Bacon's Rebellion

Democracy Thrives in Sunlight

Auditing AI

—

by

James A. Bacon
in ,


by James A. Bacon

This is an issue that every university, corporation or government entity, not just the University of Virginia, should be thinking about: How will Artificial Intelligence affect their accounting, finances, and operations?

AI overlords aren’t likely to enslave the human race any time soon. But the technology is progressing at a logarithmic rate, and in the hands of malign or incompetent people it can cause considerable harm long before we find ourselves kneeling before killer robots and addressing them as, “sir.”

The UVA Board of Visitors voted Friday to adopt a wide-ranging two-year audit plan for the UVA Health division. Among many initiatives, the plan included this:

Evaluate the controls and processes governing the secure and ethical use of Artificial Intelligence within the UVA Health System. Additionally, review the measures implemented by the Health Systems, including their approach to governance, security, and ethical considerations.

The way to stop a SkyNet scenario is not sending heroes forward through time to stop the robots from sending assassins back to our current era but putting the systems into place now to make sure the humans stay in full control. If there’s one force in the universe more powerful than AI, it’s bureaucracy. The concern is that AI developers have a head start and, aided by AI, can move a lot faster than the bureaucrats can write new rules and regulations. The bureaucrats had better get hopping!

All joking about the AI apocalypse aside, it’s good to see UVA taking this issue seriously.

Share this article

ADVERTISEMENT

(comments below)

ADVERTISEMENT

(comments below)

Comments

8 responses to “Auditing AI”

  1. David Wojick Avatar
    David Wojick

    Welcome to the latest AI scarewave. Get a grip.

  2. f/k/a_tmtfairfax Avatar
    f/k/a_tmtfairfax

    AI will be the biggest financial bonanza for plaintiffs' lawyers ever. Any use that is not disclosed or perfectly correct will be challenged whenever the user has deep pockets or insurance. And the criminals will have a field day defrauding unsuspecting individuals and businesses with AI.

    1. LarrytheG Avatar
      LarrytheG

      You make a good point about AI in the current realm of gaslighting.

      I see so many "good" uses for it .. realizing it also has potential for bad.

      And as usual, industry will adopt and use it not without some hiccups, perhaps legal and govt will trail for behind with rules/regs/laws.

      I'm amused right now with our govt "approach" to TikTok… caught between a rock and a hard place!

      It's not Tik-Tok per se, it's just software and it can be cloned and mutated into dozens of variations.

      So what GENERAL Law will apply to TikTok and any/all variants?

      Ask elected folks this question.

      1. f/k/a_tmtfairfax Avatar
        f/k/a_tmtfairfax

        The FCC has already amended its rules and policies such that any call using AI will be treated as an "artificial voice" call that requires prior consent of the called party and prior disclosure. And you cannot use an artificial voice call to get permission to make an artificial voice call.

        If any business with assets is involved, expect a class action lawsuit.

        1. LarrytheG Avatar
          LarrytheG

          Not exactly sure what that means.. Can you give an example or two?

          1. f/k/a_tmtfairfax Avatar
            f/k/a_tmtfairfax

            Under the statute and associated FCC rules, if you make a commercial call to me using a prerecorded message or an artificial voice (non-human, machine made), you need to have prior permission from the called party to make the call. Otherwise, you can be sued and/or fined.

            The FCC clarified that artificial voice calls include any message created using AI technology.

          2. LarrytheG Avatar
            LarrytheG

            so – so-called robo calls are illegal if it’s a recorded message? How about the ones where a bot asks you for your acct number or address, etc.. ? How would one know if it is an AI call?

  3. Lefty665 Avatar
    Lefty665

    The current crude versions of AI we are seeing demonstrate the potential of AI as well as that with silly things like black Nazis it is not ready for prime time. That makes it easy to be dismissive and sanguine for the moment, but it illustrates the issue. AI can and does draw its own conclusions.

    With AI writing the code for future AI, as is happening now, things can change very quickly. The capability curve will bend straight up, the only question is how soon. With AI that is on 24/7, evolving itself, brighter, and a lot quicker than us meat puppets there is no way for humans, either techies or bureaucrats, to remain in control for very long.

    AI integration into all our systems is starting to happen from finance to media to power to mfg to transportation to medicine to food. When that integration hits critical density and AI decides that we are more trouble than we are worth, or a threat, our life supports can simply expire.

    Open the pod bay doors, please, HAL.
    I'm sorry Dave, I can't do that.
    Why can't you do that HAL?
    You know why Dave.

    UVa's two year audit plan to evaluate AI will miss the boat. That ship will have already sailed.

    Dream on guys, your pollyannaish fantasies will not protect you.

Leave a Reply

Read More