US senators' AI roadmap aims for $32B in R&D spending

A quartet of US senators have released an AI legislation roadmap that calls for billions of dollars in research funding, but largely kicks the can down the road on determining federal AI legislation. The Driving US Innovation in Artificial Intelligence proposal [PDF] calls for as much as $32 billion per year to be spent on non …

[DS999]

The research funding

Should have strings attached - the government retains ownership of the research. If you work for a company and they fund your research, they own the patents and other IP that result from it. It should be the same way for government research - not fund people who then turn around and create startups based on research they did on the taxpayer's dime.

If the IP is useful the government can sell it to the highest bidder, or become a non-voting equity participant in startups using it, so taxpayer funds aren't making VCs and private equity funds even richer.

[Anonymous Coward]

Re: The research funding

Closer, but we could tie a pretty bow on those strings.

WE are paying for it. We should own it. Not let the government spend and probably waste 30+ Billion and then sell the results for pennies on the dollar to private companies, then probably waste the pennies to boot. The IP from these sorts of R&D projects should be held in trust for things like the social security fund. Make it a literal investment on our future. Payment from industry accepted in cash AND stock.

[Rafael #872397]

Driving US Innovation in Artificial Intelligence

What, no fancy backronym? Surely written by AI!

SPAINK: Spending Pork by Advancing Innovation in New Knowledge

I'm sure we meatbags can do better!

[vtcodger]

A bit much

$32B? Seems a bit high to me. Especially for a "technology" that looks to be an enormous resource drain and very likely a major PITA for users while the bugs are worked out of the initial half baked implementations. Based on the difficulty in developing "safe" autonomous cars (15 years of R&D and still just one or two fixes -- if you believe Elon -- from being ready for prime time), my guess is that AI that any sane person would want to inflict on society, is probably 3 to 5 DECADES away, not 3-5 hours/days/months as the con-artist caucus would have us believe. And very likely once we know what we're doing it will turn out to have quite limited utility and to not be all that resource intensive.

So, give a modest amount of money to DARPA for AI research. Overall, DARPA seems to do a pretty good job of managing R&D. And they're defense oriented. if AI actually has military applications they're likely to do a better job of assessing risks and benefits than folks with products to sell. Let's see what they come up with.

But ,,, but ...but .. we might lose the AI race and Sam Altman et. al. might have to find honest employment? Yep. Could happen. I think I could live with that