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 -
I'm not necessarily sure this is a bad thing: the buildout plans were so large that I didn't really think they were practical anyway. Were we really going to grow grid capacity by 300% in under a decade? I buy that it'll be game changing in lots of ways, but is the demand really there to pay for infinite tokens? The models have gotten both better and more efficient recently, too: do we need as much compute as we thought?
Slowing over-eager investment in boom-bust cycles could be stabilizing, even.
I think that we will eventually swallow all that capacity and more. Superintelligence can be used to solve, by definition, almost any problem, and there are functionally infinite problems. How to drive cars, how to design a new rocket ship part, how to identify defective potatoes on the factory line, how to build robots to paint your house, how to build more efficient chips to power the robots, how to build factories to build the chips. ChatGPT, process these dexa scans and blood work and design a diet for me to lose twenty pounds by summer. Make it more suitable for an Asian palate. Automate a weekly grocery list on a $120 budget and ship everything to my address. Find a cheap vacation rental near the beach in August. Scan this satellite imagery of Nunavik for likely undiscovered oil reserves. Design a 3D print of an art deco cornice to be glued above my mantle. Scan this image to match this color of paint. Spitball ideas for how to process and sell 3,000 pounds of spinach before it rots. Design the marketing. Anything.
That this hasn't happened yet is because 1) the AI isn't quite good enough yet 2) most people do not understand the scale of change or even what's possible.
At the same time I'm not meaningfully worried by a pause. A pause can basically be marketing. A pause might even lead to a buildout, if voters are assuaged that datacenters have been defeated forever so now we're going to give "compute campuses" a try.
Please, please, please, for the love of Ahura Mazda, stop this.
I know, it's petty, and of all the things Trump has done that I dislike, declaring that AI should be called "Superintelligence" shouldn't be at the top of my list, but it is absurdity of the highest order to issue an Executive Order based on a shower thought. "Superintelligence" actually has a meaning in the AI field, and other than trolling people I can't even see what Trump's point was in doing this. Is it supposed to "own the libs" who hate data centers? Why the heck would you immediately switch to the new term like a Party member just notified of the latest un-words?
I agreed with most of the points in your top-level and thought it was a good and thoughtful post, for whatever that is worth.
I think "deadnaming" (neologism), or not using the "prefered pronouns" (again, recently invented set phrase) was at one point declared to be against the rules of this community. As at least one not totally irrelevent AI company refers to the technology as SI, SI is at least for that company an endonym.
Is that sufficient for the technology as a whole to at least include SI in its name, if not for SI to displace it wholly?
I propose a compromise: SAI. This avoids the collision with ASI, as SI might be viewed its shortening, and also respect the self-identification of all.
Realistically you can call it whatever you want.
From a technical standpoint it is the field of Artificial Intelligence(AI) The fields goal is to reach Artificial General Intelligence (AGI) and eventually Artificial Super Intelligence (ASI).
You may call it SI or SAI all you want but the only thing you are communicating is that you don’t understand the boundary of word semantics as it relates to technical function. If you wish to convey lack of competency in favor of political tribal signaling that is your prerogative but don’t expect anyone else to join you who isn’t already in the communist party.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Sure, but the fact that we "eventually swallowed all that capacity" in network infrastructure doesn't mean Cisco was a good investment in 2000.
More options
Context Copy link
More options
Context Copy link
If a human office worker earns $70k per year, and companies are willing to pay for this, then an AI which can fully replace their job (with equal quality) is worth up to $70k in tokens to that company. An AI which can half-replace their job (reducing the need for employees of that type by half) is worth up to $35k per year in tokens.
The U.S. electric production and distribution is about 2% GDP. If AI can replace 6% of the economy then that would be worth a 3x grid capacity increase. Now, the actual tradeoff ratio would be less because the whole point is that you replace people using AI that cost less, not equal amounts. And I'm not sure that it would be practical/feasible to physically increase the grid that much that quickly. But I don't think it's an issue of lack of demand.
Surely, there is value in being independent of foreign software and the ability of human workers to learn and adapt. So the "up to" is probably doing a lot of work here. It probably varies a lot from place to place (what would it take for a hospital to feed patient data to Anthropic?), but I could easily imagine employers only being willing to replace their workers with agents if the savings are substantial.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link