This tracks with my experience. A core and integral part to make models actually shine involves post training and custom harness engineering, all specifically done for the purpose of getting them to settle into competent inputs and outputs that are relevant to you, optimized for the harness you build which better suits your domain. The weights are just a generalization. A block of marble to be sculpted.
As the compute to build adaptions for multi-billion to trillion parameter models becomes more and more available (and affordable), and the artistic techniques of fine tuning and harness engineering spread throughout the public, I think we'll come to see the "one size fits all" model for the non-starter it really is. Anybody who's even toyed around with LoRAs (let alone made their own) already knows this. That's a very deep rabbit hole, and the ceiling is determined by target domain knowledge and systems engineering.
I'm bullish on LLMs as expert tools in the same grain that computers are. You have to learn all about them to use them effectively. But what really makes the difference is how well you know what you're going to be pointing them at. There is very little room for companies like OpenAI or Google to live between us and our tools.
Sorry not an AI specialist, what exactly do you mean by 'custom harness engineering'? Some way of informing the model whether the output it is producing is good or bad based on the specific task in question?
Harnesses are the control surface the model interacts with. How it invokes tools, the tools it has access to, how agents are coordinated. It's like an interface or a shell. It's the magic that lets an LLM operate a computer. You can read more here: https://learn.microsoft.com/en-us/agent-framework/concepts/h...
> Some way of informing the model whether the output it is producing is good or bad
That's what post-training is for. At its most basic, you're giving it examples of inputs and outputs and then doing reinforcement learning to calibrate its adaptation to your examples. You need much less to fine tune a model than you do to pretrain the weights. You can build a really excellent LoRA for a stable diffusion model, for example, with 50 high quality images. LLMs are "a bit" more complicated and costly to fine tune, and you have to be mindful of the agentic loop, but the principle is the same. There's more to it than just LoRAs. Steering vectors, projection layers, custom encoders, etc. There's a fair amount to learn, but it sounds a lot scarier than it is.
Here's something to chew on: chain-of-thought doesn't exist until after pretraining! It's basically created by having <think>...</think> blocks directly in some example outputs, and this is fine-tuned into stability. It's literally not much more than a parlor trick and some careful calibration. A powerful parlor trick to be sure, though.
No, that would be a fundamental misunderstanding of the bitter lesson, which is about research bets over time. The object of comparison is the technology, the underlying substrate and fundamental architecture, and how much of it can be offloaded to computation. We're talking about the same architecture here, they're both transformers. The difference doesn't exist in a relevant way to the question.
Given that we have an embodied intelligence that is capable of being reasonably good at physics, it would be foolish to state categorically that AI can't do physics. I am however skeptical that LLMs can do physics. My experience is that it's really great at doing the stuff I can't be arsed to do, and is therefore very useful, but it has very poor "understanding" of physics
I can appreciate your original point, but this degenerated into naivete. The is-ought problem exists in (and was formulated for) natural language, and ambiguous formal languages are childsplay. There are further problems (ignorance of the consequences of reality being finite, misunderstanding the operative layers of interpretation) but these two alone are disastrous by themselves. Don't mix up convention with implicit substance. There's a very basic philosophical lesson you're missing, and I'll let you in on the secret: The labels aren't actually descriptions of any property. The distinction is indoor baseball. Notations, syntax, semantics. What you want is signal, and you can transmit that any way you want. Writing and drawing were once the same thing. Still are.
That’s simply not possible with the announced Apple products. Afaict, Apple will let people save *transcripts* of last 15s and save *summaries* of transcripts.
Do we live in different worlds? I can't think of a single meaningful way my comfort has increased in the past 25 years. In fact, with the rise of the homeless and mental health crises, the encroachment of more and more people who have no business touching computers into cyberspace, the increased relevance of flash mob power on the internet, I would say my comfort has been actively degraded on a fundamental and significant level.
On a broader timescale, we can take a look at the fact that goods which were designed to be affordable and mass-produced have been elevated to refined luxury goods, like the Herman-Miller Eames, as an indicator that material standards have fallen very far. Not the least of which because of an ever-dwindling labor share of capital (beginning in the 1970s), which silently impoverishes working households and degrades their comfort while being completely invisible to an uncareful eye on paper.
Much of the major boons of comfort were established prior to the transition to shareholder capitalism. Things like e-commerce, IoT, online banking, these are for the birds. Meaningful advances are things like HVAC and refrigeration, ubiquitous motorized transport, electrical lighting.
I believe the parent is implying that we sit uncomfortably at home, silently, rather than uncomfortably marching through the streets, demanding. Yet, I say that while uncomfortably sitting in my home, typing on a screen
Mass demonstrations are common and frequent, more than ever. The thing is that these are mild pressures at best. What came from Occupy Wallstreet? The George Floyd protests? The No Kings marches? Anti-ICE demonstrations? They're not nothing and we should continue to have them and participate in them, but we have to recognize them as what they are: background radiation. On their own, they don't affect change or bring about course correction. We clearly need something more, people inside of these institutions, sharks who are going to fight and win against the entrenched powers on their home turf. There's a very limited pool of people who are capable of doing that.
Plenty has come from all those things. But it's those in power who want you to feel like those are fruitless endeavors. That's why their strategy is to overwhelm. Good news takes time to be received and by then your focus will be on something else. But I promise you if you look them up you'll find real progress. It may not be as much as you want, but that's a different thing. Progress is made by small and successive steps. So the strategy to upend progress is to interrupt those steps and make them feel as small as possible. Don't fall for the trap
> We clearly need ... sharks
I am afraid of sharks, just as I am scorpions. It is in their nature to draw blood. Bloodshed should be the last resort. By putting sharks in power is the problem with the current administration. Replacing with a different shark doesn't reduce attacks, it just changes targets. If sharks must be used, they must be well contained.
You have a strong point, and it's probably not the one you want to make.
Non-violent (peaceful, loud, signs) protesting does little to nothing. Historically speaking, what does work is violence, and a lot of it.
Even in the USA, our own labor battles were effectively wars, with the US military fighting against the people. It's where the Pinkertons were created. And it was bloody as hell, until FDR had some sense in making the NLRB to *curb the violence* with peaceful grievances.
When those in power forget the peaceful paths were a compromise, don't be too surprised when violence rears its ugly head.
This is how you get healthcare CEOs gunned down, or ol Scam Altman's house firebombed, or city council person get their house shot up after voting for a data center, or Flocks (YC17) getting destroyed en masse.
I'm just sitting here. And it definitely is getting worse and worse, for everyone. And some people, when you corner them, do unimaginable things.
Agreed, Viable nonviolent protests are a useful signal that society has not (yet) converted to outright authoritarianism. This is their only use IMO. They cannot effect meaningful change. But their existence signifies we can still effect such a change without resorting to violence.
There is a threshold beyond which change can only come about via violence. I don’t wish to speculate how far away from that threshold we currently find ourselves. I think we’re closer now than we were 10 years ago.
To add to your point, I'm struggling to think of any non-violent movement in history that wasn't backed by a more violent movement that made the non-violent movement the more palatable option.
Unconditionally embracing nonviolence is not, properly, a goal. Nonviolence is a strategy. It must be conditional on its effectiveness.
A useful thought experiment is which 80 year period of history would you prefer to live in other than now, with the caveat that you wouldn’t know how things turn out when you lived in that time. It’s easy to pick a single brief moment in history that is in some important ways better than right now. But many of those time periods are bookended by their own periods of strife and struggle. If you had to pick a different time period in history in which to live your life for 80 years birth to death, would you pick any other time than right now? Would you try to thread the needle such that you weren’t eligible to be drafted into the Vietnam war, but still benefit from the economic growth of the 50s-70s? Knowing that the AIDS crisis was still around the corner? And the unrest of the civil rights movement? The impending doom of the Cold War? A world where you rent your phone from AT&T, where your ability to learn about and communicate with the rest of the world list limited largely to passive consumption of limited news sources and books that might be decades out of date if you aren’t in a major metro area? Where your whole town might be dependent on a single employer who might be closing up factories and moving over seas in the 80s, or just plain closing down and taking your pension (and your local economy) with them?
Would you go further back in time and risk the Korean War or the World Wars and the Great Depression? A time when electricity was new and you might very well be using an outhouse as your toilet? When modern sanitation and health laws were in their infancy and debilitating diseases like polio ran rampant?
Would you go back further?
I for one, despite a lot of the issues with today can’t really find a different period of time I would prefer my life to span. I can absolutely identify parts of different times I would prefer to now, but I can’t see myself giving up everything I would need to to actually live in that period of time, and deal with the strife and struggles of the period at the same time.
>with the caveat that you wouldn’t know how things turn out when you lived in that time.
The premise sort of gives it away, doesn’t it? If you don’t know how things turn out you can’t say how you’d respond. Everyone is a product of their time. Some people are miserable, some are resilient.
It's a necessary restriction. Part of what drives modern anxiety is you don't know the future, so it's necessary that while you get to pick a time period where you have perfect visibility into its future (for example, you know that the cold war doesn't lead to the nuclear apocalypse) the person who lives through that moment can't have that knowledge. If you picked the cold war period, it is necessary that you pick it knowing your cold war self would NOT know that, and would be as anxious as their time and circumstances would make them with the information they would have had then, not the perfect* information you have now.
Well if you don't know anything then it is just a numbers game. Most people pre-industrial has a high likely hood of you being a random subsidence farmer. And provided you didn't pick a standout point in history you would most of the time land on and live an uneventful life as a farmer.
And I know plenty of people that would take that kind of life. They would do it today if it didn't completely isolate themselves from society. But atlest in pre-modern times you could trade some random grain or booze for some other goods and pay your taxes with it. But today neither walmart nor the IRS will accept handspun yarn or a sack grain in lieu of money.
> I really hope there is, because steering them is, I feel, one of the last competencies through which I can still add value
Something as simple as output length is a hard linear floor for productivity, even putting aside the obvious context problems that you're intuiting, and it's far from being the biggest cost that arises from steering skill. Learning to make a smaller, faster model do the same work with less tokens is a technical domain that a lot of people don't seem capable of learning. I'm not just talking about "context engineering", but learning how to fine tune, post-train, create better harnesses, design inference setups, etc. If we're both using AI, but I'm beating you to market every single time and with a better product, what is your AI usage actually buying you? Yes, competency and skill is this meaningful right now, and it's highly technical. Not the least of which because you know how to describe the problem in way that gives it a smaller solution and requires less iteration.
Most of the labor who understand the technology enough to do those things lives at the companies selling you these services, but you can absolutely learn to do these things yourself right now. It's actually really fun! A hell of a lot more fun than fucking prompting that's for sure.
Where we're at, I would equate it to the early mainframe era where the programmers came with the computer. I'm placing calls that we follow a similar track and the two will end up decoupling, that "model engineers" are going to move in-house. OpenAI will have a ring to it like IBM does today.
I think you have it in reverse, it's pure soul but very little game. It's just like Noctis IV, an interactive piece of art. This one has more pretense to looking like a game with its crafting system and survival mechanics that you can thankfully disable. It also has a narrative, several well-executed threads of them. It's a very introspective game with things to say about itself and the experience of living. You might not like it, you might have found it unengaging, but to deny it lacks humanity, artistic substance and a story to tell is crazy.
There's also not very much I would say is technically impressive about it at all, and definitely not anything people normally reach for. The bandwidth of the asset streaming they accomplished on the PS4, with such a small team that still had to build the rest of the game, is the only thing that makes me raise an eyebrow. In fact, that may as well have been the real gain of crunching down how planets are described rather than the huge number of planets themselves. Even that's not particularly groundbreaking if you take away the context of the small team. And at that point we have to ask ourselves if it's groundbreaking, or just very intelligent and clever pragmatism. Impressive in a labor lens, not a technical one.
Ha! Yeah, I probably got the soul/game bit backwards. I'm not a game developer so I don't know the specifics but I've never seen something so damn ambitious pulled off to the extent I've seen in NMS for my brief interactions with it.
I think it's less obtuse to make the point that there are social-prejudicial structures encoded in the latent space. "associative pattern seeking logics" is a bit incoherent, and distributional semantics doesn't live at a level accessible to cultural analysis and theoretics, IE film critique.
If you walked into a film theory class and posited that you could derive every single encoded interpretation of a film by memorizing the positioning of the actors and objects frame-by-frame, plus the audio track in another language that you do not speak, for every single piece of video ever made, you'd probably be asked to leave. Not that I'm advocating for the position of critique here, I don't think the anti-distributional semantics crowd is ever going to recover from their humiliation that's been accelerating over the last 8 years. It's just that from the position of critique it requires a coherent narrative that human brains are capable of ingesting (IE not maximal information overload).
If you really want to go the lower level route, I think Francois Laruelle's non-philosophie touches on what you might be thinking of in a much more robust way, shining a light on the unexamined consequences of decision and dialectics of-themselves. If you can stomach the writing of continentals, that is.
Who is this crowd specifically? The stochastic parrots people? Noam Chomsky? I don't think they're good representatives of media theory at all whatsoever. The humanities are much more diverse than they're made out to be in this crap AI culture war.
> distributional semantics doesn't live at a level accessible to cultural analysis and theoretics
They might not have computational access but the theories are all about contextuality, for example Jacque Derrida's "trace" was the first thing that came to mind when I saw this headline. Those people are tuned in on the microscopic level to what LLM researchers are bumping into on macroscopic scales. I'm thinking of post-structuralists especially. But all kinds of people and I'm sure what's happening right now is way more interesting than our crude labels ("the post-structuralists", "the anti-distributional semantics crowd") could actually do justice.
> you'd probably be asked to leave
It would depend a lot on the specific school and instructor but in general I really don't think you would. I've taken classes like this and people were far more open minded and critical than you might assume. And my broader point is that there are really sharp conversations happening in these spaces for decades.
> I think it's less obtuse to make the point that there are social-prejudicial structures encoded in the latent space. "associative pattern seeking logics" is a bit incoherent
Yeah I definitely could be less sloppy but I my point is that language encodes not just word semantics but entire ways of thinking, and they're encoded at multiple levels and in superposition. By "associative logics" I hand-wavingly mean all manner of categorical thinking, "amygdala" thinking, mapping, putting things into buckets, hedging. These kinds of cognitive habits are everywhere in language, and are culturally situated "distributional semantics" style. It doesn't surprise me that when we simulate them with LLMs we'd get results like this because I've studied a little bit of cultural theory in the past and they were on this stuff forever ago.
By the way I actually do think there's more to it than just distributional semantics, but not in any way that would downplay the potency of that theory. Moreso I'm curious about generalizations of the distributional idea into TDA and category theoretic approaches. As well, there's a lot of really cool quantum-like modelling emerging in applied math that I can only see getting more relevant if/when quantum computers come around and quantum models become runnable.
> my point is that language encodes not just word semantics but entire ways of thinking
What does this mean? More importantly, why would these "ways of thinking" (what even is a "way of thinking" in this context?)
Also, keep in mind that the training data encompasses a representative sample of world languages.
My best attempt to understand you is that you are supposing that people (or other reasoning agents that manipulate language in order to reason) do pattern matching because there's something inherent to language (as a concept, in the analytical Chomsky sense: a string of symbols chosen from some predefined set, organized according to a grammar, whatever) that causes them to do pattern matching. And furthermore that to do pattern matching is inherently to be biased.
I think that is backwards on the first count (pattern matching is reasoning, and humans have language because we developed it to communicate that reasoning) and absurd on the second count (requires an unreasonable concept of "bias").
Again, I really sincerely honestly am not trying to strawman you here. If you mean something different then I'm afraid it's simply not a concept you'll be able to convey to me.
> Who is this crowd specifically? The stochastic parrots people? Noam Chomsky?
I'm thinking more the Noam Chomsky and John Searle variety. For the stochastic parrots people, which I assume you to mean the no-skin-in-the-game bloggers, I'm not concerned about them. I find a lot of the rhetoric around LLMs to be eye-roll worthy, most people slinging it often lack a coherent theory of semantics to begin with, let alone an understanding of logical induction/statistics. Take the definitional entailment that these models are ampliative. This small fact undermines quite a lot of the naive mental model people have of what on transformers even are. No intentional theory, no position driving the argument, the opposition isn't substantial. The virtue signal is valid, but from strangers is uninteresting.
As you mention the humanities is very diverse, linguistics is no exception. It's not that nobody was on the corner of distributional semantics, but it's been a long road and for much of its life results were routinely dismissed in the mainstream out of dogmatism. Noam Chomsky is a good pull because I believe he's the most prominent example of this chauvinism. Vague hand-waives about explanatory power, etc.
> I'm thinking of post-structuralists especially.
I understand what you mean. It's probably not coincidence that Chomsky is a vocal critic of the school. I think many different post-modernist camps even beyond post-structuralism actually are amenable to the implications of distributional semantics, and are probably the better equipped for it. I think there are nuanced problems unifying the two, but it's a work day so I won't elaborate.
> It would depend a lot on the specific school and instructor but in general I really don't think you would. I've taken classes like this and people were far more open minded and critical than you might assume.
I was just being cute with that, really. Although in this case, I think critique is itself the problematic lever, acceptance is more an exercise of apologetics, which is the weak point of post-structuralism in many ways.
> By "associative logics" I hand-wavingly mean all manner of categorical thinking, "amygdala" thinking, mapping, putting things into buckets, hedging.
I assumed so, but it's more or less a long phrase to repeat the same concept. Relations (IE logic), mappings, morphisms, associations, patterns, etc. Same-side, same-coin. Informal, formal, take a position of drawing no line here and the shadows scatter. The extraneous qualifiers aren't what confused me though, it's that gerund "seeking". It sticks out enough to imply additional structure, but one that isn't contextually indicated. Taken in a conservative form, I considered pattern-seeking to be interchangeable with pattern-constructing, which then returns it to just being an extraneous qualifier (hence the "bit incoherent", it's a mild interpretive non-confidence).
Even that doesn't cover it. Vaishnavism, though the most popular branch, is only one Hindu denomination. Shaktism and Shaivism don't give Vishnu the preeminant status he's given in Vaishnavism. "Hinduism" at this point in time indicates a tradition cares about some subset and some interpretation of the Vedic texts, which doesn't actually entail all that much.
> some interpretation of the Vedic texts, which doesn't actually entail all that much
If you studied the Vedas you would know why this issue exists in the first place. The Vedas talk of One Supreme Brahman who should be given Havis during a Yajna. The dispute between the sects is on who the Supreme Brahman is identified as. Every name in the Vedas is but a name of the Supreme. Vyasa composed the Brahma Sutras to unravel it. But then, as is the case with Dharmics, we have around 30+ commentaries on the Brahma Sutras itself. This, coupled with commentaries on Prasthanatrayi as well as interpretation of Puranas created sects.
All of them come within the umbrella term of Hinduism as all of it is Dharmic and native to the Indian subcontinent.
reply