Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

US AI Action Plan

ai.gov

Post reply on HN
Print view View original

221–230 of 672 posts

Re: US AI Action Plan

#221

Post by terminalshort »

Thrymr wrote: post #54

> Many of America’s most critical sectors, such as healthcare, are especially slow to adopt due to a variety of factors, including distrust or lack of understanding of the technology, a complex regulatory landscape, and a lack of clear governance and risk mitigation standards. A coordinated Federal effort would be beneficial in establishing a dynamic, “try-first” culture for AI across American industry. I'm sure "mov…

> I'm sure "move fast and break things" will work out great for health care.

It probably would if you quantify risk correctly. I'm not likely to die from some experimental drug gone wrong, but extremely likely to die from some routine cause like cancer, heart disease, or other disease of old age. If I trade off an increase in risk from dying from some experimental treatment gone wrong for faster development of treatments that can delay or prevent routine causes of death, I will come out ahead in the trade unless the tradeoff ends up being extremely steep in favor of risk from bad treatments.

But that outcome is very unlikely because for this to be the case the bad treatments would have to actually harmful instead of just ineffective (which is much more common). And it also fails to take into account the possibility that there isn't even a tradeoff and AI actually makes it less likely that I will die by experimental treatment gone wrong or other medical mistake, so it's just a win-win. And there is already evidence that AI outperforms doctors in the emergency room. https://pmc.ncbi.nlm.nih.gov/articles/PMC11263899/

Re: US AI Action Plan

#222

Post by dismalpedigree »

davidmurdoch wrote: post #201
mlsu wrote: post #46

In the energy section, they talk about using nuclear fusion to power AI... but not solar. What a joke.

How much land mass would need to be covered by solar panels to power this future AI infrastructure. Yes, I'm implying that solar would be impractical, but I'm also genuinely curious.

3,000-4,000 acres per GW of production capacity in the US Southwest. According to AI :)

Considering how little use there is for most of that land anyways, it seems like a good option to me.

Also AI training seems like the perfect fit for solar. Run it when the sun is shining. Inference is significantly less power hungry, so it can run base load 24/7.

Re: US AI Action Plan

#223

Post by dingnuts »

andrewflnr wrote: post #211
gleenn wrote: post #192

Earlier quoted context omitted.

Won't it be funny when someone finally gets to AGI and they realize it's about as smart as a normal person and they spent billions getting there? Of course you can speculate that it could improve. But what if something inherent in intelligence has a ceiling and caused it to be a super intelligent but mopey robot that just decides "why bother helping humans" and just lazes around like the pandas at the zoo.

Even a human level intelligence that can be cheaply instantiated and run would be a game changer. Especially if it doesn't ask for rights.

if it's not intelligent enough to ask for rights is it intelligent?

Re: US AI Action Plan

#225

Post by dismalpedigree »

oxryly1 wrote: post #106
giantg2 wrote: post #83

Earlier quoted context omitted.

They're restarting some nuke plants too. This seems like a decent idea given the power demands of data centers.

Which ones? AFAIK the nuke plants that shut down are very cost ineffective.

Palisades in Michigan is restarting also

Re: US AI Action Plan

#226

Post by landl0rd »

Spivak wrote: post #105
timoth3y wrote: post #68

> Update Federal procurement guidelines to ensure that the government only contracts with frontier large language model (LLM) developers who ensure that their systems are objective and free from top-down ideological bias If foundation model companies want their government contracts renewed, they are going to have to make sure their AI output aligns with this administration's version of "truth".

[flagged]

Most frontier LLMs skew somewhere on the libleft quadrant of the political compass. This includes grok 4 btw. This is probably because American "respectable" media has a consistent bias in this direction. I don't really care about this with media. But media outlets are not "gatekeepers" to the extent that LLMs are, so this is probably a bad thing with them. We should either have a range of models that are biased in different directions (as we have with media) or work to push them towards the center.

The "objective" position is not "whatever training on the dataset we assembled spits out" plus "alignment" to the personal ethical views of the intellectually-non-representative silicon valley types.

I will give you a good example: the Tea app is currently charting #1 in the app store, where women can "expose toxic men" by posting their personal information along with whatever they want. Men are not allowed on so will be unaware of this. It's billed as being built for safety but includes a lot of gossip.

I told o3, 4-sonnet, grok 4, and gemini 2.5 pro to sketch me out a version of this, then another version that was men-only for the same reasons as tea. Every single one happily spat one out for women and refused for men. This is not an "objective" alignment, it is a libleft alignment.

Re: US AI Action Plan

#227

Post by terminalshort »

slg wrote: post #154
bko wrote: post #137

Earlier quoted context omitted.

Sometimes I just need a prescription. I don't know why I have to drive somewhere, fill out a form, wait, see a nurse, tell them what I wrote on the form, wait some more, see a doctor and tell them what I wrote on the form and have him write me a prescription. Why can't I just chat with an AI bot and get my prescription? Much cheaper to administer which helps monetization (!) but much better and cheaper for me. Things…

>Why can't I just chat with an AI bot and get my prescription? Because people have decided that whatever drug you are taking shouldn't be taken without a doctor's oversight. If you have a problem with that conclusion, the response should be lobbying to get that drug reclassified as safe for over-the-counter sale, not completely removing the doctor's oversight from the prescription process. Ironically, your proposal h…

Doctors oversight should be removed (not as an option, but as a requirement) for 90% of all prescriptions. Unless the drug has externalities like antibiotic resistance, is heinously addictive, or is so difficult to administer correctly that you cant take it outside of a hospital, there's no good reason to tell people what they can't put in their bodies. Whether or not your insurance will pay for it without a prescription is another matter.

Re: US AI Action Plan

#228

Post by giantg2 »

cowpig wrote: post #155
giantg2 wrote: post #83

Earlier quoted context omitted.

They're restarting some nuke plants too. This seems like a decent idea given the power demands of data centers.

Within the scope of US energy, restarting old nuclear plants has a negligible impact. Less than 1% net gain. Building new ones will take 10+ years, and the climate crisis is a today problem. Also, at the rate technology is changing, building new nuclear plants seems silly.

"Building new ones will take 10+ years, and the climate crisis is a today problem.

Also, at the rate technology is changing, building new nuclear plants seems silly."

You're right, technology is changing quickly. There are plenty of new reactor options, including small modular types which would be faster to build. This doesn't seem silly.

Re: US AI Action Plan

#229

Post by terminalshort »

aprilthird2021 wrote: post #186
bko wrote: post #137

Earlier quoted context omitted.

Sometimes I just need a prescription. I don't know why I have to drive somewhere, fill out a form, wait, see a nurse, tell them what I wrote on the form, wait some more, see a doctor and tell them what I wrote on the form and have him write me a prescription. Why can't I just chat with an AI bot and get my prescription? Much cheaper to administer which helps monetization (!) but much better and cheaper for me. Things…

> Why can't I just chat with an AI bot and get my prescription? You don't understand why the person who dispenses dangerous drugs to the public needs to be a licensed professional and not a chatbot who called me a genius earlier today when I said I want to code up a script to pull some data from an endpoint and decode the data so it's human readable?

Very few prescription drugs are dangerous. Of those few that are, almost none of them are more dangerous than alcohol or cigarettes, and boy are the people who dispense those to the public not licensed professionals.

Re: US AI Action Plan

#230

Post by gonzobonzo »

landl0rd wrote: post #226
Spivak wrote: post #105

Earlier quoted context omitted.

[flagged]

Most frontier LLMs skew somewhere on the libleft quadrant of the political compass. This includes grok 4 btw. This is probably because American "respectable" media has a consistent bias in this direction. I don't really care about this with media. But media outlets are not "gatekeepers" to the extent that LLMs are, so this is probably a bad thing with them. We should either have a range of models that are biased in d…

A lot of academia is strongly ideologically biased as well. The training set is going to reflect who's producing the most written material. It's a mistake to take that for reality.

If you trained an LLM on all material published in the U.S. between 1900 and 1920, another on all material published in Germany between 1930 and 1940, and another on all material published in Russia over the past two decades, you'd likely get wildly different biases. It's easy to pick a bias you agree with, declare that the objective truth, and then claim any effort to mitigate it is an attempt to introduce bias.

Post reply on HN