a breach of medical confidentiality. But when the person in question is someone whose momentary confusion could cost global stability, the boundaries of privacy blur. We already trust AI to fly planes and diagnose cancer, so why not trust it to monitor those at the helm? The Trump case is only the tip of the iceberg, a reminder that the human brain is extremely fragile "hardware" that needs constant auditing.

The big question: Will automated cognitive scoring become a mandatory condition for running in elections, in an era when AI knows more about our health than we do?