This isn’t that complicated. The halo effect is real and can go to extremes when romantic relationships are involved, and most people take their sense data at face value most of the time. The sentence is meant completely literally.
Odd anon
Karma: 298
GPT-5 training is probably starting around now
Sam Altman confirmed (paywalled, sorry) in November that GPT-5 was already under development. (Interestingly, the confirmation was almost exactly six months after Altman told a senate hearing (under oath) that “We are not currently training what will be GPT-5; we don’t have plans to do it in the next 6 months.”)
The United States is an outlier in divorce statistics. In most places, the rate is nowhere near that high.
It is not that uncommon for people to experience severe dementia and become extremely needy and rapidly lose many (or all) of the traits that people liked about them. Usually, people don’t stop being loved just because they spend their days hurling obscenities at people, failing to preserve their own hygiene, and expressing zero affection.
I would guess that most parents do actually love their children unconditionally, and probably the majority of spouses unconditionally love their partners.
(Persistent identity is a central factor in how people relate to each other, so one can’t really say that “it is only conditions that separate me from the worms.”)
Brainware.
Brains seem like the closest metaphor one could have for these. Lizards, insects, goldfish, and humans all have brains. We don’t know how they work. They can be intelligent, but are not necessarily so. They have opaque convoluted processes inside which are not random, but often have unexpected results. They are not built, they are grown.
They’re often quite effective at accomplishing something that would be difficult to do any other way. Their structure is based around neurons of some sort. Input, mystery processes, output. They’re “mushy” and don’t have clear lines, so much of their insides blur together.
AI companies are growing brainware in larger and larger scales, raising more powerful brainware. Want to understand why the chatbot did something? Try some new techniques for probing its brainware.
This term might make the topic feel more mysterious/magical to some than it otherwise would, which is usually something to avoid when developing terminology, but in this case, people have been treating something mysterious as not mysterious.
(The precise text, from “The Andalite Chronicles”, book 3: “I have made right everything that can be made right, I have learned everything that can be learned, I have sworn not to repeat my error, and now I claim forgiveness.”)
Larry Page (according to Elon Musk), want AGI to take the world from humanity
(IIRC, Tegmark, who was present for the relevant event, has confirmed that Page had stated his position as described.)
Ehhh, I get the impression that Schidhuber doesn’t think of human extinction as specifically “part of the plan”, but he also doesn’t appear to consider human survival to be something particularly important relative to his priority of creating ASI. He wants “to build something smarter than myself, which will build something even smarter, et cetera, et cetera, and eventually colonize and transform the universe”, and thinks that “Generally speaking, our best protection will be their lack of interest in us, because most species’ biggest enemy is their own kind. They will pay about as much attention to us as we do to ants.”
I agree that he’s not overtly “pro-extinction” in the way Rich Sutton is, but he does seem fairly dismissive of humanity’s long-term future in general, while also pushing for the creation of an uncaring non-human thing to take over the universe, so...
Hendrycks goes into some detail on the issue of AI being affected by natural selection in this paper.
Please link directly to the paper, rather than requiring readers to click their way through the substack post. Ideally, the link target would be on a more convenient site than academia.edu, which claims to require registration to read the content. (The content is available lower down, but the blocked “Download” buttons are confusing and misleading.)
When this person goes to post the answer to the alignment problem to LessWrong, they will have low enough accumulated karma that the post will be poorly received.
Does the author having lower karma actually cause posts to be received more poorly? The author’s karma isn’t visible anywhere on the post, or even in the hover-tooltip by the author’s name. (One has to click through to the profile to find out.) Even if readers did know the author’s karma, would that really cause people to not just judge it by its content? I would be surprised.
I found some of your posts to be really difficult to read. I still don’t really know what some of them are even talking about, and on originally reading them I was not sure whether there was anything even making sense there.
Sorry if this isn’t all that helpful. :/
Wild guess: It realised its mistake partway through, and followed through it anyway as sensibly as could be done, balancing between giving a wrong calculation (“+ 12 = 41“), ignoring the central focus of the question (” + 12 = 42”), and breaking from the “list of even integers” that it was supposed to be going through. I suspect it would not make this error when using chain-of-thought.
Such a word being developed would lead to inter-group conflict, polarisation, lots of frustration, and general bad things to society, regardless of which side you may be on. Also, it would move the argument in the wrong direction.
If you’re pro-AI-rights, you could recognize that bringing up “discrimination” (as in, treating AI at all differently from people) is very counterproductive. If you’re on this side, you probably believe that society will gradually understand that AIs deserve rights, and that there will be a path towards that. The path would likely start with laws prohibiting deliberately torturing AIs for its own sake, then something closer to animal rights (some minimal protections against putting AI through very bad experiences even when it would be useful, and perhaps against using AIs for sexual purposes since it can’t consent), then some basic restrictions on arbitrarily creating, deleting, and mindwiping AIs, and then against slavery, etc etc. Bringing up “discrimination” early would be pushing an end-game conflict point early, convincing some that they’re moving onto a slippery slope if they allow any movement down the path, even if they agree with the early steps on their own. The noise of argument would slow down the progress.
If you’re anti-AI-rights (being sure of AI non-sentience, or otherwise), then such a word is just a thing to make people feel bad, without any positives. People on this side would likely conclude that disagreement on “AI rights” is probably temporary, until either people understand the situation better or the situation changes. Suddenly “raising the stakes” on the argument would be harmful, bringing in more noise which would make it harder to hear the “signal” underneath, thus pushing the argument in the wrong direction. The word would make it take longer for the useless dispute to die down.
Something to consider: Most people already agree that AI risk is real and serious. If you’re discussing it in areas where it’s a fringe view, you’re dealing with very unusual people, and might need to put together very different types of arguments, depending on the group. That said...
stop.ai’s one-paragraph summary is
OpenAI, DeepMind, Anthropic, and others are spending billions of dollars to build godlike AI. Their executives say they might succeed in the next few years. They don’t know how they will control their creation, and they admit humanity might go extinct. This needs to stop.
The rest of the website has a lot of well-written stuff.
Some might be receptive to things like Yudkowsky’s TED talk:
Nobody understands how modern AI systems do what they do. They are giant, inscrutable matrices of floating-point numbers that we nudge in the direction of better performance until they inexplicably start working. At some point, the companies rushing headlong to scale AI will cough out something that’s smarter than humanity. Nobody knows how to calculate when that will happen. My wild guess is that it will happen after zero to two more breakthroughs the size of transformers.
What happens if we build something smarter than us that we understand that poorly? Some people find it obvious that building something smarter than us that we don’t understand might go badly. Others come in with a very wide range of hopeful thoughts about how it might possibly go well. Even if I had 20 minutes for this talk and months to prepare it, I would not be able to refute all the ways people find to imagine that things might go well.
But I will say that there is no standard scientific consensus for how things will go well. There is no hope that has been widely persuasive and stood up to skeptical examination. There is nothing resembling a real engineering plan for us surviving that I could critique. This is not a good place in which to find ourselves.
And of course, you could appeal to authority by linking the CAIS letter, and maybe the Bletchley Declaration if statements from the international community will mean anything.
(None of those are strictly two-paragraph explanations, but I hope it helps anyway.)
Concerning. This isn’t the first time I’ve seen a group fall into the pitfall of “wow, this guy is amazing at accumulating power for us, this is going great—oh whoops, now he holds absolute control and might do bad things with it”.
Altman probably has good motivations, but even so, this is worrying. “One uses power by grasping it lightly. To grasp with too much force is to be taken over by power, thus becoming its victim” to quote the Bene Gesserit.
Time for some predictions. If this is actually from AI developing social manipulation superpowers, I would expect:
We never find out any real reasonable-sounding reason for Altman’s firing.
OpenAI does not revert to how it was before.
More instances of people near OpenAI’s safety people doing bizarre unexpected things that have stranger outcomes.
Possibly one of the following:
Some extreme “scissors statements” pop up which divide AI groups into groups that hate each other to an unreasonable degree.
An OpenAI person who directly interacted with some scary AI suddenly either commits suicide or becomes a vocal flat-earther or similar who is weirdly convincing to many people.
An OpenAI person skyrockets to political power, suddenly finding themselves in possession of narratives and phrases which convince millions to follow them.
(Again, I don’t think it’s that likely, but I do think it’s possible.)
It’s good that Metaculus is trying to tackle the answer-many/answer-accurately balance, but I don’t know if this solution is going to work. Couldn’t one just get endless baseline points by predicting the Metaculus average on every question?
Also, there’s no way to indicate “confidence” (like, outside-level confidence) in a prediction. If someone knows a lot about a particular topic, and spends a lot of time researching a particular question, but also occasionally predicts their best guess on random other questions outside their area of expertise, then the point-based “incentives” become messy. That’s something I like about Manifold that’s missing from Metaculus, and I wonder whether it might be possible to work in something like that while keeping Metaculus’s general system.
There’s… too many things here. Too many unexpected steps, somehow pointing at too specific an outcome. If there’s a plot, it is horrendously Machiavellian.
(Hinton’s quote, which keeps popping into my head: “These things will have learned from us by reading all the novels that ever were and everything Machiavelli ever wrote, that how to manipulate people, right? And if they’re much smarter than us, they’ll be very good at manipulating us. You won’t realise what’s going on. You’ll be like a two year old who’s being asked, do you want the peas or the cauliflower? And doesn’t realise you don’t have to have either. And you’ll be that easy to manipulate. And so even if they can’t directly pull levers, they can certainly get us to pull levers. It turns out if you can manipulate people, you can invade a building in Washington without ever going there yourself.”)
(And Altman: “i expect ai to be capable of superhuman persuasion well before it is superhuman at general intelligence, which may lead to some very strange outcomes”)
If an AI were to spike in capabilities specifically relating to manipulating individuals and groups of people, this is roughly how I would expect the outcome to look like. Maybe not even that goal-focused or agent-like, given that GPT-4 wasn’t particularly lucid. Such an outcome would likely have initially resulted from deliberate probing by safety testing people, asking it if it could say something to them which would, by words alone, result in dangerous outcomes for their surroundings.
I don’t think this is that likely. But I don’t think I can discount it as a real possibility anymore.
Relevant: My Taxonomy of AI-risk counterarguments, inspired by Zvi Mowshowitz’s The Crux List.