19 comments

  • jp57 9 hours ago ago

    I know it's from the original headline, but I feel like "doesn't cover" is confusing wording. The article means the fourth amendment disallows the use of AI-generated probable cause.

    A quote from the court (re-quoted from the article): "The Fourth Amendment to the United States Constitution does not allow a detective to delegate to AI the task of assessing probable cause. In other words, the Fourth Amendment does not permit an individual to be arrested based on nothing more than an uncorroborated conclusion by AI that the individual in question has committed a crime. And no reasonable detective in the Defendant’s position could conclude otherwise."

    Using "cover" here strikes me as weird, since the fourth amendment prohibits government action, I would interpret actions not "covered" as not prohibited and hence allowed.

    • halJordan 5 hours ago ago

      Strictly speaking the fourth amendment doesn't ban anything, it guarantees the right to privacy with an exception for probable cause. So the court is saying that an AI generated thingy does not sufficiently satisfy the requirements for the exception which the fourth amendment carved. The 4th amendment's probable cause exception doesnt cover ai generated findings

    • vibebingo 9 hours ago ago

      They have a better title in the body: "an algorithm can’t make probable cause determinations"

    • defrost 9 hours ago ago

      I've altered the HN submission title wording to your suggestion.

    • seizethecheese 9 hours ago ago

      You're not off base, I also interpreted the headline in the opposite way of what was meant and was confused by the article at first.

  • jawns 8 hours ago ago

    The right way to interpret this ruling is that a probable cause affidavit needs to offer evidence that a person has committed a crime, and model output is not evidence; it is more akin to an investigative tool.

    Photographs, witness statements, direct observations, physical objects found at the scene are all evidence: They help answer questions such as "Was this person present at the scene?" or "Did this person take the actions they are alleged to have taken?"

    A machine-learning model, even a very sophisticated one, is really at its core a probability machine. It might help you find two photographs that appear very similar, but it doesn't independently establish that the people in those photographs are, in fact, the same person. Likewise, an AI model might help you determine the similarity of speech patterns or writing styles, but those similarity scores don't actually tell you if an audio or writing sample came from the same person.

    The person who submits the affidavit is vouching for the reasonableness of the facts they are presenting, and so any probabilistic evidence needs a human judgment call.

    The wrong way to interpret this ruling is that AI can't ever be used as an investigative tool to establish probable cause. A detective certainly can use an algorithm that searches a facial-identification database, or a fingerprint database, to help identify potential suspects. But they then need to do something else to determine whether the model output actually matches the facts of the case. They can't just say, "The computer told me so."

  • AnimalMuppet 10 hours ago ago

    Good!

    Note that this isn't just a case of a suspect trying to deny a warrant. This is the suspect, after the case was dismissed, coming after the cop financially for the completely bogus arrest warrant. And the court ruling said that "no reasonable officer" could rely on AI as the sole determinant that a warrant was justified. That wording means that the officer can't use "qualified immunity" to get out of this.

    So this is a pretty big deal. However, it's just a single decision. It's not an appellate decision, which would be binding on a district, still less a Supreme Court decision, which would be binding on the whole country. Still, it's a start. A very nice start.

    • idle_zealot 9 hours ago ago

      That "sole determinant" bit is concerning. So it's established now that when the AI tells you that there's probable cause you should then ask it to parallel-construct another source and present that when questioned. I guess this at least catches cops too stupid or lazy to integrate that step?

      • dcrazy 5 hours ago ago

        It’s not “parallel construction” to investigate an AI-generated hypothesis in good faith. If the officer in this case had been acting in good faith, he would have at least tried to find some proof of his completely unsubstantiated theory that the Facebook profile was an alias for the plaintiff.

      • trothamel 9 hours ago ago

        As far as I can tell, the system seems to be a fancier version of google image search. I think it's reasonable for a detective to toss an image into google image search, and look at the top matches. What he shouldn't be doing is avoiding verifying that the images actually match.

        Assuming the underlying crime should be a crime (and check fraud seems like it should be a crime), we do want the police to catch the actual criminals.

        It strikes me that AI is a red herring here - this is about image matching.

      • vibebingo 9 hours ago ago

        This is a core AI human interaction problem in a nutshell. It genetates synthetic reasoning and then a human is either convinced by it or too busy or lazy or corrupt to think for themselves and presents it as their own thinking.

      • nextaccountic 7 hours ago ago

        This should be viewed as processual fraud

  • treetalker 8 hours ago ago

    U.S. District Court for the Eastern District of Kentucky.

  • emscb 8 hours ago ago

    According to the article, the policy already said facial recognition alone is not enough, but this still happened. Wouldn’t a process where a warrant request can’t go forward unless the verification steps are recorded be more useful? It also seems the bar for getting the warrant was a bit low.

    • AnimalMuppet 5 hours ago ago

      That would be a decent procedural requirement. But what this decision says is that the detective who did this can be sued personally for his overstep, and the qualified immunity that police have won't protect him. "Do this and go bankrupt" is not exactly a procedural restriction, but it's still a restriction with some real bite to it.

  • bitwize 6 hours ago ago

    Well, so much for the Instant Justice Machines... for now, anyway. https://www.youtube.com/watch?v=EERFGTQ2Vs0

    (this show was cyberpunk af and had no right to go as hard as it did for an 80s toy-commercial kids show)

  • Molitor5901 9 hours ago ago

    (To be clear, I don’t specifically know if this detective has these prejudices, but the chain of events in this case certainly support an assumption of this person’s assumptions, if you catch my drift.)

    This feels irresponsible. The author is basically calling the Detective a racist when that is not in evidence. It's an assumption, which this case is based on. I would have hoped the author would have avoided making the same mistake.

    • delecti 9 hours ago ago

      The typical syntax around here to show that you're quoting something is to start a line with a > character. Without that your comment looks like you are stating that parenthetical yourself, and then criticizing it.