> Are you thinking google would be willing to risk their ad revenue for a cool new tech?
University and government research is also super important. Google publishes papers not code, whereas universities often publish their code and that's doing more to move AI forward than Google.
But Google is willing to make large research bets too.
Yeah, I'm definitely not saying Google isn't doing important work. I just think this highlights the key role government funding plays in ensuring that businesses don't have to pay for all of that basic research and training.
That's basically the answer to that question: Google isn't risking their core business on this because so much of the foundation has been set. If, say, in 2003 they'd gone to investors and asked for money to fund AI research it'd have been seen as too risky, especially since the collapse of the 1980s AI bubble was still well within living memory. Once NSF, DARPA, et al. had continued to fund the scientists, however, things matured to the point where it has a more acceptable level of risk to get in on a huge field.
“derisked” is [willful?] misrepresentation. This isn't a hard argument: board early research requires a lot of speculative work which doesn't pan out for everything which does, not to mention tons of basic training and equipment investment. Companies have a hard time justifying that and, unsurprisingly, the record shows they largely don't. Government funding operates with different incentives and timescales and has a lengthy track record of success getting ideas to the point where companies are willing to foot the remaining development costs.
> Are you thinking google would be willing to risk their ad revenue for a cool new tech?
University and government research is also super important. Google publishes papers not code, whereas universities often publish their code and that's doing more to move AI forward than Google.
But Google is willing to make large research bets too.