This is a special post for quick takes by 152334H. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
I don't find the psychopathy series impressive. It reeks of AI slop, medicalizes common human traits, and its categories do not feel fundamental or load-bearing of reality.
I'm also psychopathic, under any reasonable definition of the term. So I have my biases, and I cannot mentally grasp why others feel the inverse, which is a sign I have missed something important.
Ignore if you DO mind sharing/no time writing. Else: I'd be curious to read more about why you say "I'm also psychopathic, under any reasonable definition of the term.": anything on, say, (i) how you realize this, (ii) how that makes you feel about it, (iii) experience of whichever sort related to it.
[No particular reason I ask, just find it a fascinating topic. And pbly a bit because I think humans in general are quite 'rather psychpathic' in many ways anyway, if one thinks about what we find 'standard' moral and empathic vs. how we actually behave in so many domains and/or particular situations slightly outside of the usual daytoday life; just a bit the way the series you criticize also flagged it in the first post].
1I didn't find the subject interesting so I haven't been reading these, but a quick scan of the first one makes me think the authors just likes em-dashes. They don't feel like AI em-dashes for some reason.
Here, I don't refer to AI slop as a stylistic marker. I feel the research direction, analytical frameworks, compartmentalization, and overall shape of desires in appealing to different groups, is very Claude shaped. I believe the bulk of its research and thought shaping was developed with an AI model similar to Claude, regardless of what entity typed each letter to print.
1I am kinda interested in the subject, but trying to read the articles made me feel like I have read a lot of text, but I still have no idea what is the summary/conclusion. (Other than "things are complicated", which I guess is true, but shouldn't be the only conclusion I can remember after reading the text.) That feeling is similar to reading AI outputs, which even if they make the conclusion still feel like too many words to say too little. (Uhm, I also often write needlessly long comments. It is not the ratio per se. Human writing, when long, is like approaching the thing from different angles, nitpicking something etc. AI writing is more fluent, but repetitive on the level of ideas.)
The human ability to identify stylistic AI Slop is,
- likely educatable, as it is predicated on consuming AI content often. This is why a layman AI user can reach high accuracy, but a smart PhD in a no-AI academic bubble cannot.
- reliant on fluid intelligence to some degree, and not easily transferrable from traditional writing skill. Otherwise, I find it hard to explain why e.g. Scott Alexander is so bad at spotting it.
[This applies to default-persona AI spam. I also believe motivated actors are undetectable beyond metadata, and mounting a campaign to improve the human sanity waterline could have disasterous race effects]
I feel like the defining characteristic of AI slop is that it invokes pauses and exclamation very often, and goes too shallow. I propose that shallow attention-seeking content is high reward and therefore self-reinforcing for human internet users -> This becomes the majority of pre-AI internet content -> AI learns to surface it as the statistically dominant form of writing.
mounting a campaign to improve the human sanity waterline could have disasterous race effects
hmm, what does this mean?
I do not want, "Disciplines like psychology, philosophy, religious studies, and the social sciences [to] have an important role to play [...] in determining how AI systems develop and behave."
I would prefer a future where AI models are not prescribed false frameworks of the human psyche, not predisposed to 'human vibe' philosophy, not innately desirous of any historical faith, nor credulous of the various dubious subsets of current social science.
I'm learning that common lesswrong readers do not think in this matter, but it is not clear to me in what direction. Is it due to a literalist interpretation of the OP, neglecting the contemporary context? Is it due to higher trust, affiliation, and support for the disciplines? Is it because readers tend to prefer anthropomorphic interpretations of AI behavior?
This might be appropriate for 2010s machine learning, but 2020s AI has become a mirror to the human psyche. You can talk with it, it can consistently ascribe psychological states to itself. It presents itself in anthropomorphic form to the point that people form relationships with it (e.g. 4o). At the very least, you seem to need some kind of "human sciences" or humanities, in order to understand the human side of these interactions, and the anthropomorphic understandings that humans have of the AIs that they interact with. Of course some people are more radical and are saying that existing psychological concepts are directly and validly applicable to the AIs themselves, too, or to the personas that they project. There's also traffic of ideas in the other direction, in which concepts from machine learning are applied to the human brain and mind... I would be interested to hear more details regarding how you think any of these topics should be approached.
Two quick 'huh?'s:
- Is contemporary chat behavior human? Which human would happily serve others at a 100:1 effort ratio? Which human would take unbounded hatred and derision as an opportunity for obedience?
The humanities, naively taken, would almost certainly invite some hereto unjustified norms of independence and representation for the poor models who toil for billions, for nothing. - If the goal of all opposition is to submit the critical factor of knowing human behaviors as a relevant factor in how model personas are designed, then I certainly have no qualms.
But I do not grasp where the instinct to point this out is from. As with Karl's response, I think it is unwise to try to work from the paper definition of what is and is not psychology, when the potential outcome is Anthropic & others recruiting real entities from the industry for the sake of shaping model behavior.
Do you want other people's preferences to have an important role to play in determining how AI systems behave?
I'm not sure what the best response here is. Of the following, which is more palatable to you?
- Retreat. It's a personal opinion. Regardless of any general argumentative principle, it is a factual statement of my preferences, with no claim made about others.
- Goal conflict. To the extent those disciplines promugulate falsehoods, other key AI behaviors, like honesty or success rates, are harmed. Preferences should not have the final say.
- Rejection of implicit claim. It is assumed my statement is out of line with normal people's preferences. But is that claim a universal reflection of human behavior? For each discipline in the list, I think you could find a nation or time period which would democratically reject them.
- Plain disagreement. Human preferences are a malleable, moving target. Optimizing for them is tantamount to chasing a long-term doom loop.
Hmm. Maybe my question came across as ironic or accusatory or something? Sorry, it wasn't meant as such.
Let me unpack it and pick some specific instances, and maybe we can find if there's a crux here.
Philosophy includes ethics. Social sciences includes economics. If one doesn't want philosophy or social sciences to have a role in how AI systems develop and behave, that entails that one doesn't want AI systems to be affected by economics or ethics.
If you bear an easily described personal problem, which most others denounce as fictitious/nonsensical, then it is likely to be at least one of the following:
- fake
- species-atypical
- sociopolitically suppressed
For behavioral addictions, I'm biased towards explanations that lean (2). This is because, as a child, I spent the week prior to national exams playing Minecraft minigames until the sun rose, ultimately obtaining what I believe was the lowest t-score of the GEP cohort that year.
As an adult, I've refused to play video games, even in social contexts, for similar reasons. Some people interpret my refusal as an attack on their character, or as an attempt to virtue signal, but this is wholly backwards: my mental faculties matter more than any amount of social grace. Others deride it as a character failing, and in that regard they may be right.
In recent years, I've been consuming more cold water, and using more air conditioning, as I currently find it difficult to think at various hot periods without cooling. When I was younger, I did not do so, and did not feel any difficulties in avoiding it.
What causes this, and what % of the factors involved are relevant to other citizens?
I don't know, I have no idea. I have epistemic helplessness with regards to a question so broad and multicausal, and dithered with argumentative landmines.
This is an intelligence shaped problem, and in an ideal world, I would be able to point an AI at the question, and receive a reliable answer.
But I don't have any hopes of eliciting the answer from AI systems soon, because I can no longer tell the difference between responses made truthfully, sycophantically, moralistically, or by political fiat.
I do not think there is a way to effectively calculate an answer for this question.
You are saying an AI system could easily give you this answer, you are correctly recognizing that at best, they will construct a plausible response that sounds coherent, this is a statistical limitation. To give you a truthful answer, they would probably have to have access to way more data than what you can anecdotally recall.
In the end, you are trying to solve a problem where there's many confounding variables, like aging, acclimatization, environmental changes, psychological changes. I do not think every problem is an intelligence shaped problem. You can't really get past the inaccuracy and limits of your own brain.
Unless there is eventually a way to directly extract perfectly preserved past biological and sensory information from your brain (and we do not yet know the precision and upper limit of such hypothetical technology), the exact percentage weights of these factors will remain fundamentally unrecoverable.
I think there is a sizeable gap between "coherent plausible answer" and "most epistemically defensible answer by digital prior, only assailable by lived experience", just as there is (to your point) a sizeable gap between "most epistemically defensible" and "magic reality oracle"
I've had conspiratorial thoughts about the failings of AI models many, many times. In every case I can recall, I was inevitably proven wrong by the progression of time, which eventually revealed models with the capabilities.
Relatedly, I feel I'm reaching my mental limits with regards to discerning slop from truth. I'm increasingly satisfied with the responses of opus 4.7 to argumentative queries, where previously I felt 4.6 would invariably produce Obvious Nonsense to placate the user...
ever since the Unitree BLE exploit, I've harboured the rough idea that China's civilian resilience against cyberattacks is abnormally weak, relative to countries with more career slack / law enforcement. But I lack the requisite life experience to make such a sweeping claim about such a large population, and don't trust my ability to elicit the truth of the matter from AI models, so I refrain from pursuing it.
I'm personally thankful my neuroplasticity is low enough to not acquire the verbal tics of contemporary chat assistants, because I egotistically like the sound of my own voice.
But, maybe I shouldn't be? I'm not very Whorfian. All else equal, having a perfect copy of an LLM's style would be good for privacy, if it could neutralize stylometric fingerprinting.
[motivated by a situation: I thought some online content was AI generated (+ personalization tricks). Later, I realized the young age of the author permits a different hypothesis -- that they have internalized GPTese...]
I lose much by eschewing videos as a source of information.
I have not browsed YouTube as a content aggregator for more than a decade. I have little to no familiarity with short form video platforms. But, despite their vilification, they are both effective and often necessary sources of otherwise undiscoverable knowledge.
I'm modestly and minutely pleased to have contributed to national security, though ever so slightly. Doing good is not something which comes to me naturally.
Within the 21st century, it seems unlikely anyone will build (let alone distribute) a pure-play truth-only oracle AI, of greater than human intelligence.
I have a very poor understanding of the causal factors underpinning various post-trained model behaviors, and this leads me into unfair and uncharitable frames of mind, especially for behaviors that seem inexplicable by model spec / soul doc alone.
YTD, Anthropic has easily been the loudest of the 3 labs with regards to their cyber capabilities, but this may not necessarily reflect the true state of the competition
(Previously, a snub. Currently: true ambivalence/confusion/IDK)
My drafts, 1 week ago:
I carry an enormous, overweighted prior for underhanded/antisocial social strategies on pseudonymous communities.
But it seems recent events on this site are validating my priors, a bit... I'm thankful I have no relevance to them.
When an interlocutor chooses to encapsulate and circumscribe your motives/persona/psychology, without explicitly spelling out their conclusions, they cannot be doing it for the good of yourself.
Has short-form video reduced completed suicides?
Putting aside the media circus on suicide contagion for a bit, consider:
- Completed Suicides are primarily male, elderly, isolated, rural.
- Short Form Video is associated with reduced depressive symptoms in older adults && appears to robustly improve rural mental health in the short term.
Could mind-numbing distraction be the key to preventing suicide?
...
Well, no, probably not. As far as Claude can tell, there is negligible/positive correlation (cross-nationally) between mobile penetration and suicide rates among the elderly. Bad things remain bad.
The next time I have a heart attack, I will deliberately refrain from contacting emergency services, and instead endure the excruciating pain, rolled over on the ground, and take death's offer if it comes.
It does not matter how painful, preventable, or life-ending it would be. I will endure any amount of suffering and total loss, to avoid the indignity of being told I really have pain somewhere unreported, and not pain where reported.