Why the US Needs a Moonshot Mentality for AI - Led by the Public Sector (wsj.com) 76
Fei-Fei Li and John Etchemendy, the founding co-directors of the Stanford Institute for Human-Centered Artificial Intelligence, in an op-ed on WSJ argue that AI is too important to be left entirely in the hands of the big tech companies: Among other things, 2023 will be remembered as the year artificial intelligence went mainstream. But while Americans from every corner of the country began dabbling with tools like ChatGPT and Midjourney, we believe 2023 is also the year Congress failed to act on what we see as the big picture: AI's impact will be far bigger than the products that companies are releasing at a breakneck pace. AI is a broad, general-purpose technology with profound implications for society that cannot be overstated.
[...] So what needs to happen? President Biden has set the stage, and with all this attention, it's time for Congress to act. They need to pass the Create AI Act, adhere to the elements called on by the new executive order, and invest more in the public sector to ensure America's leadership in creating AI technology steeped in the values we stand for. We also encourage an investment in human capital to bring more talent to the U.S. to work in the field of AI within academia and the government.
But why does this matter? Because this technology isn't just good for optimizing ad revenue for technology companies, but can fuel the next generation of scientific discovery, ranging from nuclear fusion to curing cancer. Furthermore, to truly understand this technology, including its sometimes unpredictable emergent capabilities and behaviors, public-sector researchers urgently need to replicate and examine the under-the-hood architecture of these models. That's why government research labs need to take a larger role in AI. [...]
[...] So what needs to happen? President Biden has set the stage, and with all this attention, it's time for Congress to act. They need to pass the Create AI Act, adhere to the elements called on by the new executive order, and invest more in the public sector to ensure America's leadership in creating AI technology steeped in the values we stand for. We also encourage an investment in human capital to bring more talent to the U.S. to work in the field of AI within academia and the government.
But why does this matter? Because this technology isn't just good for optimizing ad revenue for technology companies, but can fuel the next generation of scientific discovery, ranging from nuclear fusion to curing cancer. Furthermore, to truly understand this technology, including its sometimes unpredictable emergent capabilities and behaviors, public-sector researchers urgently need to replicate and examine the under-the-hood architecture of these models. That's why government research labs need to take a larger role in AI. [...]
The "public" sector? (Score:4, Insightful)
Does this mean government agencies who care more about providing jobs in politically important areas? Or bureaucrats who want a guaranteed high income while impeding progress? How about a free and open source option?
No (Score:1)
Bureaucrats are government employees who's job it is to implement government policy, not to conduct research. A bureaucrat is the guy who keeps your water clean, not the guy who researches how to make clean water.
Re: (Score:2, Troll)
Right, the ones appointed by the racist (see "Nixon's Southern Strategy) former Grand Oligarchic Party, right?
Re:The gay NASA route (Score:1)
Re:No (Score:5, Informative)
So, you are completely ignorant of federal employees who do research? I have shocking news for you: you should, perhaps, look up Fermilab, and Argonne, and the NIH.
Re: (Score:2)
> https://www.youtube.com/watch [youtube.com]?... [youtube.com]
tl;dw, might later
The intro to that video is long and seems to end around here - https://youtu.be/OoJsPvmFixU?t... [youtu.be]
Re: (Score:2, Troll)
Bureaucrats are worthless. The only thing they do is make busy work for other people to justify their own positions. Eventually they do this by hiring other bureaucrats to work under them so they have someone to manage. Most things would get done far more quickly without that pack of obstructioni
Bureaucrats are not useless (Score:2)
The objective of having bureaucrats is to permit strictly defined activities that are according to current policy to occur, and to require the intervention of very high ranking government officials to enable anything not perfectly aligned with current policy to occur.
If this sounds like gumming up the works, it's because it is. However, it is the true expression of government power.
Re:The "public" sector? (Score:5, Informative)
Well, it means that the entire coding team and top level training priorities will be DEI, and pronouns.
Actually, that's probably about it...actual coding and having something work using ingenuity, innovation by merit based hiring of teams is pretty much secondary or tertiary in terms of importance with the government push into AI.
Basically just another money boondoggle, waste and the program going downhill as other countries with less "values" actually use their best people to do actual work to actually try to make a superior system.
That just seems to be the norm these days in the US, sadly.
Damn...we really did used to be good at all this leadership in tech. But now, we're ultimately consumed by pondering racism in mathematics, and transphobia in everything else.
Re: (Score:2)
Oh, for Pete's sake. The US government couldn't even write up the REQUIREMENTS for this task. It took the feds longer to do an environmental impact study and issue Spacex a launch license than it took the company to build the rocket itself. Where have peopke been? Are their eyes comp!etely shut? The feds are incapable of building anything.
Re: (Score:2)
*Three* paywalled WSJ articles *in a row*? (Score:4, Informative)
Come on, /. We can't read any of these. Who's greenlighting this rampant advertising for WSJ?
Re:*Three* paywalled WSJ articles *in a row*? (Score:4, Insightful)
Wait... since when do we read the linked articles?
Hell, reading more than the headline is already considered heresy here.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Another money grab.... pfft.... (Score:4, Insightful)
If AI proves useful for such endeavors as studying nuclear fusion or curing cancer, you can be sure the entities involved will put in budget requests for it.
Right now? Nobody's even sure how any of this will pan out. We know you can use a ChatGPT type application to write pretty usable ad copy for your cookie cutter product listings online. But we also know that when you start asking it general questions, it's prone to giving you an educated- sounding detailed response that's completely incorrect. MidJourney and the like are getting repeatedly slammed by the artistic community everywhere you turn. Nobody's even sure it's fair to call it "art" when all you did was craft a detailed description of what you'd like to see -- and it renders a result based on other human artist's previous works?
I see no need for any government legislation, right now. We've got FAR too much of that and far too big a Federal deficit that they keep neglecting.
Re: (Score:1)
> If AI proves useful for such endeavors as studying nuclear fusion or curing cancer, you can be sure the entities involved will put in budget requests for it.
The gov't would probably rent existing systems to do such work; they probably wouldn't and couldn't start from scratch.
Re: (Score:2)
I've never understood that belief. So-called government waste is almost exclusively on the private side of public private partnerships. On their own, the government is incredibly efficient, the post office being an excellent example. Oh, and all while dealing with active sabotage from the right. Just ask NASA.
I have little doubt that a public AI research group could advance the state-of-the-art far more efficiently than the private sector, even starting from scratch.
Re: (Score:1)
> On their own, the government is incredibly efficient, the post office being an excellent example.
But their processes are relatively stable. Rocket tech changes too quickly.
Re: (Score:2)
I'm pretty sure NASA knows a thing or two about rockets and can easily attract top-talent. Maybe we should try funding them better instead of dumping tax dollars into the black hole that is private aerospace.
No. Gov't will bungle (Score:3)
I'm not an all-govt-is-bad kind of person, but *this* is not the kind of thing gov't does well. Apollo innovation was driven mostly by private contractors, and they are not going to give their best ideas to gov't without a pretty penny, meaning big tax, which won't fly in the US. At least walking the moon was a clear and likely achievable goal. AI's uses and abilities are too damned fuzzy right now to motivate tax payers to fork enough over to make a difference.
Re:No. Gov't will bungle (Score:4, Informative)
Wrong answer. Stop listening to Faux Noise, and read some history. There were federal employees doing a lot of the research.
Many of whom were outsourced by Raygun and the Grand Oligarchic Party, so their buddies could make big bucks, rather than pay std. GS salaries to employees.
overstated (Score:3, Insightful)
Press: Hold my beer
Let's stick to what we do now (Score:3)
National security concerns should be handled confidentially by the defence department.
Calls for public sector regulation and investment are often a way for special interests to get the taxpayer to pay for their advantage.
Re: (Score:2, Troll)
Bull. "Market forces" are why there's been inflation - "greedflation" (go look it up).
Re: (Score:3)
Greedflation (Score:2)
is caused by two things:
1) Consumers being willing to pay more for the items that they trust when the price rises unreasonably
2) The failure of the government to prevent monopolistic behaviour
So you appear to be supporting more activity by the public sector which has failed to protect the public in the past. Does that make sense?
Waves hand (Score:2)
The government will make it all better.
Hell of a plan. Hardly better than the letting the invisible hand of the free market decide everything based on short term gains.
But why would the public sector? (Score:5, Insightful)
Let's face it, the Moonshot was an endeavor that employed millions one way or another, created jobs and ensured prosperity for almost everyone in America, one way or another.
AI will eliminated jobs and enrich a few at the expense of the rest of the public.
So why exactly would the public sector not do everything it possibly could to kill off AI?
Re: (Score:1)
Re: (Score:2)
The moonshot did create quite a few jobs, direct or indirect. The US had to invent a whole lot of things, not just in engineering but also in management and logistics. The money spent here in the 60s caused a boost to the US' economy that carried it into the 80s and partly even into the 90s as the leader of innovation and productivity.
Also, the government built very little of the actual moon program. Most contracts fell on big companies like Grumman, Rockwell, Martin or Boeing. Who in turn needed engineers
Re: (Score:2)
The moonshot did create quite a few jobs, direct or indirect. The US had to invent a whole lot of things, not just in engineering but also in management and logistics. The money spent here in the 60s caused a boost to the US' economy that carried it into the 80s and partly even into the 90s as the leader of innovation and productivity.
That's the seen. You have to consider the unseen. How many jobs didn't get created because people were paying taxes to fund Apollo instead of buying products from Ford, 3M, and Walmart? How many companies didn't get started because a ton of capital was tied up building Saturn V engines?
We'll never actually know because we can't run the experiment where we hold everything else constant. However, we can be quite certain the answers aren't zero and zero.
Re: (Score:2)
Well, we can extrapolate from what happened afterwards. You know, the time when Japan took over the lead in engineering and all manufacturing went to China.
Re: (Score:2)
Well, we can extrapolate from what happened afterwards. You know, the time when Japan took over the lead in engineering and all manufacturing went to China.
To make sure I understand, your assertion is we stopped spending on Apollo around 1972 and shortly afterwards, Japan emerged as an engineering leader and China as a manufacturing center? Interesting theory, bearing in mind that correlation does not prove causation. I have a few objections.
First, spending on Apollo might be what caused Japan to develop more engineering talent relative to the US. We spent out money on rocket engines, they spent it on car engines. Car engines turned out to be a more valuable p
Re: (Score:2)
The US had an edge in manufacturing, engineering, management, logistics, you name it. All of that and more due to what the moon program developed. These people went into the industry, and they multiplied that by teaching others. The problem is that we rested on those laurels. The 70s were an incredibly productive time for the US. Nobody on the planet could hold a candle to US logistics, production or engineering, in efficiency or bleeding edge technology. If you asked anyone in the 70s and early 80s where t
Re: (Score:1)
Sigh. The Saturn V required a full three percent of the U.S. GDP to develop. Two entire generations of Americans were raised by the paychecks the space program provided, and that's just the direct investment.
We'll leave aside the aerospace, IT, research, engineering, energy, telecom and materials science advances. Then there's the telescopes and SpaceX but I've made my point.
Re:But why would the public sector? (Score:5, Interesting)
Let's face it, the Moonshot was an endeavor that employed millions one way or another, created jobs and ensured prosperity for almost everyone in America, one way or another.
AI will eliminated jobs and enrich a few at the expense of the rest of the public.
So why exactly would the public sector not do everything it possibly could to kill off AI?
The Moonshot was also something that gave the common man, at the ground floor, as it were, a goal that we could look at and say we were all a part of it. When AI comes along and sweeps the common man aside the name of increasing wealth it the hands of the few who control it, where's that moment going to be for us? It simply won't happen.
You can't get the backing of the people for something that's only going to benefit a very, very few. If AI were promoted as something more than an power/wealth aggregator for the already have plenty crowd? Maybe. But right now I'm not seeing the sell for the average person. "Help us make something smarter, so we can take more from you" isn't a resounding message for the middle class and down. Hell, at the moment it's not even a resounding message for the sorta-wealthy crowd that inhabit that space between the middle class and the uber-wealthy.
Re: (Score:3)
The Moonshot was also something that gave the common man, at the ground floor, as it were, a goal that we could look at and say we were all a part of it. When AI comes along and sweeps the common man aside the name of increasing wealth it the hands of the few who control it, where's that moment going to be for us? It simply won't happen.
You can't get the backing of the people for something that's only going to benefit a very, very few.
It is possible that technology can cut both ways. That some future good enough AI could lead to dramatic reductions in capital costs required to achieve that which was previously the province of large corporations.
Re: (Score:2)
There's that anecdote of Kennedy visiting a rocket assembly hangar and he saw a guy sweeping the floor, and he went over and asked him what he did there. The man stood to attention and said "Mr. President, I'm helping to put a man on the moon!"
True or not, it was pretty much the sentiment that people could assemble behind. Everyone could tell himself that they did something to make this happen. The farmer that delivered the grain the astronauts might eat, the factory worker that assembled the truck that wou
Re: (Score:2)
Also, no need for any governments to add money to an already massively overheated situation. It's going to be a big pop when it bursts.
A moonshot needs a moon (Score:5, Insightful)
You know why the Apollo program worked? Because it had a clear goal. Put boots on the moon before the Russians. Yeah, it was a stupid nationalistic goal, but it was a clear target with a defined endpoint and some pretty understandable milestones. So simple even a politician can understand.
What's the concrete goal for a "moonshot" AI program? What's the endpoint? What steps do we take to get there? How do we know when we're successful?
The actual moonshot was a decade's worth of crunch time. It only worked because everyone involved knew the goal and could see actual progress being made towards it. You can't sustain that kind of energy and support without a goal and clear progress. "AI" is just too nebulous.
Re: A moonshot needs a moon (Score:3)
Then lets define our AI goals here and now. America should develop an AI that puts everyone out of a job, and makes the dream of unrestrained spying on the citizenry possible.
Re: (Score:2)
Re:A moonshot needs a moon (Score:4, Interesting)
Close -
The nationalistic part was NOT stupid though. It is what enabled the clarity or at least the preservation of the clarity of the primary goal. There a was uniting organizing principle that everyone supported.
Try that with most things to day - "All electric cars by 2030" we could probably 'do that' but a lot of people don't really share a sense of purpose behind it. You'll have some people who 'want to save the planet', some others that 'want to beat China's GDP', 'want to create jobs', 'want to outsource jobs' ....
So every top line has all these secondary objectives that are actually 'primary' objectives for a lot of supporters. They are often competing objectives. So we end up with policy that gets us a network of charging stations with connectors nobody wants anymore that won't even begin to be built for some time.
We need the the 'nationalism' or some other animating component or we will continue to fail due to in fighting. If you tried the Apollo program now, nobody would have agreed to Cape Canaveral, the fact it is the most logical place to shoot things into space from would take a back seat to using the program as vehicle for some urban renewal project in the rust belt or plains states jobs program or whatever else...
TL:DR we can't do big things because not enough people actually care about the big things. In most areas the pain of or fear of failure usually put a fine point on these types of problems, but Government is large and accountability is diffuse so there is no pain. So the failure just grows and odds are the persons most responsible for the failures don't even see them as failures; the money got spent in their district after all or enough minority owned business got contracts, or funds we reallocated away from some activity they oppose.. they got what they were personally after. .
Re: (Score:3, Funny)
Hannity, git off slashdot!
Re: (Score:2)
moonshot (Score:5, Insightful)
The moonshot that was devised 60 years ago was a one-of-a-kind event that has not been repeated for *anything* since. The US is not remotely the country it was in 1963, and this "moonshot" metaphor really needs to be put to rest once and for all.
Re:moonshot (Score:4, Insightful)
The moonshot that was devised 60 years ago was a one-of-a-kind event that has not been repeated for *anything* since. The US is not remotely the country it was in 1963, and this "moonshot" metaphor really needs to be put to rest once and for all.
I like the concept of the Moonshot endeavor. The problem is, in today's America, you'll never see the entire country get behind a singular goal like that again. At least not in the foreseeable future. We're literally at a point where someone can say grass is green, and based on the politics of those involved, at least 1/3rd of the country would absolutely, vehemently deny any greenness is contained in grass. How do you get the Moonshot mentality into the public with that type of back-biting happening? You can't.
Granted, this is very much a different country than we were back then. Now all that needs to happen is you get a few multi-millionaires to agree on something, and they'll just spout nonsense about it while doing it, regardless of whether it's a good idea or not. if it fails, and humbles the rich who funded it, the government will hand them a few more dollars for their trouble and they'll wander off to find the next bumble-point. That's the modern Moonshot. And likely where the AI train is headed at some point.
Re: (Score:2)
The AI Dilemma (Score:3)
Here's a pretty compelling presentation and call for social action re: AI: https://www.youtube.com/watch?v=cB0_-qKbal4 [youtube.com]
They make a lot of surprisingly reasonable comparisons of AI to nukes, along with a pithy justification that AI is potentially worse than nukes, because nukes don't make better nukes. And like with nukes, they think effective global regulation is going to have to start with grassroots concern and activism. Can't trust companies to do it, and government is not equipped in knowledge or pacing to regulate it effectively.
They justify erring on the side of dramatic alarmism by citing some of the significant, potential consequences of not being cautious, and of our long (and recent) history of not being cautious (or even aware that we should be.)
AI (like nukes) is a plausible explanation for the Fermi Paradox. It may not be enough to rely on our paleolithic brains and medieval institutions to protect us from it.
Re: (Score:2)
Here's a pretty compelling presentation and call for social action re: AI:
They begin the presentation by saying it is a product of outreach by industry. They end it talking about how open source is a problem and EU AI act the answer. In between when they are not shitting on social media (not that it isn't deserving) they incorrectly summarize numerous papers and leave out important context.
AI (like nukes) is a plausible explanation for the Fermi Paradox.
If an alien made out of silicon diodes landed on earth would that not count? What if they held up a huge intergalactic billboard that reads aliens live here?
It may not be enough to rely on our paleolithic brains and medieval institutions to protect us from it.
Thank goodness for the EU AI Act.
Turkeys voting for Christmas (Score:2)
AI moonshot? (Score:3)
We knew what it would take to get us to the moon.
We don't know what it will take to reach any AI milestone that is not essentially already achieved.
So you can't have a moonshot. It's in the basic research category, not just development needed to finish.
Moonshot? More like Manhattan Project (Score:1)
Monumentally stupid idea (Score:2)
The "public sector" lost interest in going to the moon once we got there. It took the private sector taking up the reins to get us back to where we should have been. The public sector is subject to political whims and they stop taking risks pretty quickly if some pollster decides that something isn't worth spending money on.
Re: (Score:2)
That's simply not true.
Congress Has to Act! (Score:1)
The only problem is Congress has no constitutional authority to regulate technology.
This "Congress must act" mentality, which invites government into every alcove and cupboard, is far more concerning than the hand-wringing fiction about AI.
Meanwhile the idea that only government can be "trusted" with the big decisions needs to be rebuked once and for all. Government creates nothing but obstacles.
While we're at it, invoking the word "moonshot" is about as cynical as it gets. The average fan of top-down gover
This is such an idiotic idea (Score:2)
1) The US owns AI on the corporate sector. Its already a gold rush where anyone with any AI relevant talent has already been wooed by every cutting edge company on earth.
2) What does the US taxpayer get when they throw billions of taxpayer dollars into academic AI research? Well, the idealists and mediocrities in the academic sector can get public funding.
3) There is already a worldwide race for AI. Its about the "power" of capitalizing on AI advancements. What does publicly funding AI research do?
Re: (Score:2)