I’ve wondered for a while if anyone is working on very wide channel parallel (kind of like RAID 0) SSD for this purpose. Couple that with a tensor processor and that would be interesting.
A point I don’t often see made is: the argument that “the parents need to parent” is unintentionally classist.
Wealthier people who can afford to have one parent stay home or have babysitters or nannies who will actually supervise the kid can do a much better job monitoring and redirecting than when both parents work long hours and have no money.
So poorer kids are going to spend more time scrolling brain rot, which may lead to worse academic performance and concentration problems. They’re also more likely to get sucked into any number of garbage social media propagated cults, wacko ideologies, and dumb fads.
It also ignores social pressure entirely. If everyone in your class is on some social network and your parents parent and you aren't then you are socially outcast from a major part of your peer socialisation.
If only 50% of parents enforce the rules suddenly half your class isn't there and it doesn't become such a big deal to be missing out and it's less appealing for the kids who are allowed still.
I'd prefer it to be voluntarily organised like https://www.waituntil8th.org/ but even a bad solution is going to have a major improvement on society even if social media is less private due to the ID issue. It's not like you don't share everything with these companies anyhow. I'm pretty confident that if you are a regular user of any large social network today they could identify you 100% of the time already even without your ID.
I'll take that over everyone being forced to accept the same, standardized, government-mandated "cults, wacko ideologies, and dumb fads", thanks!
Government makes everyone poorer through their terrible policies -> parents need to work more to stay above water -> nanny state must grow to take care of kids -> terrible government policies entrench. Nope, don't want it
I realized a while back that Elon Musk isn't Iron Man. His superhero (or supervillain depending on your view) persona is ZIRP Man, the master of riding successive credit expansion and speculative waves. It's sort of ironic that he at least pays lip service to some Austrian-style quasi-libertarian economic ideas, because the Federal Reserve created him.
Now he's surfing the AI wave. We are no longer technically in ZIRP but the delayed inflationary wave is now traveling through the economy and pumping everything. He knows the best way to soak up cheap money right now is slap AI on it.
I also had the thought the other day that him hitting $1T technical net worth might actually be a harbinger of a lot more future inflation. Inflation of this type hits assets before it hits things like prices and wages, and it hits assets with fast market cycles like stocks before it hits things like Real Estate. The blast wave starts at the top and moves down and out. So maybe Elon hitting $1T really means that in 20 years that'll be more like $100B inflation adjusted. Meanwhile a loaf of bread will be $20 and a starter home $4M.
But the fact that monetary inflation starts top-down is why low interest rates exacerbate inequality. The very richest and most leveraged can use the arbitrage gap to buy everything else before the inflation wave propagates. We've been in a low interest rate environment for about two decades, and you can see during that time how the super-rich with access to cheap money have fully detached from the rest of the economy.
In other words: the reaction to the 2008 financial crisis was to inject huge liquidity at the top, which created the new Gilded Age.
Ultimately it may be somewhat intentional. One way out of a sovereign debt crisis when you also have a sovereign currency is to inflate your way out, which basically is a huge tax on every non-domestic entity that owns your debt.
To be honest and objective, I think he at least knows enough about engineering to hire people who know what they're doing, which is how he got here. It's not all chicanery.
Over time, though, I think he's drifted away from his original "make real things in the real world" focus and more toward "play money games" and "play political games."
It's sad. One common comic book supervillain arc is to start as a hero and become what you despise.
Nerds are no better or worse than other people. Dump massive money and power into an industry and you will get unhinged shit like this that bubbles to the surface.
Some of it is the mask falling off and some of it is people genuinely getting warped by it. It’s a little of both.
In finance it’s covered over by a buttoned down ivy league veneer, but the coke snorting maniac is there.
Same in politics where there’s pomp and ceremony to cover it, but when it comes out in the open there it’s probably the most ugly. Governments have armies and police.
In nerd-dom it comes in a form that’s uniquely tone deaf to the point of coming off like a comic book or anime villain.
Programming has already become this way. Opinions about different languages and architectures are taste, or sometimes even just vibes. Few try to actually ask “can I quantify whether microservices or monoliths are better in terms of either maintainability or scaling?”
A lot of this is a result of systems having long ago exceeded the complexity threshold of things people can hold in their heads. There are too many layers, subsystems, languages, APIs, all glued together. Attempts at radical simplification fail because each of those layers and subsystems has features or behaviors someone needs, and a lot of it isn’t even documented.
AI takes this to the extreme. I’ve already learned that certain models have “personalities.” Some are more likely to go with you on magical journeys into hallucination while others are more critical. Some are better at detail while others seem better at abstraction but fall over on detail. Some are better instruction followers. All their quirks are complex and the systems themselves are impossible to understand.
Computer systems are becoming organic, biological.
"Feeping creaturism" has always been a problem, for sure.
But those technologies are layers, and there are reliable things that sometimes bubble across the boundaries — type hints, better code patterns to trigger compiler optimisation, interesting tricks with key column selection — and someone with expertise from that layer below can explain why, and their advice will always work in situations that are sufficiently similar.
You are right about AI personalities. Obvious even with the open weights models. Gemma and Qwen write code and documentation like people from different cultures. Because I guess they are a bit like that.
They're almost literally "from different cultures" - because of how post-training does things.
All "personality traits" within an LLM are entangled. So when you mid-train or post-train on ESL texts, or run RLHF using people from a given culture, you risk bleeding some of the related cultural traits into the LLM itself. A lot of the resulting "personality" is downstream from different AI teams picking different datasets and training signals.
RLAF is more of a "funhouse mirror" distortion - it takes existing traits and twists them, sometimes amplifies them to comical extremes. Weird can become weirder. A verbal tic can become a style signature. Part of the reason why AI writing from GPT-4 era and to now has changed so dramatically.
Seems to be playing out.
reply