What will happen to GPT-{n} once LLMs contribute much of the language found online?

Absolutely horrifying thought. Wasn’t PageRank literally invented to solve shit like this?

This could create a funny circumstance, where LLM companies have to devise methods for automatic detection of LLMs text, but then normal people can just use those methods to filter out all LLM junk.

I did not realize I was talking to a debate-bro. My apologies.

I’m going to assume by this non-response that you’re apologising for a faulty analogy. It’s okay, I sometimes do it too.

There are plenty of conditions that change how people see reality that aren’t desirable.

This was a response towards you claiming that autism being part of somebody’s identity being “genetic-essentialism”. Of course there are plenty of conditions, like colour blindness and brain tumours. But I wasn’t arguing that autism is good because it’s different. I was arguing why autism can be part of somebody’s identity besides whatever genetic origins it has.

Are we going to medicalize the discussion of medicalization, then? You’re a champion of neurodivergence who casually dismisses an intervention by denouncing the researchers as “insane”? Dafuq?

Obviously I could’ve chosen better words and I apologise. But by “actual sane” I meant people who aren’t reactionary ableist bigots like those of Autism Speaks (who are not researchers). And at no point did I imply that the researchers themselves were such, though I wouldn’t be surprised. But although the word I used was unfortunate, I’ll still denounce interventions based on what I actually meant (bigoted/ableist/reactionary reasoning).

Diagnosed by symptoms

Which is different from being a symptom. You can’t just lump a bunch of unrelated conditions with possibly very different underlying causes because they have common symptoms. Like I said for brain tumours.

The article specifically calls out sever conditions associated with autism that they were seeking to treat in mice.

Fair point, I missed it. Here’s the line.

The male mice given the mutation were found to have lower levels of the MEF2C protein in the brain, and had symptoms that mimicked ASD-like hyperactivity, problems with social interaction and repetitive behaviour.

Those are definitely not what I’d associate with the worst of ASD. Nor are they very well defined (“problems”).

It’s strange though, this is a thread about autism erasure and “fixing” but you are the one getting flippant despite all major forces being at your side.

E: fixed a lot of bad grammar

Commenting again just to say I hate how medical science reporting often doesn’t even cite the name of the researchers. Not only is this very disrespectful towards the people who put in a lot of work into the research (and usually even publishing their results for the world to see), but it also makes it incredibly hard to verify the facts of the story.

Naming them only as “Chinese scientists” is just insulting. And now all search results have been poisoned by people only citing what’s in the SCMP article.

But that isn’t just genetic. Two color-blind people can have very different aesthetic tastes despite both “seeing” the same spectrum of color.

Another bad equivalence, colourblind people clearly have “less” vision in that they see less information. Autistic people usually have “different” cognitive functions in a way that’s hard to even describe in text to a neurotypical person.

But even then it doesn’t matter whether the neurodivergence is genetic or not, it has obvious and direct impact in how people see reality and themselves.

And if this is advertised as an “anti-autism jab” treatment rather than say a “social anxiety” one, I hope you’ll forgive me for disliking the obvious ableist implication that “curing” autism is desirable, even if it could be optimistically interpreted as “alleviating common autistic issues”.

This is miles away from a holistic rewriting of consciousness to be neurotypical.

That is still the end goal of organisations like Autism Speaks and I’d rather actual sane people were more careful when talking about the medicalisation of neurodivergence. We live in a world where it’s not even that hard to find stories about autistic people who basically grew up locked in medical institutions being put on all sorts of treatments because this is how our current systems treat neurodivergence. So we can’t pretend that “voluntary” will actually mean “voluntary” when push comes to shove.

It is diagnosed through its symptoms.

And conditions within the brain can produce both epilepsy and autistic symptoms.

It’s still not a symptom so “sharing symptoms” is a moot point. Some people with brain tumours experience sensory hypersensitivity, but that doesn’t mean it’s that related to autism (besides being neurological) or that some kind of autism cure will have any use for that. It’s not even clear from the article if their treatment is directed at “symptoms” or just behaviour.

There hasn’t been a historical drive to demonise blind people and their parents to this day, use of racially-driven pseudoscience to justify their mass incarceration and euthanasia or invent a whole conspiracy theory about vaccines that had massive consequences in the previous pandemic. And I’ve never met a blind person who prefers staying blind.

also are u forgetting many people with autism treat their symptoms.

I’m not, I’m only refusing to forget the many people experimented on without consent going all the way back to Hans Asperger, those who don’t wish for this treatment at all, or the historical pushing of drugs like risperidone for autistic children (often with lasting adverse effects) by the pharma-“advocacy group” alliance. And above all I don’t forget we currently live in a world where a bunch of countries can lock people up “for their own good” in medical institutions and apply treatments with barely any consent.

I don’t think it’s too outlandish of a scenario to imagine “experimental gene treatments” being imposed on a bunch of children due to pharma companies preying on desperate parents.

To be fair I’m not up to date on the debates among researchers nowadays, but I think it could be possible for there to still be a parallel. Until fairly recently the medical consensus on LGBTQ people was that they were mentally ill, and specific examples of people who were both mentally ill and LGBTQ were used to discriminate against the entire group.

But nowadays we have a different understanding that queer people are just people and the mental illness bit is just because people are often mentally ill (and also because of a lot of correlation with trauma, discrimination, bullying and social pressures).

I wouldn’t be surprised if a similar trajectory happened to autism and the classification of type 2-3 autism got reformulated into separate categories.

But even if my analogy was worse than I though, I think my point still stands. The most enthusiastic supporters of something like that won’t be actual autism advocacy groups, but shit like Autism Speaks, and legislators surely aren’t going to listen to actual autistic people. In the case of autism, they can even claim that “mentally ill people can’t consent” as they’ve already done with sectioning.

Since the title already has “autism jab” in it it’s worth noting that the very first “vaccines cause autism” study did a lot of unsafe, traumatic and anti-ethical tests on autisc children with basically no informed consent even from the parents.

What’s more, I wouldn’t assign my the majority of my behavior, attitude, and temperament to genetics. A big part of it is my upbringing and another big part is my immediate environment.

This might be the bit causing confusion, they (and also a lot of autisc people including me) assign a lot of who they are to how they perceive and engage with reality. Autistic people usually have different cognitive functioning, and at that point genetics is just a distant cause at best.

“Curing” autism would mean somehow changing that functioning to a neurotypical one, and that would definitely impact somebody’s identity or behaviour. Also autism is often termed a “behavioural disorder”, so if people are claiming to “cure” autism they most likely mean the behaviours go along with it.

There are plenty of syndromes that exhibit both as symptoms.

This might be just a mistake, but autism is not a symptom and even then the thing there isn’t being advertised as a “tactile hypersensitivity jab” or something like that. Autism is at worst called a spectrum “disorder” with an assortment of “symptoms” in common

Epilepsy on the other hand is indeed a disease. Lot’s of people (like me!) are really fine with staying autistic for the rest of their lives, and would much prefer research be directed at accommodating autistic people. On the other hand I think you’d struggle to find people who are glad to keep their epilepsy.

A sad development, I’ll now become an ultra-left reactionary maoist and hate on Xi specifically for hating autistic people.

I think the option should be available to autistic people, and it should be their choice to take it.

Bad analogy, but a similar thought process could be applied to being gay. The reality is that the majority of people researching, allocating funds for or marketing these “solutions” are usually neurotypical themselves. I really don’t want to see how tenuous the definition of “volunteer” is going to be if this ever gets to human trials.

There are obviously lots of applications in the realm of automation for AI, but I think where it could become game changing is in terms of large scale planning. For example, an AI could monitor usage of resources and allocate production and allocation of these resources in real time. This would allow for unprecedented level of economic planning efficiency. China already has a huge amount of automation and robotics in the industry. Imagine that being coupled with automated planning. Another important use could be watching global trends. An AI could potentially predict global economic downturns, wars, pandemics, you name it. A country that has such a predictive engine would be able to mitigate the impact of such events a lot better than others.

This is possibly the best summary on what direction I think AI should focus on. Right now we have way too many AI research orgs focusing on human-facing systems (chatbots, robots, AI art) that are neat, rather than optimisation engines that can revolutionise an industry.

I don’t know much about the history of it, but during the Cold War there was a bit of a “silent revolution” in the area of Operations Research led simultaneously by Soviet mathematicians trying to model a planned economy and Statesian military modelling their gigantic supply lines. Neural Networks (which is what people usually mean by AI) opmisation algorithms were an offshoot of that area, but sadly advanced material on stuff like “constrained non-linear optimisation” is on very few university curriculums so few students realise the connections and apply the new methods to the age old problems.

Stafford Beer (the Cybersyn guy) was one leading expert in the area.

“Towards New Socialism” by Cockshott and “The People’s Republic of Walmart” by Phillips are up next in my reading list and I haven’t read much, but seem like good books to understanding how the massive improvements in the area of mathematical optimisation (of which Neural Networks are a subset) could allow for an even better planned economy.

Both of those sound like backhanded insults lol.

They could enact this thing irl with you.

Moore’s law (and apparently now Moore himself lol) has been dead for a while and cores can’t really get that much faster due to power dissipation. On the other hand there are some really harsh physical or theoretical limits to parallelisms (besides the points in that article, I’ve heard it’s incredibly slow to have memory busses for the same memory in more than 16 cores). I wouldn’t place any bet that we’re going to get any more computing power than what we have.

The solution (to most of computing, not just AI) right now is to roll up the sleeves and start writing actually efficient software because we can’t expect that the next generation of Intel processors will necessarily make our optimizations redundant anymore.

That includes developing (possibly slightly worse but) more compute-efficient ML models, but the big AI boys are allergic to that because they rely on funding or are directly owned by the biggest cloud server providers.

I’m pretty sure this is just a hype/marketing thing for investors and laypeople. They’ve had it there since the very beginning (and always pretended it was a feasible goal), but haven’t made any strides in that direction so far.

It’s kinda like if some company CEO put forth some mission statement for his company to “use Mars colonisation to the benefit of all” then spend the following 20 years making cars and buying social media. Hypebros will look at that charter and embody that hexbear emoji of the guys pointing at a thing. IMO they’re a net negative for the serious Machine-Learning research scene.

The paper in question for the nerds: https://www.nature.com/articles/s41586-023-06558-8

Right now it seems they depend on direct optical input, so it’d be harder to adapt it to other deep-learning techniques. It might be useful down the line for stand-alone computer vision parts for micro-controllers. If they manage to develop this photonic computation tech to a commercial and more general-purpose scale, it could seriously revolutionise the GPU industry too, specially on the energy efficiency front. Really impressive stuff.

Yeah, and it gives a lot of local incentive to developing their own FOSS solutions (or learning how to pirate). It has become the norm in Brazil to replace older and more reliable linux distros in government machines for Windows, to the point where we went from having a bunch of specifically Brazilian distributions for many specific purposes to being reduced to using debian or ubuntu. Relying on foreign software can be really harmful to your sovereignty but also create a dependent IT workforce who don’t know their own national tech.

But if they blow up every house, there’ll plenty of housing to be built later on by EU companies. That’s basically the same thing, right? That’s what the citizens want, right?

This is a classic case of tragedy of the commons, where a common resource is harmed by the profit interests of individuals.

No, it’s not. It would be if all of this content was licensed as Creative Commons and had no author rights, but as could be seen a while back with the openai shadow library thing, these corporations are actively stealing content to train their models. This is corporate theft, but the culprits are too prominent and rich to ever face any repercussions.

And it’s not like it’s a new thing, Google Images was based completely around the idea that, if an image is on the web, Google somehow has the right to store it in their own servers and present it to users with ads. Small youtubers had to go through years of getting their videos randomly claimed through content ID by well known and huge scam accounts, creating a whole business out of pretending to own other people’s stuff. M$ Github trained copilot on repositories without any regard for it breaking copyleft (i.e. no attribution in case of replicated code).

It’s standard practice to simply not ask permission before potentially algorithmically wrecking somebody’s livelihood for these companies, sometimes not even informing them of afterwards (i.e. the mystical YouTube “Algorithm” that keeps changing without so much as patch notes, and the cargo cult it spawned).

For them (and for the rest of the ruling class, obviously), (intellectual) property rights exist in a hierarchical level, and so long as you are on the top everything below you is free real estate.

And the best part is that these “AI” content generation systems are still comically bad when actually put into practice, but the corporations feel the unending urge to deploy them ASAP because they’d rather have a lot of complete garbage content to be consumed, than to pay living wages. Burgers will literally employ rotting zombies if it means they can skimp on salaries and increase unemployment to drive down the rest of the wages.

Got a bit worked up there lol.

Please elaborate on what’s wrong with the article. Proclaiming “biased bias” and quoting mystical nameless “mainstream scholars” ain’t scientific.

And also explain who those Russian vacationers are who managed to singlehandedly declare two independent republics and a separatist movement. Might as well give me your Fantasyland timeline of the war against the DPR and LPR, because you are quite vague in everything you say and I can’t read your mind.

I assume you didn’t read the second one, about your claim over the snipers. Go back and read again. And who started the Donbas war?