Roman Yampolskiy argues that reality is likely a simulation and that the Big Bang may mark the simulation starting up. His more market-relevant message is that general superintelligence should not be built because it will not be controllable, while narrow AI tools remain useful.
Watch on YouTube ›Get the market thesis, key claims, assets, contradictions, and follow-up questions from any financial video — then unlock a version personalized to your portfolio, watchlist, and favorite speakers.
This interview centers on two linked themes: simulation theory and the AI control problem. Yampolskiy argues that our universe may be a simulation, interpreting the speed of light as a processor/rendering limit and the Big Bang as the simulation being switched on. He says there is no way to prove ultimate reality with certainty, only to know what is experienced locally. Even if reality is simulated, he stresses that pain, suffering, and moral responsibility remain real to the beings inside it. The conversation then turns to artificial intelligence. Yampolskiy draws a firm distinction between narrow AI tools and general superintelligence. He is supportive of domain-specific systems for things like cancer research, image generation, learning, and productivity. …
Near term, the practical risk is overextending today’s AI agents with tool access, memory, and autonomy before controls are proven. The immediate watchpoint is whether model behavior begins to create visible operational incidents that force tighter containment.
Over the next few months, the key question is whether safety claims can keep up with capability growth. If not, the narrative should tilt toward slower deployment, stronger supervision, and more skepticism around autonomous AI products.
The structural thesis is that general superintelligence would break the assumption that humans remain the top decision-makers. If that is right, the durable winners are bounded tools and constrained automation, not open-ended autonomous systems.
The universe is likely a simulation, with the Big Bang representing the system turning on.
He repeatedly says the simulation is very likely and interprets the Big Bang as a startup event.
There is no test that can give certainty that reality is authentic or not simulated.
He says authenticity cannot be proven, only falsity can sometimes be shown.
General superintelligence should not be built because it cannot be safely controlled.
This is his central AI safety thesis throughout the interview.
What are your biggest unanswered questions about the simulation?
The guest answers: 'What's outside the simulation?'
Would it not be great to create a simulation such as the life we're living now to test AI?
The guest says that is exactly what they think is happening — we are in a simulation right now, and the most likely reason for this specific time to be alive is that it is the most interesting time, as we are creating new worlds, virtual reality, and learning how to create intelligent agents.
Why do you believe we are in a simulation? What is your thesis and how do you test it?
The guest explains it is all philosophy and theory, not testable with instruments. The argument is statistical: we have thousands/millions of simulations and only one real world, so you are more likely to be in a simulated world. He references video games and simulations for science, saying if every kid has hundreds of video games, statistically you're probably in one.
Unlock the full claims, asset map, scores, related transcripts, follow-up questions, and AI chat — shaped around your portfolio, watchlist, favorite speakers, and risks.
