You're accused of a crime - who do you want to judge you: man or machine? Hannah Fry explains

Поделиться
HTML-код
  • Опубликовано: 5 сен 2018
  • Mathematician Hannah Fry is here to explain the change of heart she's had about one particular quandary. If you were guilty of a crime who would you rather determined your fate? A human judge or a machine. Based on her knowledge of algorithms (and humans) your instinctive reaction may not be the right one.
    Discover more about her fascinating book, Hello World, here:
    SIGNED copies here: bit.ly/2NRDuHa
    F O L L O W U S
    TWITTER --} / waterstones
    FACEBOOK --} / waterstones
    INSTAGRAM --} / waterstones
  • РазвлеченияРазвлечения

Комментарии • 29

  • @ChiktarWorshipper
    @ChiktarWorshipper 5 лет назад +9

    I'm still desperately waiting for Hannah to make some A.S.M.R. videos. Her voice is pure gold.

    • @DamianReloaded
      @DamianReloaded 5 лет назад

      Do you people really fap to that? To me it's almost as weird as eating dirt. ^_^

  • @DamianReloaded
    @DamianReloaded 5 лет назад +3

    It would be very VERY interesting to do a similar experiment with the judges but also giving the cases to current expert systems to get some statistics on how they would perform based on the evidence presented to them.

  • @philipripper1522
    @philipripper1522 5 лет назад +3

    It depends ENTIRELY on who wrote the algorithm.

  • @MultiEcko15
    @MultiEcko15 5 лет назад +6

    Can I just apologize n mean it

  • @gitman3486
    @gitman3486 3 года назад

    Being a well presented young woman is far more of an advantage in front of a judge than any machine impartiality

  • @danecobain
    @danecobain 5 лет назад +3

    I'm glad you hinted at algorithmic bias because algorithms might not be all they're cracked up to be. Seems to me like the best mixture is AI+humans combined. I read somewhere that judges are more lenient right after lunch and at their strictest right before lunch. I wonder what we'll do when an algorithm is accused of committing a crime. Who will be responsible?

    • @Locutus
      @Locutus 5 лет назад

      I agree, it's going to be a mix of AI and Humans. We're now using something similar in medicine with IBM's Watson. You have Drs inputting what's wrong with the patient into Watson, and Watson gives them a treatment plan. They have had good results using Watson.

  • @nicolahocking5981
    @nicolahocking5981 5 лет назад

    Such an interesting decision. I'm not sure which one I'd choose. Maybe both??

  • @darcipeeps22
    @darcipeeps22 5 лет назад

    I got her book free since it was my first on Audible. I love it

  • @usandmexico
    @usandmexico 5 лет назад

    The variability even within the same judge is more evidence that supports David Hume: Reason is the slave of the passions. We shouldn't pretend that our sentiments don't affect our reasoning, even when we are being most reasonable.

  • @gmshadowtraders
    @gmshadowtraders 5 лет назад +1

    The flaw in her argument is stated right at the beginning. You do not know for certain that the person is guilty of the charge. Hence there is an element of guesswork. This is why there will always be room for human judgement in the equation, even if it is flawed I believe it's many times more superior than relying completely on a machine-learning algorithm. A computer is useless at detective work when you think about it properly. Let it stick with matching dog and cat faces to images please.

    • @DamianReloaded
      @DamianReloaded 5 лет назад

      guesswork (in a court room) is based on bias not fact, also this is about judges, not detectives.

    • @gmshadowtraders
      @gmshadowtraders 5 лет назад +1

      @@DamianReloaded A judge has to decide on the merits of the case, whether or not there is truth to an allegation. Judges may have started as detectives or investigators earlier in their career. A 'detective' to me, can be anybody that must weight up the evidence and make a decision. Decision-making, or a rules-based system, is thus something shared in common between a human and a computer. There are further issues at play here, but I won't go into them. Suffice to say, this argument about which method is better will probably never be settled, and we will continue to have AI shoved down our throats, and continue to have judges who make mistakes. Until something better comes along.

    • @DamianReloaded
      @DamianReloaded 5 лет назад +1

      @@gmshadowtraders if the merits of the case are based on logical facts, an expert system ought to arrive to a logically sound veredict. I'm not saying it must be the final decision (at first). It could be a helper that would assist judges for minor offenses and such and where the human judge would supervise at the end and ponder special cases.

  • @nawaf6993
    @nawaf6993 5 лет назад

    i'd still pick human because most likely he'd see my regret or remorse and lighten the sentence
    so what i'm trying to say is even if people did the same crime the emotions they show and actions should and do influence the sentence but where i'm willing to compromise is if the machine would be an advisor to the judge

  • @MrDaanjanssen
    @MrDaanjanssen 5 лет назад +1

    Computer says no

  • @femsadbaks9993
    @femsadbaks9993 4 года назад

    I 100% agree! It's law, it needs to be done accurately. It'll still be humans training whatever model you deploy but A.I will take in all the relevant factors and produce an accurate result. I believe in humans but I trust A.I more than them.

  • @Robocop-qe7le
    @Robocop-qe7le 4 года назад

    Why there are two “hello world” books?

  • @ghollisjr
    @ghollisjr 4 года назад

    If I program the machine, then can I be my own judge?

  • @cornelbalau9584
    @cornelbalau9584 3 месяца назад

    Also consider this vitamin B17 vs cancer!

  • @balloonist477
    @balloonist477 5 лет назад

    Where do lawyers come in? Perhaps they need to have their own competing probability algorithms and let the judge have the final say.

  • @reverse5461
    @reverse5461 5 лет назад

    WOW

  • @rmason4358
    @rmason4358 5 лет назад

    Very interesting. Thank you. I've considered AI for such tasks based on conformity and equal justice for all. Assuming it's matured and with safeguards against manipulation. It could cross-reference more aspects of our activity and history, and also other cases of similar nature. Cars are now capable of self control and navigation. I favor this as it takes human error, road rage, tiredness and foolishness out of the driving experience. If all vehicles were autonomous but aware of each others location and direction they could avoid colisions more easily than humans and keep traffic flowing more easily. No more jammed up junctions caused by in considerate or thoughtless drivers.
    It's almost a shame AI couldn't be used to ascertain the remaining wealth of the planet and be 'Storeman' to allocate on demand. There is far too much wastage.

  • @nozzerjuice
    @nozzerjuice 5 лет назад

    i love you

  • @moonlitegram
    @moonlitegram 5 лет назад

    Disagree entirely and its specifically due to the machines consistency.
    I think the desire for a fully consistent application and interpretation of law in court cases comes from the desire to believe that there is a perfect legal system and that there is a singular correct answer for what laws should be. And to me this is on par with believing utopia can exist. It's never going to happen, just like we're never going to get every person on the planet to completely agree on what laws should be, how they should be applied and essentially what makes perfect justice.
    There's always going to be people that are going to call a hearing incorrect, oppressive, unjust, tyrannical etc while other people will defend a verdict passionately. The same thing goes with legislation etc. So for me, I've come to the point where I don't think the optimal strategy is to keep trying to force the square peg through the round hole. We're never going to achieve that kind of utopia or level of agreement among people. So I think the optimal system is to embrace this diversity and embrace decentralized societies where you have different regions with different laws and different interpretations of legal concepts etc. And let people congregate around the laws they prefer to live under. (If you're reading this and you think this sounds crazy, realize that we do have this to a certain degree already today. We already have different countries, states, towns etc with different laws. It's really not as foreign as it sounds when you first confront this perspective)
    And in order for such a system to work, you need to accept and allow for judges to make different judgements and interpretations. As long as we support liberty enough for people to live in and move to areas where the law most suits their perspective, this kind of divergence in adjudication isn't such a bad thing. Is it a perfect system? Of course not, utopia isn't possible. But I think its far more optimal strategy considering the diversity of perspectives and thought that's always going to exist between such a large population of individuals.
    With that said, what I would actually love to see in regards to incorporating algorithms into law is some kind of legal process inspired by bitcoins mechanisms for keeping record and communicating information in a decentralized network. Such a system would allow us to achieve an even more diverse, yet still fair, system of law that's less tied to region than it currently is. Imagine a system where individuals could keep a record of their desired legal principles to live under that could be compared against another individuals records in a court hearing during a legal dispute rather than forcing each party in that case to bend to the will of some external third party that was responsible for x legislation written at y time. Instead, a judge could make a ruling based on the two involved parties desires for law, and adjudicate accordingly to the best of his ability.
    Still not a perfect system, but it would be a system that would apply law and create legal outcomes at a level that is much closer to the personal preferences of both parties involved. And it makes sense to place their preferences at the highest regard in that individual case since they're the two parties that are actually going to deal with the consequences of the decision. Rather than the system we have now where the fates of two parties could be decided by the reasoning of an individual or group of individuals who aren't even alive anymore because they wrote the peice of legislation that led to the judge or juries decision years or even centuries ago... We accept this as the norm today without much thought, but when you really think about it, it's pretty preposterous.

  • @ldsaunders8021
    @ldsaunders8021 5 лет назад +1

    You are completely wrong. The reason why we have people on juries is because they can understand the reasons why people act the way they do. As far as the claim that the same case to the same judge gives a different answer? Name any two cases that are the exact same. I've been a trial attorney for decades and have never, ever, seen two identical cases. The problem is you may be a mathematician, but you are not a trial attorney and don't know what takes place in a trial.

    • @lmcc9636
      @lmcc9636 5 лет назад +4

      In the experiment she's referring to, judges were presented with multiple hypothetical cases, some of which were the same but with names changed. They came to different judgements on what was literally the same case :)