> Typically that's solved with R&D happening in academia or semi-public space.

It was, but not any more. The crazy capital investments often come from private companies as well. And that's not a problem. They're spending their money rather than your money.

> They're spending their money rather than your money.

Ok, and from whom did their money come? They didn’t magic it into existence.

Google spending money that came from advertisers who got it from people buying their products isn’t much different than a research team spending public money, just with fewer middle men and a (generally) lower negative impact on society.

The money came from people selling (or trying to sell) products that other people want to buy for the price that includes the ads, as it's valuable. This is far better than the old school ad industry, which made advertising much more expensive and products more expensive.

$400 billion was spent on advertising alone in the US in 2023. In 1970, that number was $19 billion ($150 billion in 2023 dollars). Relative to GDP, that’s and 0.014% and 0.018% respectively. Depending on how you want to argue, there is either more spending on advertising (in dollars) than there was 50 years ago, or about the same (in relative GDP). Neither would account for products in the past being substantially more expensive than they are today.

That is ignoring the many moral arguments against targeted advertising, unnecessary consumerism, etc.

They could still spend the money in academia if they didn't the option to spend it themselves. When I was in college we had labs named after different companies, and those companies sponsored the research in those labs and had first dibs on commercialization.

so who are the academics raising 1b+ dollars to build and operate GPU farms to work on the frontiers of transformer models?

I'd raise you the same question on which private company is spawning its own LHC to research fundamental physics.

Firing particles at each other to disprove the next 500 variants of string theory is nowhere near as useful.

What a shortsighted comment