Published on January 21, 2026 10:30 PM GMT

It is interesting that as soon as I distract myself from the field of AI risks for just a week or two, my psyche starts playing the same record called “everything will be fine, don’t worry.”

This can perhaps be explained by the fact that I have been studying AI only relatively recently (since the beginning of last year). And probably my concern about existential risks has not yet had enough time to become a routine for my brain.

But nevertheless, it is interesting that even after persistently studying the arguments of Yudkowsky, Bostrom, Bengio, and others for (with breaks) nine months, I am still able to simply forget them.

Ordinary life, in which Gemini can still say that the President of the United States is still Biden, pulls me in and shifts my focus of attention to things far less urgent than carrying out all possible actions to ensure the survival of the human species.

It is possible that a similar tendency of the psyche also forces people working in AI companies to neglect safety. They may be too involved in the race / competition / specific technical tasks, or simply in everyday life problems.

But I am lucky. Something inside me, something rational, simply remembers the numbers: what if we have only 10 years left? What if 5? Or maybe 3?

And these numbers wake me up. They oblige me to remember everything I have studied recently.

They oblige me to ask further questions in order to understand that the anxiety which from time to time goes into the fog is real and objective.