site banner

Culture War Roundup for the week of September 5, 2022

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

105
Jump in the discussion.

No email address required.

First volley in the AI culture war? The EU’s attempt to regulate open-source AI is counterproductive

The regulation of general-purpose AI (GPAI) is currently being debated by the European Union’s legislative bodies as they work on the Artificial Intelligence Act (AIA). One proposed change from the Council of the EU (the Council) would take the unusual, and harmful, step of regulating open-source GPAI. While intended to enable the safer use of these tools, the proposal would create legal liability for open-source GPAI models, undermining their development. This could further concentrate power over the future of AI in large technology companies and prevent research that is critical to the public’s understanding of AI.

The definition of "GPAI" is vague and unclear, but it may possibly differ from the commonly-understood usage of "AGI" and may include systems like GPT-3 and SD.

I will be very curious to see how much mainstream political traction these issues get in the coming years and what the left/right divide on the issue will look like.

As @arjin_ferman observes, this is in line with my more pessimistic scenarios. What is AGI, people ask? Why don't they just click the link? But to be fair, it took me a little while to discover the actual definition, here («WITH PRAGMATISM AGAINST POPULISM & STAGNATION», lmao):

'general purpose AI system' means an AI system that - irrespective of the modality in which it is placed on the market or put into service, including as open source software - is intended by the provider to perform generally applicable functions such as image and speech recognition, audio and video generation, pattern detection, question answering, translation and others; a general purpose AI system may be used in a plurality of contexts and be integrated in a plurality of other AI systems;

It's trivial to realize how this applies to large language models like the GPT series, to say nothing of multimodal systems. We don't have to get even to GATO sorts of multitask training. If anyone thought the cooling effect will only start close to what we intuitively recognize as human performance: think again.

I advise people to notice how synchronized the push against individual agency enhancement is, and it's not, contra the insistence of quokka-economists, explained by innocuous market reasons like economies of scale and data moats. In the US, you have the EA movement with their longtermism, fearmongering, advocacy for «compute governance» and «pivotal acts», and independently from that – politicized corporate AI safety/fairness divisions that'll probably be used to distinguish «responsible actors» and delegitimize smaller ones on the next legislation cycle (like with Oscars: not everyone can afford the demanded diversity package). In the EU, you have this regulation circus building on pop of earlier anti-American big tech rackets masquerading as customer protection. Of course Bostrom's hand is traceable to both sides of the pond, via WEF in the Old World and LW cluster in the New. In Russia... well, if we'll have Russia still on the map in two years, they'll do good if they don't start burning their remaining ML talent for witchcraft; they're also shut out of international markets and can't acquire new compute. Japan is «LOL», as @gwern (not with us I assume) puts it – they don't do any AI R&D worth mentioning, aren't sovereign, and will meekly follow Western lead. In China, as gwern again points out, the newest American export regulations will increase the relative (although not absolute) capacity of central government and big tech, which are already paranoid and illiberal to the highest degree:

The second-order effects here would seem to confirm Chinese autarky and trends towards secrecy, and further, to shift power from Chinese academia/small businesses/hobbyists/general-public to Chinese bigtech and thus, the Chinese government. If you've been following along, the big megacorps, especially in the wake of the attempted US execution of Huawei, have been developing their own DL ASICs for a while with an eye towards exactly this sort of scenario. [...]

If you are rich and well-connected and can finance the lobbying and guanxi and paperwork, you'll be able to get access to compute, one way or another, while the small guys can no longer click 'buy' on nvidia.com or just negotiate their usual datacenter orders and will pay higher costs or go without. It's the same reason why things like GDPR always wind up hurting FANG less than the activists expect (and hurt small actors like NGOs or startups much more), why 'regulatory capture' exists and why big actors often actively lobby for more regulation. It's going to be much harder and more expensive to get Nvidia GPUs or to get proprietary hardware (can you buy a TPU from Google? no, you cannot), therefore, small actors like hobbyists will be systematically disadvantaged and many priced out.

The rest of the world (sorry fellow rest-of-worlders) is comprised of some shades of shithole and Western cryptocolony, wracked by climate disasters, brain drain and, crucially, global economic crisis triggered in no small part by the EU/American/Chinese COVID policies and now the war, extremely vulnerable even to half-hearted sanctions, and won't have the wherewithal to do ML research at scale.

Well, there are exceptions of course, hilariously two exceptions validating priors of Russian conspiracy nuts.

One is a dystopian surveillance state with legendary intelligence services and diplomatic acumen, a history of attempting and partially pulling off ludicrously illiberal tech regulations, but not (yet) any de facto obstruction on advanced AI research for smaller actors; the island where core DeepMind staff is physically located, and Stability.AI incorporated.

The other is a militarized ethnostate with infamously capable intelligence agencies, world-class lithography fabs, world-leading STEM&software talent, brazenly self-interested and defiant of international regulations, not beholden to NATO or really any other alliance, with a good track record in clandestine WMD development and non-signing of non-proliferation treaties and their equivalents.

So: USA, UK, Israel, maaaybe China if (and that's a big if) it doesn't immolate itself with its own bureaucracy and Special Military Operation in a few years. These three and a half – more like one and a half – actors will split the future of the light cone, the way it's going.

That's to be expected of course. Individual agency is a threat to big structures, always has been. Even allowing escape is a threat, an infinitely big threat when multiplied by longtermist numbers and existential anxiety. There used to be a great Motte-adjacent blog, 451somethingsomething, a few years ago, with a good article of the alienation the author felt when he noticed the vibe of eusociality and hivemindedness in the society around him, his own obsolescence as a stubborn independent cell. Not edgy, just desperate. That's kind of how I feel now.

Everyone loves weaponization narrative. Sure, every action, performed by a rational actor (even more so by coalition of actors) is calculated to secure their cozy status quo or disrupt rival one. But know what? Absence of any EU legislation would signify the same weaponization, successfully carried out by other actors like AI-powered businesses. They would have lobbied their way toward just right degree of individual agency, basically any degree you wish to pay for. Monthly agency subscription, pretty interface, but they would also hoard some details of your precious agency in the background, for its safety and for better recommendations, and maybe some other things.

Every disruptive technology would be weaponized, rest assured. And not only technology itself – that’s the bread and butter of technocrats – folks of more modest means weaponize the mere threat of technology. Even your natural claim for agency is already part of a standard sjw toolset you would sneer at in other circumstances.

Instead of pamphleteering away regulatory motion, I’d first explore the strategies they devise, and where exactly this tide is moving. Or do you already have a good strategy of decentralized resistance?