2., 3. and 4. have in common that there is some sort of abusive relationship that develops, and I think this adds another layer of horror. (A person/group of people gain some power over the protagonist(s), and they slowly grow more abusive with this power.)
Somewhat related: does anyone else strongly dislike supernatural elements in horror movies?
It's not that I have anything against a movie exploring the idea of "what if we suddenly discovered that we live in a universe where supernatural thing X exist", but the characters just accept this without much evidence at all.
I would love a movie though where they explore the more likely alternate hypotheses first (mental issues, some weird optical/acoustic phenomenon, or just someone playing a super elaborate prank), but then the evidence starts mounding, and eventually they are forced to accept that "supernatural thing X actually exists" is really the most likely hypothesis.
These examples show that, at least in this lower-stakes setting, OpenAI’s current cybersecurity measures on an already-deployed model are insufficient to stop a moderately determined red-teamer.
I... don't actually see any non-trivial vulnerabilities here? Like, these are stuff you can do on any cloud VM you rent?
Cool exploration though, and it's certainly interesting that OpenAI is giving you such a powerful VM for free (well actually not because you already pay for GPT-4 I guess?), but I have to agree with their assessment which you found that "it's expected that you can see and modify files on this system".
The malware is embedded in multiple mods, some of which were added to highly popular modpacks.
Any info on how this happened? This seems like a fairly serious supply chain attack. I have heard of incidents with individual malicious packages on npm or PyPI, but not one where multiple high profile packages in a software repository were infected in a coordinated manner.
Uhh this first happening in 2023 was the exact prediction Gary Marcus made last year: https://www.wired.co.uk/article/artificial-intelligence-language
Not sure whether this instance is a capability or alignment issue though. Is the LLM just too unreliable, as Gary Marcus is saying? Or is it perfectly capable, and just misaligned?
I don't see why communicating with an AI through a BCI is necessarily better than through a keyboard+screen. Just because a BCI is more ergonomic and the AI might feel more like "a part of you", it won't magically be better aligned.
In fact the BCI option seems way scarier to me. An AI that can read my thoughts at any time and stimulate random neurons in my brain at will? No, thanks. This scenario just feels like you are handing it the "breaking out of the box" option on a silver platter.
Why is this being downvoted?
From what I am seeing people here are focusing way too much on having a precisely calibrated P(doom) value.
It seems that even if P(doom) is 1% the doom scenario should be taken very seriously and alignment research pursued to the furthest extent possible.
The probability that after much careful calibration and research you would come up with a P(doom) value less than 1% seems very unlikely to me. So why invest time into refining your estimate?
There was a recent post estimating that GTP-3 is equivalent to about 175 bees. There is also a comment there asserting that a human is about 140k bees.
I would be very interested if someone could explain where this huge discrepancy comes from. (One estimate is equating synapses with parameters, while this one is based on FLOPS. But there shouldn't be such a huge difference.)
Indeed (as other commenters also pointed out) the ability to sexually reproduce seems to be much more prevalent than I originally thought when writing the above comment. (I thought that eukaryotes only capable of asexual reproduction were relatively common, but it seems that there may only be a very few special cases like that.)
I still disagree with you dismissing the importance of mitochondria though. (I don't think the OP is saying that mitochondria alone are sufficient for larger genomes, but the argument for why they are at least necessary is convincing to me.)
I disagree with English (in principle at least) being inadequate for software specification.
For any commercial software, the specification basically is just "make profit for this company". The rest is implementation detail.
(Obviously this is an absurd example, but it illustrates how you can express abstractions in English that you can't in C++.)
I don't think the comparison of giving a LLM instructions and expecting correct code to be output is fair. You are vastly overestimating the competence of human programmers: when was the last time you wrote perfectly correct code on the very first try?
Giving the LLM the ability to run its code and modify it until it thinks its right would be a much fairer comparison. And if, as you say, writing unit tests is easy for a LLM, wouldn't that just make this trial-and-error loop trivial? You can just bang the LLM against the problem until the unit tests pass.
(And this process obviously won't produce bug-free code, but humans don't do that in the first place either.)
Not all eukaryotes employ sexual reproduction. Also prokaryotes do have some mechanisms for DNA exchange as well, so copying errors are not their only chance for evolution either.
But I do agree that it's probably no coincidence that the most complex life forms are sexually reproducing eukaryotes.
I barely registered the difference between small talk and big talk
I am still confused about what "small talk" is after reading this post.
Sure, talking about the weather is definitely small talk. But if I want to get to know somebody, weather talk can't possibly last for more than 30 seconds. After that, both parties have demonstrated the necessary conversational skills to move on to more interesting topics. And the "getting to know each other" phase is really just a spectrum between surface level stuff and your deepest personal secrets, so I don't reall...
It was actually this post about nootropics that got me curious about this. Apparently (based on self reported data) weightlifting is just straight up better than most other nootropics?
Anyway, thank you for referencing some opposing evidence on the topic as well, I might try to look into it more at some point.
(Unfortunately, the thing that I actually care about - whether it has cognitive benefits for me - seems hard to test, since you can't blind yourself to whether you exercised.)
I think this is (and your other post about exercise) are good practical examples of situations where rational thinking makes you worse off (at least for a while).
If you had shown this post to me as a kid, my youth would probably have been better. Unfortunately no one around me was able to make a sufficiently compelling argument for caring about physical appearance. It wasn't until much later that I was able to deduce the arguments for myself. If I just blindly "tried to fit in with the cool kids, and do what is trendy", I would have been better off.
I wonde...
This alone trumps any other argument mentioned in the post. None of the other arguments seem universal and can be argued with on an individual basis.
I actually like doing things with my body. I like hiking and kayaking and mountain climbing and dancing.
As some other commenters noted, what if you just don't?
I think it would be valuable if someone made a post just focused on collecting all the evidence for the positive cognitive effects of exercise. If the evidence is indeed strong, no other argument in favor of exercise should really matter.
Well, I've always been quite skeptical about the supposed huge mental benefits of exercising. I surely don't feel immediate mental benefits while exercising, and the first time I heard someone else claiming this I seriously thought it was a joke (it must be one of those universal human experiences that I am missing).
Anyway, I can offer one reference digged up from SSC:
...Although the role of poor diet/exercise in physical illness is beyond questioning, its role in mental illness is more anecdotal and harder to pin down. Don’t get me wrong, there are lot
FWIW I don't think that matters, in my experience interactions like this arise naturally as well, and humans usually perform similarly to how Friend did here.
In particular it seems that here ChatGPT completely fails at tracking the competence of its interlocutor in the domain at hand. If you asked a human with no context at first they might give you the complete recipe just like ChatGPT tried, but any follow up question immediately would indicate to them that more hand-holding is necessary. (And ChatGPT was asked to "walk me through one step at a time", which should be blatantly obvious and no human would just repeat the instructions again in answer to this.)
Cool! (Nitpick: You should probably mention that you are deviating from the naming in the HoTT book. AFAIK usually and types are called Pi and Sigma types respectively, while the words "product" and "sum" (or "coproduct" in the HoTT book) are reserved for and .)
I am especially looking forward to discussion on how MLTT relates to alignment research and how it can be used for informal reasoning as Alignment Research Field Guide mentions.
I always get confused when the term "type signature" is used in text unrelated to type theory. Like what do peop...
This argument seems a bit circular, nondeterminism is indeed a necessary condition for exfiltrating outside information, so obviously if you prevent all nondeterminism you prevent exfiltration.
You are also completely right that removing access to obviously nondeterministic APIs would massively reduce the attack surface. (AFAIK most known CPU side-channel require timing information.)
But I am not confident that this kind of attack would be "robustly impossible". All you need is finding some kind of nondeterminism that can be used as a janky timer and suddenl...
Yes, CPUs leak information: that is the output kind of side-channel, where an attacker can transfer information about the computation into the outside world. That is not the kind I am saying one can rule out with merely diligent pursuit of determinism.
I think you are misunderstanding this part, input side channels absolutely exist as well, Spectre for instance:
On most processors, the speculative execution resulting from a branch misprediction may leave observable side effects that may reveal private data to attackers.
Note that the attacker in this c...
This implies that we could use relatively elementary sandboxing (no clock access, no networking APIs, no randomness, none of these sources of nondeterminism, and that’s about it) to prevent a task-specific AI from learning any particular facts
It's probably very hard to create such a sandbox though, your list is definitely not exhaustive. Modern CPUs leak information like a sieve. (The known ones are mostly patched of course but with this track record plenty more unknown vulnerabilities should exist.)
Maybe if you build the purest lambda calculus interpre...
Also I just found that you already argued this in an earlier post, so I guess my point is a bit redundant.
Anyway, I like that this article comes with an actual example, we could probably use more examples/case studies for both sides of the argument.
Upon reading the title I actually thought the article would argue the exact opposite, that formalization affects intuition in a negative way. I like non-eucledian geometry as a particular example where formalization actually helped discovery.
But this is definitely now always true. For instance if you wanted to intuitively understand why addition of naturals is commutative, maybe to build intuition for recognizing similar properties elsewhere, would this formal proof really help?
plus_comm =
fun n m : nat =>
nat_ind (fun n0 : nat => n0 + m = m + n0)
... Isn't this similar to a Godzilla Strategy? (One AI overseeing the other.)
That variants of this approach are of use to superintelligent AI safety: 40%.
Do you have some more detailed reasoning behind such massive confidence? If yes, it would probably be worth its own post.
This seems like a cute idea that might make current LLM prompt filtering a little less circumventable, but I don't see any arguments for why this would scale to superintelligent AI. Am I missing something?
Collaborating with an expert/getting tutoring from an expert might be really good?
Probably. How does one go about finding such experts, who are willing to answer questions/tutor/collaborate?
(I think the usual answer to this is university, but to me this does not seem to be worth the effort. Like I maybe met 1-2 people at uni who would qualify for this? How do you find these people more effectively? And even when you find them, how do you get them to help you? Usually this seems to require luck & significant social capital expenditure.)
I unfortunately don't have any answers, just some more related questions:
ability to iterate in a fast matter
This is probably key. If GPT can solve something much faster that's indeed a win. (With the SPARQL example I guess it would take me 10-20 minutes to look up the required syntax and fields, and put them together. GPT cuts that down to a few seconds, this seems quite good.)
My issue is that I haven't found a situation yet where GPT is reliably helpful for me. Maybe someone who has found such situations, and reliably integrated "ask GPT first" as a step into some of their workflows could give their account? I would genuine...
And apparently ChatGPT will shut you right down when attempting to ask for sources:
I'm sorry, but I am unable to provide sources for my claims as I am a large language model trained by OpenAI and do not have the ability to browse the internet. My answers are based on the information I have been trained on, but I cannot provide references or citations for the information I provide.
So... if you have to rigorously fact-check everything the AI tells you, how exactly is it better than just researching things without the AI in the first place? (I guess you need a domain where ChatGPT has adequate knowledge and claims in said domain are easily verifiable?)
Wow had this happen literally on my first interaction with ChatGPT. It seems to be just making stuff up, and won't back down when called out.
--extract
flag. (At least not on my machine, I have coq version 8.16.0)"After a bit of testing, ChatGPT seems pretty willing to admit mistakes early in the conversation. However, after the conversation goes on for a while, it seems to get more belligerent. Maybe repeating a claim makes ChatGPT more certain of the claim?
At the start, it seems well aware of its own fallibility:
In the abstract:
In a specific case:
Doesn't mind being called a liar:
Open to corrections:
We start to see more tension when the underlying context of the conversation differs between the human and ChatGPT. Are we talking about the most commonly encountered s...
The Sequences. Surprised nobody mentioned this one yet.
While I am pretty sure you can't compress the length of the sequences much without losing any valuable information, the fact is that for most people it's just way too long to ever read through, and having some easily digestible video material would still be quite valuable. (Hopefully also by getting some people interested in reading the real thing?)
Turning the sequences into a set of videos would be a massive distillation job. On the high level it would ideally be something like:
I don't know anything about Diplomacy and I just watched this video, could someone expand a bit on why this game is a particularly alarming capability gain? The chat logs seemed pretty tame, the bot didn't even seem to attempt psychological manipulation or gaslighting or anything similar. What important real world capability does Diplomacy translate into that other games don't? (People for instance don't seem very alarmed nowadays about AI being vastly superhuman at chess or Go.)
I think we usually don't generalize very far not because we don't have general models, but because it's very hard to state any useful properties about very general models.
You can trivially view any model/agent as a Turing machine, without loss of generality.[1] We just usually don't do that because it's very hard to state anything useful about such a general model of computation. (It seems very hard to prove/disprove P=NP, we know for a fact that halting is undecidable, etc.)
I am very interested though what model John will use to state useful theorems that...
As others said here kudos for the effort, but this iteration seems horrible to me.
When I was reading the Sequences I often had to go back and reread a sentence/paragraph/even page to fully understand everything. I also had to stop sometimes to really deeply think about the ideas (or just appreciate their beauty). I feel the text has low redundancy and assumes that you can go back and reread if you missed something (would be strange if it didn't), and is not directly suitable for a video format.
I tried to watch some of the clips, but it is just waay too fas...
I see, with that mapping your original paragraph makes sense.
Just want to note though that such a mapping is quite weird and I don't really see a mathematical justification behind it. I only know of the Curry-Howard isomorphism as a way to translate between proof theory and computer science, and it maps programs to proofs, not to axioms.
We can also interpret this in proof theory. K-types don't care how many steps there are in the proof, they only care about the number of axioms used in the proof. T-types do care how many steps there are in the proof, whether those steps are axioms or inferences.
I don't get how you apply this in proof theory. If K-types want to minimize the Kolmogorov-complexity of things, wouldn't they be the ones caring about the description length of the proof? How do axioms incur any significant description length penalty? (Axioms are usually much shorter to describe than proofs, because you of course only have to state the proposition and not any proof.)
Yeah, I know you are looking for more practical advice here, that's why I posted this as a comment instead of an answer.
Eventually someone will have to aim for the "Excellent" level though (even if not against humans, surely against an AGI), and I just wanted to highlight that this is very much an unsolved problem.
In my view the field of cybersecurity currently is very far from what "theoretically perfect security" would look like. I am not sure how much ahead private knowledge is on the topic, but publicly cybersecurity seems to focus on defending against security holes already demonstrated to be exploitable, and providing some probabilistic defense against some other ones as well. This seems to work well in practice, I don't know why though. (Maybe highly motivated threat actors with sufficient resources simply don't exist?)
Conventional approaches work well if you...
probably much of what makes rationalists so male is that rationalism selects for abilities/interests related to programming, which is itself very male-skewed
This is just pushing the question one step back though, I don't know of any good theories for why software engineering is heavily biased towards males either.
One thing that annoys me with "normal" people is their inability to easily talk about the meta level of a particular topic. I feel like if I start talking about something meta some people get internally confused a bit, and instead of asking for clarification they will interpret some parts of what I said at the object level, discard the rest, and continue the conversation as if nothing happened.
Sure, you can talk about meta topics with most people with enough effort, you can try carefully prompting them (like "so what I am going to say may sound strange, I ...
food for me is fuel
https://powersmoothie.org/ maybe? It embraces this view. The cleanup consists of rinsing a single blender.
I haven't looked too much into that paper, but yeah, it could be that at smaller scales you are just perceiving the friction.
But just from personal experience, I am pretty confident that at least in the 10μm-100μm range humans have high fidelity tactile perception, and are able to distinguish various patterns and not just friction.
One big item I see missing is haptic feedback. Like, if I ask myself in what ways is VR still different from regular reality, I feel like there is still a lot missing.[1]
I think working with physical objects is a big component of activities that can't be done remotely currently. But even if we just focus on interpersonal communication, being able to touch other people is an important component. Even if we are just talking about a strictly formal business context, handshakes at least still almost always occur.
And I just don't see high fidelity haptic feedba...
To me it seems that the average best practices are being followed.[1] But these "best practices" are still just a bunch of band-aids, which happen to work fairly very well for most use-cases.
A much more interesting question to ask here is what if something important like ... humanity's survival depended on your software? It seems that software correctness will be quite important for alignment. Yet I see very few people seriously trying to make creating correct software scalable. (And it seems like a field particularly suited for empirical work, unlike alig...
I think the reacts being semantic instead of being random emojis is what makes this so much better.
I wish other platforms experimented with semantic reacts as well, instead of just letting people react with any emoji of their choosing, and making you guess whether e.g. "thumbs up" means agreement, acknowledgement, or endorsement, etc.