Polish programmer Psyho defeated the OpenAI model in Tokyo in the most important programming competition - and he puts it bluntly: today we don't know the methods that guarantee AI obedience. In a candid conversation, he tells how he won against the machine, why the technology race is unstoppable, what the alignment problem really is, where the information chaos around AI comes from, and why a single mistake with super-AI could be our last. The behind-the-scenes of working at OpenAI (Dota 2 project), leaving in 2020, life without a degree, creative freedom and plans: education and support of Polish technology companies.