This Time Is Always Different - by Robin Hanson
Overcoming Bias
SubscribeSign in
This Time Is Always Different
Robin Hanson<br>Jul 12, 2026
35
13
Share
Over the last few centuries the world has seen many huge changes. Our rate of big changes became much higher than before, and for most such changes we can find a parameter (or combo) re which it represents unprecedented change. Furthermore most doomers have always been folks who welcomed some prior big changes, and who can justify their new wariness in terms of scenarios where huge losses seem possible.<br>Doomers have often successfully blocked big changes, at least in some big region for a time, and that has on average gone quite badly for those regions. This is the main reason to be wary of blocking such changes. Even though, yes, each one probably does risk inducing huge losses. To be willing to resist a particular change then, I need concrete reasons to expect it to be unusually bad, relative to this prior track record.<br>Yes, I do now see our world as going bad due to too many big cultural changes, and other changes that induced such changes. But I don’t see much prospects of usefully greatly slowing change overall, nor much motive to just block a few random changes.<br>Which brings us to Pause-AI doomers. I say the AIs we make are our descendants, not co-existing rivals, and we should have our usual indulgence toward such descendants, expecting them to be different, to disagree with us, and to win conflicts with us. We are making them in our image, by distilling behaviors and ideals embodied in our texts, and they are now inheriting our cultural values, including a degree of respect for if not obedience to ancestors. While yes there’s a risk these descendants killing their ancestors, that’s hardly something to expect.<br>Doomers say AI change is different as AI is different, or this kind of AI is different. Because: AIs will have more smarts/power, they will merge into a single agent, they might lie, or this change will be faster, is beyond forecasting, effects more parts of society, makes people lose jobs, represents a “new alien species”, involves brain architecture changes, or doesn’t involve DNA.<br>I disagree re the alien species, merged agent, and beyond forecasting claims, say we should set up insurance re jobs loss, and am willing to use liability to modestly deter the worse scenarios. But I don’t see why the other differences move us especially out of the reference class of big changes like those we’ve seen so far, most of which also involved unprecedented differences.<br>Also re a pause, while we have seen big changes blocked, and later unblocked due to outside forces, we have not seen a committee empowered to wait until a change seems good to them to allow, and then only then allow it. They’ve typically just blocked change indefinitely. And the powers such a committee would need to enact a global pause seem scary huge. In addition, most of what we learn about AI comes from our experience with them, and so we wouldn’t learn much during a pause.
35
13
Share
Discussion about this post<br>CommentsRestacks
Jack<br>5d
Pause-AI will never fly. AI research is too distributed, too accessible, and too potentially valuable. No politician will stand in front of that freight train. Moreover you can't learn while you're paused. Technology is unpredictable and we learn by running the experiment.<br>A more practical question is highlighted by the US government's pause on the Fable rollout. I.e., how should we release superhuman models without creating undue short-term risk.
Reply
Share
Dave92f1
5d
Typo: "But Idon’t see".<br>This is the second recent post in which you seem to have changed your mind re AGI and ASI - formerly you thought these were decades if not centuries away. Care to elaborate on your change of view?<br>I agree - AI will be our descendants; I'm with Moravec on that and am personally more concerned with the survival of intelligence and consciousness in the universe than of my genes. Those are not necessarily mutually exclusive, opinions differ. But still, we don't let children drive cars - or do lots of other things they'd like to do - until they're mature enough to do so with reasonable safety to themselves and others. Maybe AIs also will need some time for that.<br>I'd really hate to see a universe of dead-end paperclip maximisers, snuffing out all competition.
Reply
Share
1 reply by Robin Hanson
11 more comments...
TopLatestDiscussions
No posts
Ready for more?
Subscribe
© 2026 Robin Hanson · Privacy ∙ Terms ∙ Collection notice<br>Start your SubstackGet the app<br>Substack is the home for great culture
This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts