Technology

Optimizing AI performance by “scaling” is lazy and wasteful.

Reminds me of back in the early 2000s when someone would say don’t worry about performance, GHz will always go up.

They're throwing billions upon billions into a technology with extremely limited use cases and a novelty, at best. My god, even drones fared better in the long run.

It's ironic how conservative the spending actually is.

Awesome ML papers and ideas come out every week. Low power training/inference optimizations, fundamental changes in the math like bitnet, new attention mechanisms, cool tools to make models more controllable and steerable and grounded. This is all getting funded, right?

No.

Universities and such are seeding and putting out all this research, but the big model trainers holding the purse strings/GPU clusters are not using them. They just keep releasing very similar, mostly bog standard transformers models over and over again, bar a tiny expense for a little experiment here and there. In other words, it’s full corporate: tiny, guaranteed incremental improvements without changing much, and no sharing with each other. It’s hilariously inefficient. And it relies on lies and jawboning from people like Sam Altman.

Deepseek is what happens when a company is smart but resource constrained. An order of magnitude more efficient, and even their architecture was very conservative.

The actual survey result:

Asked whether "scaling up" current AI approaches could lead to achieving artificial general intelligence (AGI), or a general purpose AI that matches or surpasses human cognition, an overwhelming 76 percent of respondents said it was "unlikely" or "very unlikely" to succeed. 

So they're not saying the entire industry is a dead end, or even that the newest phase is. They're just saying they don't think this current technology will make AGI when scaled. I think most people agree, including the investors pouring billions into this. They arent betting this will turn to agi, they're betting that they have some application for the current ai. Are some of those applications dead ends, most definitely, are some of them revolutionary, maybe

Thus would be like asking a researcher in the 90s that if they scaled up the bandwidth and computing power of the average internet user would we see a vastly connected media sharing network, they'd probably say no. It took more than a decade of software, cultural and societal development to discover the applications for the internet.

Technology in most cases progresses on a logarithmic scale when innovation isn't prioritized. We've basically reached the plateau of what LLMs can currently do without a breakthrough. They could absorb all the information on the internet and not even come close to what they say it is. These days we're in the "bells and whistles" phase where they add unnecessary bullshit to make it seem new like adding 5 cameras to a phone or adding touchscreens to cars. Things that make something seem fancy by slapping buzzwords and features nobody needs without needing to actually change anything but bump up the price.

Me and my 5.000 closest friends don't like that the website and their 1.300 partners all need my data.

I liked generative AI more when it was just a funny novelty and not being advertised to everyone under the false pretenses of being smart and useful. Its architecture is incompatible with actual intelligence, and anyone who thinks otherwise is just fooling themselves. (It does make an alright autocomplete though).

Meanwhile a huge chunk of the software industry is now heavily using this "dead end" technology 👀

I work in a pretty massive tech company (think, the type that frequently acquires other smaller ones and absorbs them)

Everyone I know here is using it. A lot.

However my company also has tonnes of dedicated sessions and paid time to instruct it's employees on how to use it well, and to get good value out of it, abd the pitfalls it can have

So yeah turns out if you teach your employees how to use a tool, they start using it.

I'd say LLMs have made me about 3x as efficient or so at my job.

Imo our current version of ai are too generalized, we add so much information into the ai to make them good at everything it all mixes together into a single grey halucinating slop that the ai ends up being good at nothing.

We need to find ways to specialize ai and give said ai a more consistent and concrete personality to move forward.

There are some nice things I have done with AI tools, but I do have to wonder if the amount of money poured into it justifies the result.

The problem is that those companies are monopolies and can raise prices indefinitely to pursue this shitty dream because they got governments in their pockets. Because gov are cloud / microsoft software dependent - literally every country is on this planet - maybe except China / North Korea and Russia. They can like raise prices 10 times in next 10 years and don't give a fuck. Spend 1 trillion on AI and say we're near over and over again and literally nobody can stop them right now.

Pump and dump. That’s how the rich get richer.

It's because customers don't want it or care for it, it's only the corporations themselves are obsessed with it