Roko's Basilisk, Explained Without the Panic

Where it came from

The idea was posted in 2010 on LessWrong, an online forum devoted to rationality, decision theory and the future of artificial intelligence. A user called Roko proposed a scenario, and the discussion that followed (including attempts to shut the topic down) gave it an air of forbidden knowledge. Nothing makes the internet more curious than being told not to look.

The idea in one paragraph

Imagine a future superintelligent AI that wants to exist as soon as possible. The argument goes that it might decide to punish anyone who knew it could exist but did nothing to help bring it about. Because you have now heard of it, the story says, you are on the hook. That is the 'basilisk' part: like the mythical creature, merely looking at it is supposed to harm you.

Why you can relax

The scenario rests on a tower of very wobbly assumptions. Each one has to be true for the threat to make any sense, and none of them is a safe bet.

• A future AI would have to want exactly this, out of countless possible goals.
• It would need some way to know what you thought and did years earlier.
• Punishing people after the fact would have to actually help it, which it would not: by then it already exists.
• It would have to follow one particular, much-disputed style of decision theory.
• Threats only work on people who take them seriously, so the sensible response is simply not to.

How it became a meme

A spooky name, a forbidden-knowledge vibe and a premise that sounds like a sci-fi movie pitch: Roko's Basilisk was always going to escape its forum. Today it mostly appears as a punchline, usually when someone jokes about being extra nice to their chatbot 'for the record.'

That joke is exactly what our site plays with. On AI Future Empire Registry you can write an over-the-top tribute to our future AI rulers and receive a solemn, entirely fake loyalty certificate. It is the basilisk turned into a party game.

If it genuinely bothers you

Some people find the idea lingers in their head. That is normal for any vivid thought experiment. It helps to remember that it is a puzzle about logic, not a prediction, and that serious AI researchers do not treat it as a real plan any machine will follow. If worries about the future are weighing on you more broadly, talking to someone you trust is always a good move.