Comment by rayiner
5 hours ago
> Like do people think that these AI companies will create some superintelligence, suck up all the financial benefits of that, and then just decide to share it with the rest of us out of kindness?
I don’t get this. It’s like worrying in 1960 that the companies that invented computers will hoard the benefits of computing. It doesn’t make any sense. There is no secret formula to any of this. The math underlying the models is widely known, and there are tons of competitors, including foreign competitors.
> The math underlying the models is widely known, and there are tons of competitors, including foreign competitors.
Foreign militaries investing in autonomous warfare does not assuage my concerns about my country investing in autonomous warfare.
Also, have you been paying attention to median wages vs median CEO wages since the 1960s? The benefits of computing really have gone to the captains of industry.
Yet, that 80s CEO, practically a peasant by modern standards of pay, had the option to cross the Atlantic in a Mach 2 supersonic airliner. If that's not the most obvious demonstration of technological prowess, I don't know what is. In contrast, the F-35, believed to be the world's most advanced fighter jet today, has a top speed of Mach 1.6
Today's CEOs get a gold-plated sky bus.
The Concorde was cramped. JFK to Heathrow took five hours (3 in the air, an hour in each airport) plus the time for first and last miles. JFK to Heathrow today is 7 to 8 hours with the same provisos.
If you're going to spend a quarter of the day travelling, why not spend a third of the day and do it more comfortably?
The F-35 isn't the fastest jet; fighter jets aren't business jets; apples are not rutabagas.
> Also, have you been paying attention to median wages vs median CEO wages since the 1960s
CEO pay isn’t a good proxy for “captains of industry.” What you want to look at is the labor versus capital share of income. That’s been very stable since the 1960s: https://taxfoundation.org/blog/labor-share-net-income-within....
Clearly that's not a good proxy either then... even the article itself says:
> Ultimately, concerns over inequality should focus on differences within labor compensation rather than the split between labor and capital.
The main issues I can see are:
- This "labor share" includes multi-million dollar executive pay packages, so it's heavily skewed towards the 1%
- It also completely ignores unrealized capital gains and loans against them, which is how the ultra-wealthy actually fund their lifestyles tax-free, with a tiny income on paper
1 reply →
Tech isn't special. The value of pretty much all productivity improvements since ~1979 have been captured either via the stock market (97% owned by the top 10% in wealth) or land rents.
If tech isn't special, why are nine of the ten richest Americans in tech?
You could say that about any big tech product and yet we've all seen power and wealth become concentrated on an incredible scale since the 60s. The immense resources needed to train frontier models gives the few companies that can manage it more of a moat than most tech products. So empirically I expect that we'll see the current wealth hording keep going at at least the same rate.
>It’s like worrying in 1960 that the companies that invented computers will hoard the benefits of computing.
Isn't that what happened? There was enough competition among computing companies that they weren't able to completely monopolize all the productivity improvements, but the financial benefits were mostly captured by the capital class in one way or another.[1]
TVs might be cheaper today and we all like watching Netflix, but I'm skeptical of the idea that the financial wellbeing of the average American has been improved by computers.
[1] - https://www.cbpp.org/research/poverty-and-inequality/a-guide...
>I'm skeptical of the idea that the financial wellbeing of the average American has been improved by computers.
Really? Even with all the new avenues of education and communication that they opened up? You think the positives and negatives balance out to close to zero?
The stats I linked above show that those “new avenues of education and communication” didn’t actually benefit the average American financially in at least the relative sense. Do you have something to counter that evidence?
5 replies →
>It’s like worrying in 1960 that the companies that invented computers will hoard the benefits of computing. It doesn’t make any sense.
Not only is that exactly what happened, they weren't satisfied with accumulating most of the wealth produced by it, they've also taken it upon themselves to take over democracy and media and act like a state within a state. You only need one statistic to understand America today, that working class Americans without college degrees have had their purchasing power stagnate since the 1970s(https://www.epi.org/publication/charting-wage-stagnation/)
People who used to have good jobs can now drive for Doordash and what the last wave of digitalization did over 30 years the AI gurus now promise to do in 10 again, and not just to the working class. The only reason to be optimistic is that they're snake oil salesmen.
> You only need one statistic to understand America today, that working class Americans without college degrees have had their purchasing power stagnate since the 1970s(https://www.epi.org/publication/charting-wage-stagnation/)
Having a college degree was a much more selective measure in 1970 (about 10% of the population) than today (almost 40%). Consider a large law firm. In 1970, the only 4-year degree holders would have been the attorneys. A lot of work would have been done by well-compensated paralegals with high school diplomas. Today, everyone down to the receptionist will have a degree and the non-degree holders would basically be the maintenance and cafeteria staff.
So using your math, the income of non-college workers at the law firm could be stagnant from 1970 to today, even if the income of each specific position had increased significantly.
Look at the macro data. The overall median wage for all American workers combined has also drastically lagged behind economic productivity since the 70s.
The 60s were 60 years ago. There were regulations back then that protected the common folk and capitalism wasn't as ruthless and aligned with the will of the 0.1% as it is now.
I want to be optimistic and agree with you, but I don't think the parallels are as strong as you say they are. We already have Anthropic withhold Mythos from the public, the governement now allowing the use of Fable, I don't think its farfetched to think that the US will start regulating access to Chinese/open-source models, pricing for compute isn't slowing down. The problem isn't AI, but who controls the compute that powers it.
What regulations do you feel protected common folk from capitalism during the development of computing, and how have they weakened?
I was also struck by this. We are a far more regulated society today than in the 1960s. The total body of codified federal law, statutory plus regulatory, grew from roughly 78,000 pages in 1960 to roughly 246,000 pages by 2020. The content matters of course, but I’m skeptical that under-regulation is the problem.
1 reply →
Antitrust enforcement.
> The 60s were 60 years ago. There were regulations back then that protected the common folk and capitalism wasn't as ruthless and aligned with the will of the 0.1% as it is now.
No company in the 1990s ever achieved as much market power as IBM did in the 1960s-1980s. IBM controlled 70% of the computer market at the time.
Can you give me a concrete example that’s relevant to the deployment of computers?
The computer market was also much smaller and society much more independent from it at the time.
> IBM controlled 70% of the computer market at the time.
The computer market was microscopic at the time.
> Can you give me a concrete example that’s relevant to the deployment of computers?
Nvidia. 92% of domestic GPU market, 85% of AI datacenter market.
The AI investment bubble is exploding the cost of everything that contains RAM or SSD storage. This is having and will have detrimental effects on the global economy, as supply shortages enormously increase the cost of consumer electronics, cars, anything which requires memory.
Not because AI is making so much money, far from it - it's a money sink. Rather, it's because the global supply of money got so large and so unequal that it ran out of vehicles which would provide sufficient returns for the indolent investment class. So they jumped into speculation - first crypto, now AI and the tech that enables it.
To the benefit of whom? Mass layoffs, tech industry consolidation, new products being cancelled left and right.
Ironically there's no equivalent in the modern era. This is peak social inequality breaking the economy for everyone outside the .001%. And it didn't happen because of a shortage of a raw material, or physics breaking Moore's law. It happened because big tech centralised power into a handful of gigacorporations - creating a handful of ultra-billionaires - who are enormously incentivised to centralise technology itself.
I mean. From a financial PoV that's exactly what happened
Did they, though? Intel, IBM, DEC and Honeywell were big names, but they're not FAANG.
Other than Nintendo, how many games companies today were around back then?
... so your argument is "no actually the second generation did that"?
End result for 100-16 % of Americans is the same
1 reply →