Can't AI Be Used for Something Important?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • marcus
    Member
    • Jul 2008
    • 9239

    Can't AI Be Used for Something Important?

    From https://www.thejakartapost.com/cultu...le-errors.html
    Title : ChatGPT faces complaint for 'uncorrectable errors'
    By News Desk (AFP) , Vienna April 29, 2024

    A Vienna-based privacy campaign group said Monday it would file a complaint against ChatGPT in Austria, claiming the "hallucinating" flagship AI tool invents wrong answers, which creator OpenAI cannot correct.

    NOYB ("None of Your Business") said there was no way to guarantee the programme provided accurate information...

    ChatGPT has openly acknowledged it cannot correct inaccurate information produced by its generative AI tool and has failed to explain where the data comes from and what ChatGPT stores about individuals, said the group.

    Such errors are unacceptable for information about individuals because EU law stipulates that personal data must be accurate, NOYB argued.

    "If a system cannot produce accurate and transparent results, it cannot be used to generate data about individuals," said Maartje de Graaf, data-protection lawyer at NOYB...

    ChatGPT "repeatedly provided incorrect information" about the birth date of NOYB founder Max Schrems "instead of telling users that it doesn't have the necessary data", said the group.

    OpenAI refused Schrems's request to rectify or erase the data despite it being incorrect, saying it was impossible, NOYB added.

    It also "failed to adequately respond" to his request to access his personal data, again in violation of EU law, said NOYB, and the firm "seems to not even pretend that it can comply"...
  • mamaliga
    Member
    • May 2024
    • 12

    #2
    As a daily user with several subscriptions to ChatGPT and its specialised offsprings, I can confirm that it often feels like you're talking to a 5 years old.
    However, while not trivial, the issue raised by Austrian politicians is really a non-issue, imho. ChatGPT is simply going to decline to provide specific details about EU individuals, quoting privacy concers (or EU laws), the same way it refuses to answer questions related to crime, prostiution, etc. Remember implementing GDPR compliance costed companies tens of billions, and its ongoing.
    This is very a reductive argument, but OpenAi could jsut feed GPT4 pieces or legislation and say "read this law and try not to break it".
    I agree that to some extent AI is still nascent, but it's already changing the world in countless ways. One of them is it's ability to adapt practically instantly (eg EU laws). Implementing changes at global level takes seconds for tech giants. Amazon's AI jacks up the prices for products with high repeat customers (or god forbid subscriotion), knowing that customers won't notice a 3%increase. A decate ago, this would take months of market research and analysis, now it's seconds. That's the true power of AI atm. Unfortunaltely, it mostly benefits corporate giants, whih is ultimately your point...

    Comment

    • marcus
      Member
      • Jul 2008
      • 9239

      #3
      Originally posted by mamaliga View Post
      ...One of them is it's ability to adapt practically instantly (eg EU laws). Implementing changes at global level takes seconds for tech giants. Amazon's AI jacks up the prices for products with high repeat customers (or god forbid subscriotion), knowing that customers won't notice a 3%increase. A decate ago, this would take months of market research and analysis, now it's seconds. That's the true power of AI atm. Unfortunaltely, it mostly benefits corporate giants, whih is ultimately your point...

      Yes , I know AI can do more important things (I was just being sarcastic) .

      But you say "it's ability to adapt practically instantly (eg EU laws)", what you mean by "adapt" ?
      I am constantly reading Indonesian Laws and Regulations , and to me , most of the time , I just need to read them to understand (very few times they are not written clearly and so you need to make further analysis) . Maybe AI can help , for example , by informing all related laws/regulations about a specific subject .

      What I see is that an AI system will "behave" according to the information it is "fed" with . So if somebody feed it with wrong information , clients will get a bad "advice" , isn't it ?

      Comment

      • mamaliga
        Member
        • May 2024
        • 12

        #4
        Yes , I know AI can do more important things (I was just being sarcastic) .
        I got that. I was was, too, but in a different way. This is gonna sound like ima conspiracy theorist/paranoid, etc, but AI at the moment is used in an exploitative way, and its goimg to facilitate the bigest transfer of wealth, upwards. Let me expand.
        Latest inovations were always available first to the afluet, which granted them advantages: financial, lifestyle, etc.
        But for the first time in history, the new tech is deployed (and has been deployed for at least a decade) to actively exploit the masses, in a systematic, realtime, comtinuous way. Sure the assembly line made Ford rich, but id argue he didnt do it (much) at the expense of the masses. ATm AI is a new tech purely focused on extracting revenue from the masses.
        Forget ChatGPT, which is a commercial general purpose tool. Speciffic purpose AI is the real moneymaker. I think it was about 15 years ago when Wired magazine published an article (i really wished i could find it) about one of the airlines supercomputers that was basicslly calling all the shots about routing traffic. Zero human intervention, Of course it woudl take in thousands of parameters like weather, comgestion, etc, and act accordingly ?but ultimately its metric was dollars. it made decisions like: a flight is going to be 10 mins late, which will cause delays for a 2 of our other flights. so we cancel those 2 flights. and reroute those pasengers who booked a 2 hour flight, to an indirect route with a 9 hour layover in Atlanta. Theres an FAA fine for delayed flights, but there isnt one for rerouting. and there was not much consumer protection. thye were actively cancelinag and rerouting flights for no other reason that it wpuld save them a few dollars. and screw the passangers. Or they wouldt mimd keeping the plane on the tarmac for 5 hours, even if they know it cant take off becuse other factors... theres no FAA penalty for that. And the system had built in all the regulations, fines, costs for any change. Efficiency and customer satisfaction were not even considered. it was purely dollars and cents. It was "fed" the regulation.

        now imagine that, 15 years later, with global merchants like amazon who adjsut their product prices in realtime based on demand. i had an amazon subscription to Gillette razor blades for like 2 years, until i noticed theyve doubled the proce. The price per/pc of Tide pods in a 11pc box is half compared to a 20pc bag. coz eveybody has been buy?ing the 20pc for years, and they dont bother to check the price.

        other examples: Cambridge Analitica. Ukraine war disinformation. Ads selling you shit that yo don't need.

        I am aware of AI powered legal tools, but theyr still limited to finding relevant precedents. i would tend to agree tat indomesian regulation is not concise or consistent in implementation.

        What I see is that an AI system will "behave" according to the information it is "fed" with . So if somebody feed it with wrong information , clients will get a bad "advice" , isn't it ?
        Always keep in mind that youre talking to well read a 5 years old. dont take advice from him.

        Maybe AI can help , for example , by informing all related laws/regulations about a specific subject .
        Atm ChatGPT out of the box will refuse to give any specific information about, legal, medical,, matters. theyll post generall stuff from the web, and "better consult a specialst".

        im not saying its all bad, were definitely on the verge (well, 2 gears in actaully) or an industrial revolution. lots of people are gonna lose their jobs to AI. the new jobs will be "AI prompters" people who can extract most value out of general purpose AI. for example ive been relatively successfull in automatimg a large part of my job by teaching an AI persona to respond to cliemt emails in the fashion and style that my Wharton School of Business CEO would respond. i can give you more details on Dm if youre interested,

        Comment

        Working...
        X