• Earthwormjim91
    link
    fedilink
    English
    arrow-up
    8
    arrow-down
    2
    ·
    1 day ago

    A 99% false positive rate is not the same thing as a 99% failure rate though. You’d need to look at the false negative rate too. Since the goal is to find weapons and prevent school shootings, the false negative rate would be the actual failure rate. Like the TSA when they have undercovers try to bring weapons through, and 95% of them make it through.

    A 99% false positive rate isn’t good, but it’s not a huge problem if you also have a false negative rate of 0%. It just means the model is too overzealous. The problem would be that eventually those responsible for reviewing and reporting the hits would start ignoring them.

    If it were a 99% false positive rate and a high false negative rate, then yeah the whole thing is useless.

    • irgendjemand
      link
      fedilink
      arrow-up
      1
      ·
      17 hours ago

      I mean it’s right there in the title, I would say it’s a problem

      • Earthwormjim91
        link
        fedilink
        English
        arrow-up
        3
        arrow-down
        1
        ·
        23 hours ago

        The problem would be that eventually those responsible for reviewing and reporting the hits would start ignoring them.

        I literally addressed that.

        • LilB0kChoy@reddthat.com
          link
          fedilink
          arrow-up
          2
          arrow-down
          1
          ·
          23 hours ago

          Which you prefaced with:

          A 99% false positive rate isn’t good, but it’s not a huge problem if you also have a false negative rate of 0%

          It is a huge problem when it’s a purpose built system to identify firearms in a school and it’s ineffective at its only job.

          • Earthwormjim91
            link
            fedilink
            English
            arrow-up
            1
            ·
            23 hours ago

            Again, that does NOT mean it’s ineffective. That would be the false negative rate.

            If it makes 1,000 hits, falsely identifies 990 items as guns that are not, but correctly identifies 10 real guns, then it’s doing its job. It’s just overzealous on what it’s hitting on. It’s still finding all the actual guns, which is the only metric that is really going to matter in terms of public safety. The only real problem is that humans would start ignoring it when it hits.

            If it makes 990 hits and all are false positives, and it misses 10 real guns entirely, then that’s an issue.

            • LilB0kChoy@reddthat.com
              link
              fedilink
              arrow-up
              1
              arrow-down
              1
              ·
              22 hours ago

              The only real problem is that humans would start ignoring it when it hits.

              Unless it’s only job is to detect them and do nothing then this makes it an ineffective system and is a major issue.

              I work in IT and we see this happen often. AI is exacerbating the issue considerably, particularly in cyber security .

              An alert that’s wrong 990/1000 times gets ignored 1000/1000 times so those 10 where it’s right don’t matter. And apparently those 990 have a chance of generating a lawsuit.

              From a statistical perspective you may be right but from a practical perspective it is an ineffective and broken tool.

              • Earthwormjim91
                link
                fedilink
                English
                arrow-up
                2
                ·
                22 hours ago

                And now you’ve circled back to what I already addressed lmao. I literally said that that would be the problem, the humans starting to ignore the warnings. That’s different than saying it has a 99% failure rate because the failure rate would be judged on catching guns, not false positives.

                Have a good one dude.

                • LilB0kChoy@reddthat.com
                  link
                  fedilink
                  arrow-up
                  1
                  arrow-down
                  1
                  ·
                  22 hours ago

                  I didn’t say it had a 99% failure rate. I said it was ineffective at its purpose and responded to you because you downplayed the negativity of 99% false positives for its intended use.

                  And now you’ve circled back to what I already addressed

                  “People would ignore it which would be bad but, ya know, not that bad actually. I mean, sure those guns it detected would never actually be noticed but the system sure did identify them all.”

                  On top of the fact that it doesn’t fucking work at its only job it also would serve to create a false sense of security.

                  Add to that, in this specific case, it has the chance to cost desperately underfunded schools money in the form of lawsuits. Which is great because tax payers don’t pay taxes to put that money towards supplies, or teachers but to cover lawsuits.

                  • Earthwormjim91
                    link
                    fedilink
                    English
                    arrow-up
                    1
                    ·
                    21 hours ago

                    Yes, I’m aware. You jumped into a conversation about something else and decided to just keep yelling that you’re right and everyone else is wrong. I get it.

                    Have a good one.

      • Pieisawesome@lemmy.dbzer0.com
        link
        fedilink
        English
        arrow-up
        3
        arrow-down
        1
        ·
        24 hours ago

        The comment you are replying to is correct in regards to statistics and evaluation. There is more too it that “crying wolf”.

        False positive AND false negatives is the true measure of failure. True negative and true positive are the opposite measures.

        These four can be optimized for depending on the consequences of being incorrectly positive or negative. The consequences for negatively identifying an object incorrectly can outweigh the consequences of incorrectly identifying an object positively.

        Depending on that breakdown, a model might be optimized to falsely identify a bag of chips as a gun “just to be safe”. The consequences of missing a weapon might have outweighed the consequences of being incorrect. Crying wolf might be the desired outcome that the school or company wanted because they deemed the risk acceptable.

        That said, this is still stupid and police should not have come out after the alter was cancelled and CERTAINLY should not have behaved that way. 99% is way too high of a false positive rate

        ACAB, but I like statistics and wanted to explain it.

        • LilB0kChoy@reddthat.com
          link
          fedilink
          arrow-up
          2
          arrow-down
          1
          ·
          23 hours ago

          Crying wolf might be the desired outcome that the school or company wanted because they deemed the risk acceptable.

          A 99% false positive rate is a broken system. Mitigating a broken use of shoehorned “AI” by saying it’s not good but not that bad is ridiculous.