Nick Bostrom

1. 1998Co-founds the transhumanist movement

   With David Pearce he co-founds the World Transhumanist Association. He later left it, and says he is neither an uncritical transhumanist nor anti-technology.

   Wikipédia 2025 · Nick Bostrom2
2. 2003The simulation argument

   His paper 'Are You Living in a Computer Simulation?' argues that one of three propositions must be true. He says we currently lack grounds to rule out any of them.

   Wikipédia 2025 · Nick Bostrom2
3. 2005Founds the Future of Humanity Institute

   He creates the Oxford institute that studies existential risks and long-term futures. It closed in 2024.

   Wikipédia 2025 · Nick Bostrom2
4. 2013The hypothesis versus the argument

   He stresses the common mistake of mixing the simulation hypothesis with the simulation argument, and says the hypothesis itself is less than 50% likely. Years later he still declines to put a number on it.

   In their words

   « I think that is less than 50% likely to be the case »In 2013 he says the simulation hypothesis itself is less than 50% likely, while he believes the simulation argument (trilemma). watch

   « the most common misunderstanding is to conflate the simulation hypothesis with the simulation argument »He distinguishes the simulation hypothesis (we are simulated) from the simulation argument (one of three propositions is true). watch

   « all the really big existential risks are ones that arise in one way or another out of human activity »He argues the biggest existential risks come from our own activity, notably synthetic biology, nanotechnology and machine intelligence. watch

   Video 2013 · Nick Bostrom - The Simulation…2
5. 2014Superintelligence

   His book argues that a self-improving AI could trigger an intelligence explosion, that intelligence and goals are independent, and that AI must be aligned with human values.

   Wikipédia 2025 · Nick Bostrom
6. 2019The vulnerable world hypothesis

   He suggests some technologies could destroy civilization by default once discovered.

   Wikipédia 2025 · Nick Bostrom
7. 2021Digital minds

   With Carl Shulman he studies how to share the world with digital minds that could have superhumanly strong claims to resources, and argues moral reflection is needed now.

   Article 2021 · Sharing the World with Digital…
8. 2024Deep Utopia: meaning in a solved world

   He asks how to live well once technology has solved practical problems. Work and effort might lose their meaning, so purpose may have to come from new kinds of goals.

   Wikipédia 2025 · Nick Bostrom2
9. 2024Moral status without consciousness

   Around Deep Utopia he says suffering is enough for moral status but consciousness may not be necessary: a mind with goals, a sense of self over time and reciprocal relationships could also deserve it.

   In their words

   « my guess is the answer to both is yes that you could have morally significant beings on digital substrate and also that you could in principle become a digital mind »He thinks digital minds can be morally significant and that a person could in principle become a digital mind, e.g. by gradual replacement. watch

   « I think is a lean towards thinking um conscious is not even necessary for moral status »He leans towards thinking consciousness is not even necessary for moral status, though suffering would suffice. watch

   « people have misunderstood what I was like thinking that I was kind of gung-ho about establishing some total World surveillance system »He says readers misread his vulnerable world paper as advocating total surveillance; he presents the idea from several sides. watch

   Video 2024 · Nick Bostrom on…2
10. 2025The cosmic host

    He presents 'AI Creation and the Cosmic Host': a superintelligence we create may enter a world with other super beings (aliens, simulators, God), so it should be able to get along with them.

    In their words

    « the fate of you know the gorillas depend a lot more on our choices than on what the gorillas choose to do »He compares us to gorillas to explain why a radically smarter AI could decide humanity's fate. watch

    « it does look like we are sort of within striking distance it seems to me with the continued scale-up »He thinks superintelligence is within striking distance, while insisting on probability distributions. watch

    « I haven't really expressed a particular P(doom) because I think it might also turn out to depend quite a lot on your definition of doom »He declines to give a P(doom): the weird middle outcomes may be likelier than clear doom or clear utopia. watch

    Video 2025 · AI FUTURE THAT CAN DESTROY US |…
11. 2026A fretful optimist, against overshooting

    He calls himself a fretful optimist and a moderate fatalist: alignment may be easy or impossible, and effort matters in between. He favors building superintelligence carefully and fears calls for an AI stop could overshoot.

    In their words

    « you could say um that I'm a moderate fatalist »He says he is a moderate fatalist about AI: the outcome depends largely on how intrinsically hard alignment is, though our effort could matter if it is intermediate. watch

    « there's a big cost to delay which is maybe easier to perceive because it's less vivid than some particular catastrophic risk »He warns that a long AI pause has costs, as people die and suffer every day, so we should not delay longer than necessary. watch

    « I think sentience is a sufficient condition for having moral status »Sentience suffices for moral status, but probably is not necessary: a sense of self over time and life goals could also suffice. watch

    Video 2026 · Prof. Nick Bostrom: The truth…23

The arguments 2

The site's arguments they defend or criticize, and why. Click a title to see the full argument.

Signature ideas