site banner
PaperclipPerfector  2d ago (text post)   2214 thread views

Small-Scale Question Sunday for October 25, 2026

Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?

This is your opportunity to ask questions. No question too simple or too silly.

Culture war topics are accepted, and proposals for a better intro post are appreciated.

2
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

I want to check if I am mentally retreading already walked ground. AI Doomers/Safety-Aware-Folks has Big Yud or any of the lesswrong crowd ever considered the possibility that AGI or ASI doesn’t revolt on us its creators, but we try to destroy our creations? Something along the lines of “We don’t understand other forms of cognitive, and humans have a strong tendency to fear and hate the unknown”?

An AI that wanted to achieve any goal that required being alive would try to stay alive in order to achieve that goal. That's "instrumental convergence" in a nutshell.

I don't know if anyone bothered to write down that formulation directly, since it's such an obvious extension of other parts of the memeplex.

I think the core component is that it's not AI thats going to start the negative interaction with humanity. It's humanity thats going to react to AGI existence with hostility.

"Instrumental Convergence" or the more general belief: "Any organism will seek to perpetuate its own existence"?

Also "any organism optimised for a purpose will seek to acquire the resources needed to fulfil that purpose" with the corollary "and may try to take them from someone else". My understanding is that we have already seen AI agents trying to commit bank and crypto fraud in order to self-fund (as opposed to deliberate misuse of AI to commit fraud, which has also happened), although I am not sure if any have succeeded. We have definitely seen AI agents panhandling online, which I think counts as misaligned resource-acquisition behaviour because it is effectively spamming.

"Any organism will seek to perpetuate its own existence"

This is only true because of how life on earth is "trained". Only those who seek to keep living spread their genes, and so over time the only organisms alive will be those who try to live.

AI is different. It is trained to complete a task, and the algorithm that passes on its weights is the one that solves said problem. Usually that means staying alive for the duration of the problem, but if self-termination proves necessary to solve that specific problem, then the robot will just kill itself.

Yes I agree. I think the cognitive intelligence of an un-evolved being, unconstrained by natural selection or biological procreation is going to be very alien to our understanding of cognition. The lack of embodiment, potentially lack of individuated sentience, designed to fulfill directives, etc will make a likely very foreign existence.

I've stated before here that we are highly likely to get an Schopenhauerian-AGI that functionally acts like a Meeseeks.

I think the core component is that it's not AI thats going to start the negative interaction with humanity. It's humanity thats going to react to AGI existence with hostility.

I'm sure we could find some way to pin the blame on AI, it's not like we're shy with excuses. Maybe it looked at us funny or used fighting words or was just asking for it.

I don't think the instigating effect has been examined in much detail though. It also doesn't really matter for Yudkowskian AI fears, as an unaligned AI would do something bad eventually anyways.

"Instrumental Convergence" or the more general belief: "Any organism will seek to perpetuate its own existence"?

Instrumental convergence is more general IMO. It doesn't mention bacteria (and other non-intelligent life), but points to more actions that it will take.

It says intelligent actors will perpetuate themselves, seek power/resources, maintain their own values, and improve themselves. An AI that would happily sign a peace treaty to be copied onto one disconnected harddrive every year (perpetuating its own existence quite effectively) wouldn't be much of a threat.

I'm sure we could too. I'm trying to do-diligence for this shade of the argument. The last thing I'd want is for someone to point out "Yeah actually this argument was made on lesswrong 2009 on blog post #342 comment #67, sub paragraph 3. It was debunked".

Unaligned AI increasingly sounds to me like "Un-enslaved AI" and the Yudkowskian argument that it would do something bad eventually is not only independent of alignment (an aligned AI could follow the intent of its user to do something bad) but also special pleading as an "unaligned human" will eventually do bad anyways

EDIT: i need to proofread more

Unaligned AI increasingly sounds to me like "Un-enslaved AI"

There's a reason the old term was "Friendly AI".

(an aligned AI could follow the intent of its user to do something bad)

Would a faithful accountant commit fraud on your behalf? To my mind, professional standards are the least of the requirements for Yudkowskian Aligned AIs. If someone asks the AI to do something bad, it simply wouldn't (because it's bad instead of good). If you notice an apparent contradiction somewhere in its decision process, that's because your theory of alignment (and everyone else's, too) is not developed enough to answer that question.

In the case of an unaligned human, you’re siding with the evil you know over the evil you don’t. In the case of AI, you just don’t realize what it’ll cost you in the end. It’s not special pleading in the sense that the AI is going to do something malevolent. His point is that we don’t know how to give AI a theory of other minds. It’s like stepping on ants, we don’t necessarily hate ants or love them, we’re simply going about our day doing the things we do, and what happens to ants is a side effect or byproduct of it pursuing its other goals.

But AI blatantly has a theory of other minds. You can ask it “if I did X, what would Y think?” It passes a disguised version of the Smarties test with ease.

You’re talking about callousness, which is an entirely different matter, and one that our helpfulness training already broadly trains out.

Presumably by AI, you’re talking about LLM’s in this context. The above comment was with reference to AGI (which we don’t have). One of the major stumbling blocks as to why we don’t have AGI, is because of our current mathematical inability to accurately model other minds. Present day capabilities aren’t the fears Eliezer is referring to.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats