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.
Jump in the discussion.
No email address required.
Notes -
Color me skeptical that this is much more than a mildly iterative improvement at best. LLMs have largely either stagnated or even slightly regressed since the big hype cycle last year, at least for everyday uses. There have even been a handful of people breathlessly claiming we'd hit the singularity, only for their claims to be proven as nothingburgers within a few days.
There's a chart later in the press release that shows whether people subjected to a blind test prefer the current models or the new models. For writing there was essentially no difference, for programming and data analysis there was a minimal difference, and only for math was there a moderate difference.
It does seem like a lot of progress now is consolidating gains.
As I mentioned before, LLM's already had the ability to write a program to calculate the number of R's in strawberry. But they still gave the wrong answer instead of using the program. Similarly, a lot of "hallucination" could be fixed by simply incorporating databases.
I think there's a lot of low hanging fruit. But maybe the "one weird trick" of just increasing model size has hit its limit.
The point of "low hanging fruit" is that it's easy and quick to pluck, which means rapid gains. This is the opposite of that. This is grinding away at the margins for tiny, nearly imperceptible improvements.
They've pulled out the ladder and are scaling to the top of the tree to scrounge a handful of "strawberrries".
The US tech titans are collectively ploughing at least $100 billion into AI capital spending per annum. They are absolutely determined to reach the top of this tree, no matter the cost.
Tech companies ploughed insane amounts of money into stuff like the metaverse as well. Heck, Facebook even changed its name to Meta, yet all we got were some dead malls.
One tech company ploughed moderate amounts of money into the metaverse (about $20 billion total?), all of them are pumping insane amounts of money into AI.
There's a qualitatively different atmosphere between AI and the metaverse, you don't see the US restricting VR tech exports like they are AI tech. AI is just better, LLMs are used in so many places (writing, images, music, code, translation...) whereas the metaverse only exists in VR.
Facebook was hardly the only company investing in the metaverse. If VR is included as well, which is at least adjacent to concepts of the metaverse if not outright intertwined, then the amount of investment almost certainly rivals what LLMs are getting (though data is sparse and a true apples-to-apples comparison is difficult).
I have more faith in the long-term viability of AI than I do of the metaverse + VR. That said, I find the idea of a near-term AI breakout to be highly improbable and decreasing with every mediocre release. At best, LLMs have years or perhaps even decades of research ahead before we get to human-level intelligence. There's also the possibility that the current generation of LLMs will eventually be seen as a dead end and AI focus will shift elsewhere, like how game playing used to be the forefront capturing headlines in the 90s before tapering off into relative obscurity.
More options
Context Copy link
More options
Context Copy link
While I agree that the 'metaverse' was useless crap, much and most of the Facebook money has been going more to VR hardware and infrastructure. That has been getting remarkably good, both in terms of direct stats (resolution, image quality, refresh rate, headset weight, tracking fidelity, passthrough video quality) and user experience. The StreamLink/AirLink over Wifi6 works with surprising quality, and even five years ago that was a ridiculous unbelievable pipe dream to get as even some glorified iPad Game gimmick.
It's just the business case kinda sucks.
VR can become a lot better, but it will never solve the fundamental technical hurdle that has always held it back: It's supporters bill it like it will eventually usher in Ready Player One, but without physical movement the tech just sucks. All demos have to make critical compromises with how the player or user moves around, and they usually do it by teleportation or by being on-rails since the alternative is a one-way trip to the floor as the user's cochlea decides that remaining upright is no longer an option.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link