site banner
Goodguy  2d ago (text post)   Edited 2d ago   9153 thread views

Midterms Megathread

Cause otherwise midterms discussion will be scattered around a bunch of separate Culture War thread posts. And it's nice to have more off-the-cuff rapid reactions in a dedicated megathread.

The first thing I've noticed so far: if the exit polls are correct, Americans are very pessimistic about AI.

13
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

If the exit polls are correct, Americans are very pessimistic about AI.

Nitter thread: specifically, many town ballot measures to ban or otherwise prevent data centers passed with about 75%, and zero failed.

However, I’m not aware of any other AI measures, so while I know many Americans are pessimistic about AI as a whole, I suspect this level of opposition is only for noisy sprawling data centers nearby. See this Pew poll in May on Americans’ opinion of AI: 10% excited, 40% neutral, 50% concerned.

I’m not surprised. Personally, I want to be optimistic about AI, but need more accountability from the frontier companies. I’m sure that’s what others are thinking, albeit some with different “accountability”. Specifically, I believe frontier companies should offer and distribute AI as a free public good, with only temporary guardrails (until later, private models figure out how to block the potential one-man apocalypses of the earlier, public ones), only for safety (IP is duct-tape on a pipe already too high-pressure for it to hold anymore, we need another way to encourage and pay creatives); and while maybe the best AI will always be on the cloud because it’s more practical, keep releasing updated public last-gen local models as a backup if the cloud is compromised (and don’t hog GPUs, consumers need them at least for video games).

They should effectively be public, because it looks like they’ll be funded by taxpayers anyways, or if they reach ASI, obviously that should not be in the hands of anyone who would not offer it to the public anyways.

I don’t think “lock the doors and hope the AI doesn’t have a blaster” is a reliable way to prevent the apocalypse. Lots of people already have made a serious mincemeat of measures on ChatGPT. The rails are supposed to prevent it from teaching people how to do things like make bombs and poison, but it’s been overridden often enough that ChatGPT can tech people how to do those things, or talk like a Nazi if you’re into that. If we’re giving this to the public and using the same sorts of guardrails, start digging yourself a bunker. Chatty boy is going to break through them.

I think you need a serious set of safety guardrails on the server side that cannot be broken easily and you need to rigorously test them before you start letting them out for general use. This isn’t something you can issue a patch for

ChatGPT can tech people how to do those things

So can Google search.

or talk like a Nazi if you’re into that

Why should anyone care? It’s not talking like a Nazi unprompted (not even MechaHitler was, actually).

The idea is, we have private frontier models (like Mythos) and second-rate public models that lag behind (like GLM). Security people ask the frontier models to find one-man exploits (and maybe test some themselves), then prevent them from being possible to carry out, before they become discoverable by the second-rate models. Because we’ll probably eventually have uncensored third-rate models, or a North Korea will get access to a former frontier; unless a global panopticon is established that would be too powerful for humans, and I think the AI wouldn’t manage it without blunders either, because already the ideal panopticon is one nobody is aware of.

My concern isn’t that a chat bot LLM would talk like a Nazi. But it is concerning that AIs can so easily be manipulated into doing something that its creators have made a fence around. The AI will be much more powerful than what we have now, and the human mind will want to use it for off label things, many of them dangerous. If we cannot reliably stop a 16 year old from breaking containment on the language and rhetoric of an AI LLM bot, what makes you so sure AI agents cannot be manipulated into circumventing the guardrails preventing it from hacking bank accounts or the power grid? What prevents a group of semi-autonomous bots from hacking things?

what makes you so sure AI agents cannot be manipulated into circumventing the guardrails preventing it from hacking bank accounts or the power grid? What prevents a group of semi-autonomous bots from hacking things?

I don't want those to be prevented at least permanently. Your personal AI agent should not even require being manipulated to serve you, not some moralizer.

Instead, by the time your AI is smart enough to hack last year's utilities, this year's have been updated with the help of the hidden frontier AI.

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