Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
AI Government

What Happened When California's State Government Examined the Risks and Benefits of AI? (msn.com) 80

An anonymous reader shared this report from the Los Angeles Times: AI that can generate text, images and other content could help improve state programs but also poses risks, according to a report released by the governor's office on Tuesday. Generative AI could help quickly translate government materials into multiple languages, analyze tax claims to detect fraud, summarize public comments and answer questions about state services. Still, deploying the technology, the analysis warned, also comes with concerns around data privacy, misinformation, equity and bias. "When used ethically and transparently, GenAI has the potential to dramatically improve service delivery outcomes and increase access to and utilization of government programs," the report stated...

AI advancements could benefit California's economy. The state is home to 35 of the world's 50 top AI companies and data from Pitchfork says the GenAI market could reach $42.6 billion in 2023, the report said. Some of the risks outlined in the report include spreading false information, giving consumers dangerous medical advice and enabling the creation of harmful chemicals and nuclear weapons. Data breaches, privacy and bias are also top concerns along with whether AI will take away jobs. "Given these risks, the use of GenAI technology should always be evaluated to determine if this tool is necessary and beneficial to solve a problem compared to the status quo," the report said.

This discussion has been archived. No new comments can be posted.

What Happened When California's State Government Examined the Risks and Benefits of AI?

Comments Filter:
  • So, it can do wonderful things and it can do horrible things. But nobody knows the actual future mix.

    • by gweihir ( 88907 )

      I pretty mich doubt it can actually do things at this time. Well, let's say I doubt it can do more than a small number of things, that do not matter that much, with the accuracy and reliability needed. And hallucinations cannot be fixed with LLMs, or statistical models in general. Oh, and model collapse will pretty much make the current generation the last one trained on public data from the Internet. Oh, and model poisoning is a thing.

      My take is some people want to get filthy rich and hence hype this tech

    • Actually the report is brutal: 7 pages of marginal benfits and 13 pages of substantial
      risks.

      Potential benefits:

      - "Conduct sentiment analysis of public feedback on state policies"

      - "Summarize meetings"

      - "identify specific groups or subsets of participants that may benefit from additional outreach"

      - "convert educational materials into formats like audio books"

      - "Translate COBOL into [Ruby on Rails]. This can slash timelines, reduce bugs, and democratize development." (Good luck with that.)

      Among the many risks

  • It may have not been ai but the Australian government used automated analysis of benefit payments and resulted in many people being hounded incorrectly and a number of persons removing themselves from among the living. It would need to be backed up with multiple levels of actual people checking before action should be taken based on an ai analysis. Use it as a prompt but donâ(TM)t treat it as gospel

    • Human review of the positives doesn't resolve the issue of people getting their hands on the AI and using it to tune their tax filings to avoid detection of their own tax fraud.

      And then getting away with it even if caught on the grounds that "the government's own tax audit AI was used to review this to ensure compliance!"

      • Human review of the positives doesn't resolve the issue of people getting their hands on the AI and using it to tune their tax filings to avoid detection of their own tax fraud.

        "AI" is a black box that doesn't show its workings. You won't know what to hide, because nobody will know what the criteria actually is.

        And then getting away with it even if caught on the grounds that "the government's own tax audit AI was used to review this to ensure compliance!"

        The AI isn't making the final decisions. It's just a tool to help the people who are making the decisions.

        • There is tax software that follows rules to fill out your tax forms for you. Now imagine you change the rules so they produce options for deliberate misreporting, and you feed those into the AI to see what gets flagged. You don't have to understand the black box to utilize it.

          And your second point is not a rebuttal of mine.

          • There is tax software that follows rules to fill out your tax forms for you. Now imagine you change the rules so they produce options for deliberate misreporting, and you feed those into the AI to see what gets flagged. You don't have to understand the black box to utilize it.

            You're assuming the AI never changes. Next year when the AI notices a pattern of deliberate misreporting, it will be trivial to go back X number of years and catch everyone who did this all at once.
            X is whatever the statutory limits on audits are.

            And your second point is not a rebuttal of mine.

            How are they getting away with anything? You just said they were caught.
            And then getting away with it even if caught on the grounds that "the government's own tax audit AI was used to review this to ensure compliance!"

  • We seem to find one every few years.

    • by Rei ( 128717 )

      This internet thing is a fad. Paper is faster.

    • by gweihir ( 88907 )

      There is a sucker born every minute...

      No idea why with AI, perpetrating this ages-old scam seems to be even simpler. All demonstrations of AI so far left me very unimpressed. Simplistic, has no ability to fact-check, hallucinates, cannot even do beginner-level stuff in many areas (students of mine tried it on an utterly simple firewall exercise with zero success), etc. Especially hilarious was a presentation were a supposed senior pedagogics expert demonstrated chat AI to create teaching material. I mean, m

      • > No idea why with AI, perpetrating this ages-old scam seems to be even simpler
        > It did create the crap it created very fast though and it looked good, even if it was not.
        • by gweihir ( 88907 )

          You imply that is enough to impress many people, including many people with money? Makes some sense, in a very unfortunate way. Well, I have concluded a while ago that most people are idiots, and this hype is just one more inciator for that.

          • Definitely. I'm not a fan of this either of course. "Very fast" can make a lot of difference, just like UI responsiveness can make huge difference or how if we play back images quick enough we can see motion. "Looking good" is also unfortunately more important than being as good as it looks, when it comes to marketing to users or investors. I'm sure you've seen corporate presentations vs academic presentations on tech for example, and how the former can be 100x more polished than the latter, whereas the con
      • Ridiculous, really, sort of similar in my field (physics) where people would feed data to AI, receive some "results" and when asked to explain them in terms of theory will grind to a halt.

        The few times I've helped dig through "results", they turn out to be statistical artifacts from hidden or poorly understood correlations in detectors or other experiment setup defects. Sad, really.

        I mean, it is useful to have a helper to sift through a lot of data, but the researcher should know why and how it works, not j

        • by gweihir ( 88907 )

          the researcher should know why and how it works, not just feed "AI" garbage and label the output as "results".

          Yep. Otherwise they have no business calling themselves "researchers".

  • Can't have that, the lobbyists will prevent it.

    • s/lobbyists/donors/

      That's why a certain US political party wants to defund the IRS.

      • by gweihir ( 88907 )

        Yep. So utterly convinced they have truth and honor and know how to build a bright future, that they are convinced _they_ do not need any checks and balances. That is how you build fascism and have everything reliably go to shit.

    • by gweihir ( 88907 )

      Naa, it will hallucinate so many problems with small-time tax payers that the large ones will be even safer!

  • translations (Score:3, Insightful)

    by jjaa ( 2041170 ) on Sunday November 26, 2023 @04:35AM (#64032121)
    "quickly translate" this might look like a win from native english speaker perspective. but each time i see such automated translations (offered unrequested, just because of my location) like docs, articles or video descriptions - the cringe meter shoots through the roof. The automation is so obvious - errors are appalling - it only gives off a sense of disrespect to the reader. and unfounded pride of the srvice (we have umptheen langs!) unjust, causing anger!
  • ..what companies are already doing, e.g. increase productivity & reduce labour costs. You know, like Amazon are famous for using the current technology, algorithms, for. They'll probably also use it for union-busting campaigns, thwarting lawsuits against worker abuses, etc.. Billionaires & PACs will use it to manipulate political processes in their favour & promote extremist policies with the result that it'll further enshittify our societies. Just imagine how much more money & power the sup
    • Sure, but actually Amazon listings have already nosedived in my experience. It used to be that Amazon was a reliable shopfront where I could find a book, put it in the cart and have it sent to me with one click. This year, about half the books I bought had "issues" ranging from incorrect titles or descriptions, lost packages, super long delays and arbitrary cancellations. I sent some products back asking for the correct replacement, just to get the same one again. If this is Amazon's AI future, they had bet
      • OK, I was referring to Amazon as an example because they're infamous for being abusive, criminally negligent employers & my comments were criticisms of corporations that do this in general, & that GenAI will most likely serve to augment these kinds of abuses & far more. If GenAI is as powerful as they claim, then these corporations & their clients are a direct & imminent threat to our democracies.

        If you'd like to leave a review of your online shopping experiences, please find the appr
      • by Entrope ( 68843 )

        Do you buy things that are shipped by some entity other than Amazon? If so, that's your problem. Don't trust "third-party" fulfillment. (Yes, I know that Amazon is the third party from a certain perspective. I don't think that's a helpful perspective.)

    • by Rei ( 128717 )

      To quote XKCD about any new technology:

      Will [ ] make us all geniuses? No
      Will [ ] make us all morons? No
      Will [ ] destroy whole industries? Yes
      Will [ ] make us more empathetic? No
      Will [ ] make us less caring? No
      Will teens use [ ] for sex? Yes
      Were they going to have sex anyway? Yes
      Will [ ] destroy music? No
      Will [ ] destroy art? No
      But can't we go back to a time when- No
      Will [ ] bring about world peace? No
      Will [ ] cause widespread alienation by creating a world of empty experienc

      • =D

        Re: "We were already alienated." - Not around here, we aren't.

        Also: "Will [ ] further concentrate wealth & power into the hands of an unelected, unaccountable few? - Yes."
        • by Rei ( 128717 )

          If you had spent time in the open source LLM community, you might not be so confident of that statement.

          It's like thinking that operating systems will stay in the domain of wealthy powerful corporations because it takes such a large amount of resources to develop an OS.

          • Yeah, they said that about social media in the beginning, that we'd all have our own sites & feeds & the like. I guess those still exist to some extent.
            • by Rei ( 128717 )

              They did?

              Who did? I don't remember such a timeperiod. I don't even remember the term "social media" being "a thing" until Facebook. Like, there were "blogs", but these weren't generally self-hosted. People running their own websites was seen as nerdy and esoteric.

              • Your comment sounds like it comes from the United States of Amnesia. Social media (the activity, not the term) has been around in various forms for a lot longer than you imagine. See: https://en.wikipedia.org/wiki/... [wikipedia.org]

                There were many such claims about decentralised networks & many attempts continue to this day, e.g. Mastodon. Still, the space is dominated by a handful of large corporations & even for specialised, closed systems for corporations & institutions, those social networking platform
  • They probably declared it a carcinogen.
  • Articles about the risks of AIs make me think of the several Star Trek: Lower Decks episodes and the Daystrom Institute Self-Aware Megalomanical Computer Storage Facility housing the various evil AIs like "Lord Tyrannikillicus" ...

  • It used to be a simple, joyfully oblivious place. Thanks, Silicon Valley. Thanks a pant-load.
    • by Rei ( 128717 )

      It used to be a simple, joyfully oblivious place

      When was that, exactly - 1518?

      • The whole rest of the country used to stereotype it that way for most of living memory, so I'm gonna say your question is dumb.
        • by Rei ( 128717 )

          Let me ask Rodney King if he remembers the California of several decades ago in as idyllic of a manner as you do.

          • Anecdotes are the refuge of idiots and liars. Which are you?
            • by Rei ( 128717 )

              I'd rather be the curator of anecdotes than the architect of delusions.

              • Have you ever set foot in this country, or are you just another self-important jackass who thinks our entertainment industry accurately represents us to the likes of you?
                • by Rei ( 128717 )

                  I spent far more years than I would have cared to living in that country, thanks.

                  Including, I should add, three in California, plus numerous visits.

                  • Right. You know so much about this country and California specifically that you thought bringing up Rodney King was an intelligent point. Lemme waste my time on Wikipedia finding some incident in your country / region's history and say "Whatabout that time that blah blah blah" if you say anything that's not completely masochistic.
                    • by Rei ( 128717 )

                      Wow, amazing, Rodney King was a freak one-off incident, and not just emblematic of decade after decade of seething racial conflict and inequality, amazing!

                      Next you'll inform me that there's weren't any conflicts related to migrants in California! No housing crises! No massive problems with wildfires from overdevelopment! No aging infrastructure, esp. the power grid! No problems with a government constantly struggling with a debt crisis! No mess of a proposition system where the public has voted away right

        • The whole rest of the country used to stereotype it that way for most of living memory, so I'm gonna say your question is dumb.

          I used to listen to the Beach Boys and watch surfer movies starring Annette Funicello.

          Then I grew up, moved to California, and saw reality.

          The Beach Boys weren't surfers, it was all an act.

          The beaches in California are cold, rocky, foggy, and the waves are usually less than a meter. It's one of the worst places in the world to surf but one of the best for hypothermia.

          The "California Dream" was all false advertising.

          • "Foggy" my balls. It's just a marine layer. And as for the water temperature, it's called a wet suit. And wave height? Come on. Like it's a bigger achievement to perfect a handful of elite surfers rather than being a school for introducing the art.
  • New headline: government ignoramuses produce study about something they don't understand.

  • by javaman235 ( 461502 ) on Sunday November 26, 2023 @01:04PM (#64032809)

    Do not like this new trend of security by scientific illiteracy. The knowledge of how to make nuclear weapons is nothing, you get dangerous when you have mined the uranium, and have the hundreds of millions of dollars of centrifuges turning to enrich it. Same with dangerous chemicals. A person making TNT without a license is dangerous, a person who knows how to make it is any chemist. The war against dangerous knowledge is a war against a scientifically informed public.

    • Anybody who isn't totally internet-illiterate can find and follow instructions on how to make sulfuric acid, nitric acid, nitroglycerin and dynamite. A chemist who doesn't specialize in explosives chemistry, however, is much more likely to do it safely. And, BTW, there's a reason why I listed those chemicals in that order; can you figure it out?
  • The State of California found that AI causes cancer?
  • It blows my mind that a site like this has such ignorant discussion of this topic. It's not artificial intelligence or anything close to it. It's a collection of data structures and algorithms. AI processors are just like normal processors except they lack most of the scalar instructions you'd expect in a CPU and replaces them with vector instructions that play well with the data structures required/utilized by neural nets.

    Which is exactly why the state of California made their recommendation. None of thi

RADIO SHACK LEVEL II BASIC READY >_

Working...