> I think Bregman skirts close to the "Effective Altruism" movement and his work has similar problems of choosing flashy, exciting, elitist projects over boring, uncomfortable, policy changes.
I'm sorry, but what? The most prominent EA projects focus on cost effective interventions that save and improve lives.
Give Well's current top recommendations are medicines and nets to prevent malaria, vitamin A supplements and promoting regular childhood vaccinations. None of them are flashy or elitist.
You don't need to apologize! People like GP are just liars.
It's incredible the consistency with which you can go from [negative opinion about EA] to [revealed extreme ignorance of EA]. Just let the person talk for another sentence or two and they'll out themselves as having literally zero familiarity whatsoever with EA.
"EA sucks because it pursues flashy, exciting projects" is an insane position that, truly, only a liar could write.
Here's a preview since apparently there are hordes of overconfident people too "busy" to follow simple steps like these:
3 cause areas: Global health & Wellbeing, Animal Welfare, Global Catastrophic Risk
Global Health:
* Malaria
* Malaria
* Vitamin A for children
* Vaccines for children
* GiveWell, an aggregator for charities similar to the prior 4
Animal Welfare:
* Fund aggregators to increase animal wellbeing by the dozens, hundreds, or thousands, or millions of living creatures' suffering
Global Catastrophic Risk:
* Research on AI risk, bioengineering risk, and nuclear risk – all of which
AI is not only listed last, but doesn't even have its own dedicated page. It's lumped in with a set of other things that many people believe are long-term risks rather than immediate term massive problems, like childhood vaccinations, malaria, or industrial animal abuse.
> The creosote in toothache drops administered to a New York boy cured the pain, but killed the boy. This recalls the entry in the register at Bellevue Hospital, which reads; "Operation successful. Patient died."
> you may run some models locally if only from a cost perspective
I have a hard time believing running a model on a laptop will be cheaper than running it in a datacenter. Why wouldn't economies of scale apply here as with every other computation?
AI models will pretty undeniably affect your electricity bill; yes you already own the computer, but it will cost more to run it if it's doing inference!
To a point, but we're talking a laptop, not a server farm. Even if you're going fullbore wide open 24/7 that's about $150/yr in electricity bills at average rates. Not quite nothing but in terms of AI costs that's pretty close to rounding to zero.
This is assuming that you'll be priced the fraction of computing that you consumed. But you are actually paying for their infrastructure, for the R&D (and also the computation that went into training the model) etc.
It is not clear that, for your own small computations, this kind of costs are needed, but you will still pay your share in the investment the provider made so that they could serve everyone's computation needs.
But, currently ... you're not. AI companies are operating at a loss, and are being subsidized by their investors.
Local may or may not be cheaper than remote now, depending on the details, but the factors you describe won't affect the math nearly as much as they will once that subsidization ends.
In that analogy bigtech AI is currently investing in cleaner air for all of us? We _could_ breath it through their hose, but might as well breath it outside.
The datacenter setting has huge economies of scale for low-latency, just-in-time inference using extremely large models, but that's not the only viable use of AI. Batched, unattended inference of possibly smaller and weaker models, while theoretically viable in a datacenter setting, is far from the best use of that hardware. This is where local AI is at its best.
A laptop is really a pretty bad form factor to run LLMs. Worst cooling, more expensive memory that you cannot replace, resell value depreciating fast. It’s fine for tinkering, small scale research, and demos but it’s definitely niche.
The vision NVIDIA is selling is pure marketing IMHO
Does it apply for every other computation? Purely for the computation part?
You can host all kinds of things locally cheaper right now than in the cloud, no? (At least pre memory price hikes.)
It does, of course, come with its downsides like availability/reliability, less convenience, scaling options,..., but purely the computing price - I don't see why it wouldn't be cheaper in the future - at least for some use cases.
What "every other computation"? I seem to have a lot processing power at my disposal here, between my cell phones, laptops, gaming PCs, various other hardware devices.
You're going to need to analyze the problem much more deeply because it sound like the standards you are implicitly applying would result in "economically, everything should be centrally hosted" but that is clearly not the result that obtains. Even a modern mid-grade cell phone is no slouch; you may not be running a current-gen frontier AI on it but you certainly can do a lot of other rather intense things locally that would have been laughable 10 years ago, like suprisingly high powered games.
That's interesting, thanks. I only read the abstract so far but was immediately reminded of this recent HN submission[1] and the whole thing that certain ideas go together, and so they are adopted together, but the resulting bundle of ideas might be poorly suited to certain problems.
I think the original argument is too limited in its scope. The wider AI market, which is primarily fueled by OpenAI, Anthropic, Google and the large frontier labs (are there any other in the West, except for these 3?) is spending how much... $900bn this year in DC buildouts? After the spent $500bn last year and they're probably planning to spend just as much the next few years if things go remotely their way.
So yeah, I wouldn't be shocked that in the 2023 - 2033 timespan total AI investment worldwide will be around $5tn, maybe even going towards $10tn.
All that money will have to be repaid, and it will have to be repaid 10x, otherwise heads will roll.
The enshittification we've seen so far is nothing compared to what's coming.
OK so let's say build out this year is $900b. Depreciated over, IDK, 6 years (mix of 3 year GPUs and 20 year buildings)? That's $150B a year, but you want the investment to be profitable, so let's call it $250B a year.
That seems... Pretty reasonable? Like Anthropic is at $45B annual revenue, let's say they enter next year with $100B annual revenue? Let's say they have 30% gross margins (no idea), so $70B goes to data center owners/operators. That's one company doing roughly 1/3rd of what's required to pay the investments off. And you have Ant+OAI+GDM+Internal AI at GOOGL/META/etc.+all the servers for open models.
I'm sure there's a world where you can paint a picture that requires $5-10T but that would require capex increasing significantly NEXT year. And the cloud companies won't do that unless revenue keeps growing.
> The computers were very old IBM PC compatible machines, mostly with monochrome cathode-ray tube (CRT) monitors. They had no hard disks at all. They had a few hundred kilobytes of RAM. Every time, we performed the same ritual. Insert a 5¼-inch floppy disk to load MS-DOS into memory. Then insert another disk to load LOGO.COM. Then write small Logo programs and watch the turtle move.
I'm a few years younger than OP and grew up in a large Indian city, but this matches my earliest experience exactly, right down to having to take our shoes off before entering the computer lab.
I'm sorry, but what? The most prominent EA projects focus on cost effective interventions that save and improve lives.
Give Well's current top recommendations are medicines and nets to prevent malaria, vitamin A supplements and promoting regular childhood vaccinations. None of them are flashy or elitist.
reply