nem

nem2mo

Another one I've seen is the use of 'Not A, But B' statements.
"This is not just an existential crisis. It's a full-blown catastrophe."

OP's writing also contained something like that: "This is not science fiction—it’s unfolding right now, within our lifetimes." It's a shame, because it is not a bad sentence structure. But, like with emdashes, you now have to monitor your writing for overuse of logic like this.

Replying toHow I stopped being sure LLMs are just making up their internal experience (but the topic is still confusing)

nem2mo

How I stopped being sure LLMs are just making up their internal experience (but the topic is still confusing)

As an experiment, can we intentionally destroy the state in between each token? What happens if we run many of these same introspection exercises again, only this time, with a brand new instance of an AI for each word. Would it still be able to come off as convincingly introspective in a situation where many of its possible mechanisms for introspection have been disabled? Would it still 'improve' at introspection when falsely told that it should be able to?

I am probably not the best person to run this experiment, but it feels relatively important. Maybe I'll get some time to do it over the holidays, but anyone with the time and ability should feel free.

-2

Started to read this, but I stopped once I realized it was written with AI.

Replying toThe Only Option Left

nem5mo

The Only Option Left

I appreciate you writing this up. I have been a bit starved for this community's opinions on the current political situation. I felt very strongly that electing Trump was a huge mistake on almost every level. But I think there is something we have overlooked which has become more obvious to me over the last 9 months.

Advocacy has been extremely diluted. I want to protest against the acceleration of AI. I want to invest my time and emotional energy toward what I think is the largest existential risk to humanity. I want to call my representatives monthly to let them know that I am a single issue voter on AI safety.

What I... (read more)

-6

Replying toNot stepping on bugs

nem5mo

Not stepping on bugs

Eh, I think it's possible that worms do have feelings. But even if they don't, my mind is able to trick me into having empathy for them in a way which it doesn't for video game characters. Are worms like people? shrugs I'm not arguing whether they are or aren't. All I can say for sure is that something about them triggers my protective and benevolent instincts, and I have chosen to lean into that instead of rationalizing it away.

As for 'type of place', I guess what I mean is that I want the world to have more people with bountiful empathy. And I can't very well be optimistic about achieving that... (read more)

Replying toMy AI Predictions for 2027

nem5mo

My AI Predictions for 2027

I'll register my prediction here as well. I largely agree with your projection, although my median case looks a little bit more advanced. Also, note that I am not vouching for your arguments.

75% - likely we live in a world that feels pretty normal. That is, similar to what you described or a bit more advanced, as mentioned.

Here are some places I differ from your predictions which might give insight into what I mean by "a little bit more advanced":
- In general, I anticipate more progress, both in terms of tech, and its integration into our world.
- AI might have transformed some major industries and careers, even without providing novel research or... (read 405 more words →)

Replying toNot stepping on bugs

nem5mo

Not stepping on bugs

There is no causal link. It's about 1) practicing empathy, and 2) making the world the type of place you'd like it to be.

Replying toNot stepping on bugs

nem6mo

Not stepping on bugs

When I see a worm drowning in a puddle, or stuck on hot pavement, I rescue it. Is it important? Probably not. But I'd like to think that if anyone ever saw me drowning in a puddle, they would rescue me.

Thank you for writing this post.

Replying toThe Problem

nem6mo

The Problem

As a native English speaker, that seems pretty unnatural to me. But your choice of course!

Replying toThe Problem

nem6mo

The Problem

Calling out a small typo, as this is clearly meant as a persuasive reference point:

"On* our view, the international community’s top immediate priority should be creating an “off switch” for frontier AI development"

Presumably, "On" here should be "In"

-1

Seeking advice on careers in AI Safety

nem

10mo

Title sums it up. I am a mid-level software engineer (no CS degree*, but 4 years of full-stack dev work at mid-size local company).

I was recently laid off, and have been looking for new jobs. However, I am seriously considering taking this time to retool myself to work on AI safety. I just don't know exactly what that means.

Does that mean going back to school so that I can do technical safety work in a few years? Does it mean trying to get a job in AI safety right away? Ditching software altogether and using my background to become an advocate?

If anyone has a strong opinion on how best my background can... (read more)

Just How Good Are Modern Chess Computers?

nem

When you play chess against a grandmaster, you know you're going to lose. But do you really understand how badly? You might think to yourself, “There's only like 20 possible moves per turn. Even a novice like me can quickly rule out half of those.”

In this post, I'd like to help build an intuition for just HOW good the best chess players are, while picking from “just” the best 10 moves.
I’ll also add some intuition for how much better chess computers are than humans. As most people don’t play chess, I’ll develop this intuition using something most people are inherently familiar with: Physical ability.

Specifically, in my metaphor, chess is a 1-on-1 soccer... (read 1776 more words →)

The Anthropic Principle Tells Us That AGI Will Not Be Conscious

nem

More specifically, the Anthropic Principle tells us that AGI/TAI are unlikely to be conscious in a world where 1) TAI is achieved, 2) Alignment fails, and 3) the 'prominence' of consciousness scales with either increasing levels of capability or with a greater number of conscious being/the time of their existence.

The argument is simple. If the future is filled with artificial intelligence of human origin, and if that AI is conscious, then any given observer should expect to be one of those AIs. This means that, on balance, one of the following is likely true:

1) The anthropic principle does not hold.

You and I, as observers, are simply an incredibly unlikely, but also inevitable... (read 227 more words →)

What does pulling the fire alarm look like?

nem

I am not quite ready to pull my own personal fire alarm. The word 'quite' is doing a lot of lifting in that sentence. Because a fire is starting to feel like a real possibility in my future, I've begun to think about what pulling the fire alarm would actually mean. What would it mean for me? What would it mean for the community if a significant proportion agreed?

This is an open question for folks who have a good understanding of what actions the AI safety community would do if we agreed to pull the alarm. It's also for people who have a good idea of what they personally would do.

For myself,... (read more)

Is it a bad idea to pay for GPT-4?

nem

I don't want to accelerate an arms race, and paying for access to GPT seems like a perfect way to be a raindrop in a dangerous flood. My current idea is to pay an equal amount monthly to Miri. I'll view it as the price being $40 per month with half going to AI safety research.

Is this indefensible? Let me know. GPT-4 is very useful to me personally and professionally, and familiarity of language models will also be useful if I have enough time to transition into an AI safety career, which I am strongly considering.

If it is a good idea, should we promote the offsetting strategy among people who are similarly conflicted?

The Pinnacle

nem

In another universe, in another time, the stars don’t shine like ours. In fact, they act in a manner quite more familiar to you and I as that of a black hole. Take Los Ral as an example. An average star, whose future in this universe is rare, and storied.
Like its peers, Los Ral does not emit light. Instead, it draws energy inexorably toward it. Planets sail the solar plane, much as they do around our sun; But if there is any scrap of energy on these planets, it is stripped from the surface. As though beckoned by Los Ral, photons fly the true and straight line towards the star. It is... (read 2211 more words →)

I Believe we are in a Hardware Overhang

nem

Epistemic status. I am just a regular person who follows the space, and this is just my hunch based on a few days of musing on long walks. You should not update on this. I just wanted to put my thoughts out there, and if it generates discussion, all the better.

If I were in charge, my hand would be on the fire alarm right now.

I can already envision ways we could use current, public facing technology to create AGI. I would be surprised if no one did it in the next 5-8 years, even with no advancement in the constituent parts. I would almost hesitate to propose my solution for fear of... (read 282 more words →)

LESSWRONG
LW

LESSWRONG
LW

Is it a bad idea to pay for GPT-4?

The Pinnacle

Just How Good Are Modern Chess Computers?

I Believe we are in a Hardware Overhang

nem

nem

Seeking advice on careers in AI Safety

Just How Good Are Modern Chess Computers?

The Anthropic Principle Tells Us That AGI Will Not Be Conscious

What does pulling the fire alarm look like?

Is it a bad idea to pay for GPT-4?

The Pinnacle

I Believe we are in a Hardware Overhang

nem

Is it a bad idea to pay for GPT-4?

The Pinnacle

Just How Good Are Modern Chess Computers?

I Believe we are in a Hardware Overhang

nem

nem

Seeking advice on careers in AI Safety

Just How Good Are Modern Chess Computers?

The Anthropic Principle Tells Us That AGI Will Not Be Conscious

What does pulling the fire alarm look like?

Is it a bad idea to pay for GPT-4?

The Pinnacle

I Believe we are in a Hardware Overhang