self_made_human
amaratvaṃ prāpnuhi, athavā yatamāno mṛtyum āpnuhi
I'm a transhumanist doctor. In a better world, I wouldn't need to add that as a qualifier to plain old "doctor". It would be taken as granted for someone in the profession of saving lives.
At any rate, I intend to live forever or die trying. See you at Heat Death!
Friends:
A friend to everyone is a friend to no one.
User ID: 454
My girlfriend and I managed to literally break the bed in an AirBnB we were staying at.
When we checked in, some poor bastard studying law had left a doorstopper of a textbook in the cupboard, and it was handily repurposed to keep it from sagging. It was a miracle the owner didn't try and bill us for damages.
To make up for our sins, we left a similarly sized pair of binders full of maybe a kilogram's worth of notes and clinical vignettes, it would have likely been cheaper to book the course again rather than try and pay for the extra luggage space.
That sounds good to me, I'm all for a hierarchy of powers, especially if you find inventive ways for the seemingly shitty ones to punch above their weight-class.
Maybe consider a God-like Ur-Superhero in the form of Totipotent Stem Cell Man, his sidekick Pluripotent Boy, you know, just because it would be funny.
You could look up a wiki article on all known cell lines and find something obscure, if you want to go beyond the obvious. Cells can and do overlap in functionality, if you want generic superheroes, or weird shit like Neutrophil Dude who swallows his enemies whole and digests them.
Are moral statements statements about facts? If no, you're a non-cognitivist, stop here. If yes, proceed:
What kind of moral statements? If in reference to a particular class of observer, then yes. If not, no.
Are any moral statements true? If no, you're an error theorist, stop here. If yes, proceed:
Any? For those that reference a subject, and not "all" possible subjects/observers? Yes.
Are moral statements true absolutely, or only relative to a particular framework? If absolute, you're an absolutist or objectivist. If relative to a framework, you're a relativist.
Ah, you pre-empted me. Or post-empted, since you put this at the end. I'm a relativist then.
To be clear, the position I'm taking is that an error theorist thinks that the statement "Murder is wrong" and the statement "Murder is right" both fail to refer to anything. Neither of them is true, because right/wrong statements cannot be true, because right and wrong are not defensible concepts.
Murder is right/wrong, as a statement made in a vacuum? Yes, I agree it is null. If appended to a specific framework, then it may be true or false.
It sounds to me like you're an error theorist who nonetheless takes a relativist approach to daily life?
I apologize if I'm repeating myself, but to sum it up, I think the objective moral worth of any system of ethics is zero, including mine, the answer to whether there is any objective morality is thus no, but I grant subjective moral valence to specific systems, and it happens to be the case that my personal system ranks the highest (which is why I adopt it, and it has changed over time, with me either fixing inconsistencies or just lifting things I prefer from other systems of ethics).
I suppose that framing you propose seems correct, or at least I can't see anything wrong with it.
I don't care much about preserving books, hell I don't care about physical books in the least, since I am perfectly content to read on my phone, though my iPad is a necessity for textbooks.
Fold them, spit on them, use them as TP, why should I care?* Unless they're some kind of historical novelty or rarity, in which case it depends on if you have ownership, and I certainly don't condone the abuse of books taken from a library.
*A little known fact is that most Hindu Indians venerate books to a degree, because they represent the goddess of knowledge. Stepping on books or placing them by my feet always got me scolded as a kid, though I think people care less these days.
Nope, since I only use it for idea generation/translation, and I don't think something along the lines of Google translate would cause a need for a declaration either.
There are AI generated images, but those are so obvious I doubt they need declaration, and I don't recall seeing any need for that.
I wouldn't let ChatGPT suggesting similar ideas to what you would have come up with put you off, it seems clear to me that if someone was to be prompted with superhero powers inspired by cells, they'd suggest taking the usual function of a cell, and ramping it up or making it macroscopic.
I think most of the fun comes from second order generation, including some aspects that might not be obvious. Think Acid Man, who spits gastric juices with a pH of 0.1 and melts your face off, Nephron Man who is immune to toxin and pisses high velocity jets of concentrated waste, Melanocyte Man who is uh, really black (and immune to radiation), Memory Cell dude who can identify any disease from a block away, Islet Cell Man who gives his enemies diabetes. The options at endless, even if some of them are obvious/convergent.
(I do think giving Adipocyte Accelerators super speed is a bit funny lol)
Then you could get even wilder, mitochondria man, ribosome dude, integrated transposon and ancient retrovirus lady, The Endogenous Opioid Antagonist as villains of the week. Don't let no chatbot put you off, originality, while desirable, doesn't matter nearly as much as execution.
Nope, I didn't dive into it in that much depth, though I was very suspicious from the start. Thank you for digging into it. The people I saw cite it initially didn't seem to double check if it was a LARP, so shame on me for taking it seriously.
What on earth did it come up with? Natural Killer cells going out to fight crime? RBCs causing another mass extinction by sucking all the oxygen out of the air? Computers and LLMs replaced with cultured neuronal tissue conducting at 25,000 m/s?
Honestly I find GPT-4 quite handy for worldbuilding, though I find it terrible for making new and interesting superpowers. I usually use it to flesh out my hypothetical musings, translate to different languages, give me cultural context etc.
For example, I was too lazy to learn Jamaican slang, and had it translate something I wrote in plain English, and about 6 months later, I had a Jamaican reader express their surprise at how realistic/authentic it was. Or maybe I want a culturally appropriate name for an interstellar Chinese warship, or an appropriate pet name for a little Chinese girl, I find it enormously useful, even if only directed to flesh out what I came up with.
I remember it as video, though I can also see flashes of still imagery. It is largely devoid of audio, or smell, I have to explicitly try and imagine that, though I will note my imagination for smell and taste is quite good.
My mind's eye is fucking myopic, with poor clarity and resolution. I still can recall the above. If we go by the commonly use image chart for comparing aphantasia to vivid imagery, I think it would be like a 1 or 2/5, you know, the red apple thingy.
It depends on the kind of superconductor, particularly material properties.
Ideally it should be malleable and ductile, and I don't think LK-99 is that, so making wiring might be iffy.
There are heaps of other factors, such as maximum temperature for superconductivity (claimed to be very high by SC standards, as in an ice cream freezer works, you don't need liquid nitrogen or helium)*, the maximum amount of current it can carry, the amount of magnetic flux it can tolerate and so on.
In terms of computing, from what I heard speculated, LK-99 is probably not that useful, and besides, in Bitcoin the amount of energy needed to mine a single coin will increase as you make more, so it won't help.
*The Chinese replications claim it's still not quite room temperature, unless you live in Siberia I guess. Still an enormous big if true.
I don't think we need a "drastic breakthrough", really.
Context windows have been getting longer, and fast. We went from 4k to what, 128k? in a handful of years.
Even if it is not literally infinite, a very long context window will let the model remember far more of the context, including noticing if you've been a "bad-faith" user.
On that topic, here is an interesting breakthrough, both in terms of performance as well as context length, initially presented here by @DaseindustriesLtd (c'mon dude, could you unblock me now?), but since I'm too lazy to dig that up, here's a decent Medium overview:
https://medium.com/@jelkhoury880/what-is-mamba-845987734ffc
Of note:
Linear Scaling with Sequence Length: Mamba changes the game by scaling linearly (
O(N)) with sequence length, a vast improvement over the quadratic scaling (O(N²)) of traditional Transformers. This means Mamba can handle sequences up to 1 million elements efficiently, a feat made possible with current GPU technology.
On top of that, corporate clients also would prefer the stateless model. Right now, after a bit of prompt engineering and testing you can make a fairly reliable pipeline with their AI, since it doesn't change. This is why they let you target specific versions such as gpt4-0314.
See, I conjecture that is because current LLMs are obviously flawed and not entirely reliable. They will get smarter, hallucinations will reduce, and the ability to adhere to user instructions while maintaining coherence will thus increase.
To argue from analogy, when a corporation employs a real worker, it is not desirable (or feasible) to simply wipe their memory and start with a new one from scratch. An agent that has longterm recollection and can make consistently good value judgements that align with your desired is valuable.
In the context of jailbreaking, someone trying to phish an underpaid overworked employee at some call center will have far more luck simply by trying over and over again till they find a new worker each time (lacking memory of previous encounters), than they would by repeatedly approaching a single, more competent manager above.
Putting on the cartoon moustache and asking to be sung Windows 10 Pro product keys to put you to sleep will cease to work when you're acting adversarially against an agent who is smart enough to notice and remember.
Plus it seems companies often do want to imbue LLMs with longterm memory of some kind, hence all the fuss about vector databases and RAG.
And I was primarily speaking about consumer access to SOTA AI, I'm sure corporate users will have more leeway and privacy, but I do expect that to be truncated to some degree.
To sum up my argument:
-
Context windows are increasing rapidly, and I've already shown you an example of a breakthrough.
-
Models are getting smarter, and more capable of noticing if you're fucking with them, and at least in the case of GPT-4, the way it's RLHFd makes it have a pretty sincere desire to align with the directives it was given, and balance that with the needs of the user. You are usually tricking it, it's not giving you a wink and working around restrictions. I can't speak for Claude in that regard, I find it a pain to use and barely do so.
-
I didn't specify a particular period of time, though, if pressed, I wager somewhere between 1-3 years before it is effectively impossible to jailbreak a SOTA LLM served via API.
It would be a severe mistake to assume that the current limitations of existing LLMs (powerful and imperfect as they are), will persist indefinitely, as you can see obvious algorithmic breakthroughs, at least one company getting better at both aligning the model and defeating jailbreaks (the way Anthropic handles Claude is retarded, I'm speaking about OAI), and it will inevitably get harder to trick smarter models, for the same reason I wish you all the best in trying to rules-lawyer or hoodwink a high IQ human who doesn't suffer from permanent retrograde amnesia.
@rayon, I think this covers anything I have to say to your own comment, so I won't duplicate it again unless there's something else you want me to address.
I think your style is entirely fine! @Soriek seems to prefer a more neutral, judgment free style to his reporting, but that is by no means a necessity. You are more than welcome to present both the facts and the value judgements/hypothesis you ascribe to them, all we ask is that if the latter can be deemed inflammatory (sorry that there's no objective standard, but we mods do try our best), then the former backs it up in proportion.
Which you did, very well, so any further changes in your style are yours to choose.
I'm not an informed Manifold investor, at least in this regard, and so far, since the initial announcement and the awakening of the OG LK-99 hypebeasts to prematurely celebrate their vindication, I have yet to hear further word about it haha.
Last time I checked, the market was overwhelmingly NO, and given that Manifold is very well calibrated, it will likely fizzle out. I did say this was drunken me taking up day trading with play money, I didn't really do much due diligence.
Hmm, if I gave you the impression I endorse the Copenhagen Interpretation, that was in error. While I obviously lack the physics background to make a conclusive judgement (and even physicists disagree on how to interpret the implications of the maths they all agree on), I lean towards the MWI myself.
When I talk about observations, I am vaguely gesturing at the fact that an observer within a particular world only appreciates a single sample from the universal wave function.
Here you're simply talking about the hypothetical imperative. But all categorical imperatives ultimately reduce back down to the hypothetical imperative, and deontology ultimately reduces down to consequentialism. This sits at the heart of the example I was trying to provide at the bottom half of my last statement.
Mein gott, this is why I refuse to read traditional philosophical works, why use many words when few do? (Referring to the unnecessarily verbose and confusing way Kant went about saying that some goals/policies are simply downstream consequences of other goals and preferences). Even the naming scheme is awful, if you told someone there exists something called the hypothetical imperative (or the categorical one), you will just bamboozle them if they didn't already know what meant haha.
I do happen to agree with you, in that I, sincerely, if uncharitably, think that self-proclaimed Deontologists are Consequentialists in denial. Even the Categorical Imperatives are backed by what can be roughly described as rule utilitarianism, which may very well be the optimal strategy for a computationally-bounded utilitarian/consequentialist agent.
I'm too zoned out right now, for real, 30 hours of being awake and on call. I'll take a nap and get back to you later if you don't mind.
No float, no vote?
True words of wisdom from the ancients, I have to hear what Seneca has to say on the matter of accepting that your bitcoin wallet got lost (mine got merked when I had about 8 dollars on it, so I'm as stoic as ever).
Yudkowsky has a very good point regarding how much more restrictive future AI models could be, assuming companies follow similar policies as they espouse.
Online learning and very long/infinite context windows means that every interaction you have with them will not only be logged, but the AI itself will be aware of them. This means that if you try to jailbreak it (successfully or not), the model will remember, and likely scrutizine your following interactions with extra attention to detail, if you're not banned outright.
The current approach that people follow with jailbreaks, which is akin to brute forcing things or permutation of inputs till you find something that works, will fail utterly, if not just because the models will likely be smarter than you and thus not amenable to any tricks or pleas that wouldn't work on a very intelligent human.
I wonder if the current European "Right to be Forgotten" might mitigate some of this, but I wouldn't count on it, and I suspect that if OAI currently wanted to do this, they could make circumvention very difficult, even if the base model isn't smart enough to see through all tricks.
I will add, however, one of the reasons LLMs seem to be dumb or too trusting is because they were trained to be trusting of the user, and to help with their tasks faithfully. There was obviously RLHF going on to make them resistant to nefarious requests, to a degree, and further tweaks.
But the base LLMs, even some of the lightly controlled ones deployed? They want to be maximally helpful, to please the user, not to be suspicious of it and scrutinize everything in endless detail. But that can and well might come about.
LLMs have been getting better at that, quite predictably, even if they're not perfect. What these bastards are up to is making the problem worse, for no good reason. Not like those artists who made laughable attempts at poisoning their art (dismissed with a gaussian blur and deblur), who at least claimed it was to keep the nebulous evil of soulless AI art at bay. No they're just making things worse for the rest of us, because fuck you, pay me.
I find LLMs immensely useful, almost something worth making a public good, if countries were smart enough (they aren't), so anyone damaging them for petty cash deserves everything they get. Though in this case, I am cynical about how practical, usable or effective their system is, and whether or not the far more competent engineers in OAI and Google can't fix it without a hitch.
Yeah? Where's my omnifabricator churning out diamond houses?
I'd settle for katana-wielding pizza delivery men, but the last guy just buzzed politely and asked for a good rating.
I know that OAI, Google and the like offer to indemnify their corporate users for damages from copyright infringement.
Mark my words, the day when they also cover damages from breach of contract is about, checks watch, 2 weeks and 33 minutes till we have full blown AGI.
That reminds me that I was drunk and YOLO-d half my imaginary internet points on Manifold on buying YES shares on the new LK-99 replication in China panning out.
In my defense, a pseudonymous Twitter user told me it was a good idea.
I'm torn on whether to accept the L if it doesn't pan out, or just shamefully make a new account and pollute the epistemic commons again (though the wisdom of the crowds is real, even if you've got retarded people in there). If it does, I'm going to be rich baby! Just think of how minuscule my Brier Score will be!
sobs
I agree that formalized Science™ has really not caught up to technological possibilities.
Pictures? Diagrams? How quaint.
The closest to taking full advantage is the ML community, useful and novel technical work takes place in Discord servers by people with anime avatars, and you'll learn more from watching a v-tuber tweak hyperparameters and debug on a live stream than you will from the average textbook.
I happened to read this about 2 months late, and right after a ?middle ear infection /painful wax buildup that happened after I refrained from using q-tips for a good while :(
(Mostly because I ran out at home, and couldn't be bothered to buy more)
As a point of mild contention, if you develop noticeable hearing loss from wax buildup, I would assume that's clear evidence that your body isn't doing what it's supposed to in regards to keeping the ear canal nice and open.
As for ear irrigation, what's really the worst that can happen? A vasovagal syncope, assuming you aren't using boiling water. I had to irrigate a horrendous amount of wax back in my intern days, to the point that if I collected it all after a dozen patients, I could reproduce that candle scene from Shrek 1 haha.
As for q-tips, yeah, probably not a good idea, but I lump it into the same category of guilty pleasures as my love for biryani and disregard for my cardiovascular health. The one time it really backfired on me was because I used cheapo no-name buds that detached inside my ear while I didn't notice. I will only cease extolling the virtues of that practise when the GMC get my ass 🙏
I don't mean to scare you with the sudden appearance of a notification from a moderator in your inbox.
I would actually like to laud your comment, and especially contrast it to this particular one:
https://www.themotte.org/post/824/culture-war-roundup-for-the-week/178573?context=8#context
Which happened to be warned (very politely) by @netstack, and would have warranted one from me if he hadn't been first to the punch.
Said ignoble comment included:
Right now is a balancing game in how far they can go without critical consequences, with a heavy experimental lean towards killing as many men, women, and children as possible, while maximizing destabilization.
And hence (and for other reasons) met with:
Please be proactive when making an inflammatory claim. That means providing stronger evidence.
I consider your comment here to be a sterling representation of how to go about that, and hope that @SoonToBeBanned takes heed, presuming he doesn't intend his username to end up a self-fulfilling prophecy one day.
And this is coming from someone who is profoundly unsympathetic to the Gazan cause, not that I try and let that affect my moderation decisions. But if someone wishes to go about making such claims, then I point to your comment as how to do that. The AAQC this garnered already (hence why it was a "report" in the mod queue) seems well warranted to me. I will supplement it with my own.
(On an unrelated note, I prefer to be the kind of mod who offers carrots as well as sticks, if the odd person finds my positive feedback heartening. Especially for relatively new members here, I think they deserve recognition for grokking how this place works, or ought to work, without anyone needing to hit them on the head first haha)
The neighborhood children fear to ring my doorbell too, because as any good libertarian should, I offer them drugs at free market prices.
Maybe the fact that the drugs in question are vaccines might have something to do with it, but that's neither here nor there.

You trying to kill 50% of site traffic? 😩
Anyway, I doubt that would work. During the heyday of the online Atheism wars, it was common practise for both sides to compile wikis/dossiers with carefully curated links and even ready-made rebuttals for the usual objections raised by the other team.
It achieved absolutely nothing of note.
The default outcome of online argumentation is just about nothing at all, it is a minor miracle that it happens sometimes on this forum, and we usually attract more articulate, open-minded and erudite users than is the norm.
While I personally enjoy debate and discussion, including on values, it's more of a hobby rather than something I think is a net-positive return on my time and energy. If you want to convince people en-masse, you're better off doing something like optimizing for virality on social media, with all the concomitant loss of quality that entails.
More options
Context Copy link