back to article Microsoft president: We said no to Cali cops' face-recog tech – and we won't craft killer robots

Microsoft president Brad Smith has revealed that the company turned down an order from California cops for its facial recognition technology over human rights concerns. Speaking at Stanford University this week, Smith said Microsoft had concluded that its technology would probably lead to innocent women and minorities being …

  1. I.Geller Bronze badge

    Why instead of using the term bias don't switch to BIOS

    Why instead of using the term bias don't switch to BIOS - a set of computer instructions in firmware which control input and output operations? And begin to make Lexical Clones (collections of structured texts and images) as the foundations for AIs?

    1. I.Geller Bronze badge

      Re: Why instead of using the term bias don't switch to BIOS

      Three thumbs down! Who re these people? Google?

      1. This post has been deleted by its author

  2. Borg.King
    Coat

    "All your faces are belong to us!"

    attrib. Arya S. Braavos.

    (Mine's the one with the Valyrian steel dagger in the pocket.)

  3. Anonymous Coward
    Anonymous Coward

    If the updates to the facial recognition software are on par with their Windows updates then the cops are going to be cuffing lots of innocent people and missing the crims.

  4. a_yank_lurker

    When AI Will Be Ready

    Artificial Incompetence is touted to ready in 5 to 10 years but it seems the 5 to 10 years is always 5 to 10 years away. This reminds of navel gazing about fusion power generation being about 20 years away over 50 years ago and it is still '20 years away'. AI is the fusion power of the IT industry; always promised but never arrives.

    One of the problems alluded to in the post is the quality of the dataset and the suitability of the dataset for 'training' the AI system. They are not identical problems. Datasets are often not as good as assumed so there is always a certain amount of garbage in that will polluted the system. Another problem, even if you have relatively high quality data, is often it was not originally collected for use with an AI system. So it is incomplete for the purposes of AI but may be more than adequate for other purposes.

    1. JohnFen

      Re: When AI Will Be Ready

      "Artificial Incompetence is touted to ready in 5 to 10 years but it seems the 5 to 10 years is always 5 to 10 years away. "

      At this point, it's pretty much tradition. AI has been 5-10 years away for all of the 30 years that I've been in this industry.

      1. I.Geller Bronze badge

        Re: When AI Will Be Ready

        AI answers Factoid and Definition questions, which is NIST TREC definition. IBM Watson answers both, plus Google Waymo answers and applies the found patterns driving its cars. Thus AI exists.

        1. a_yank_lurker

          Re: When AI Will Be Ready

          Sounds like redefining AI to be something that can be done rather anything resembling intelligence. Keep on redefining it and any dbms is now an AI system; the marketing wallahs will love it.

          1. I.Geller Bronze badge

            Re: When AI Will Be Ready

            The current AI was originally conceived as a response to the challenge of NIST TREC. Yes, originally conceived as a search technology and a purely commercial project. Yes, this search technology is the basis for AI.

        2. JohnFen

          Re: When AI Will Be Ready

          The NIST TREC definition is not the final or authoritative word on what AI is or whether it exists. It's not even authoritative outside of certain limited fields of interest.

          1. I.Geller Bronze badge

            Re: When AI Will Be Ready

            Perhaps... I should tell there I came from, shouldn't I? Now you know: Analytic Philosophy and NIST TREC.

      2. Drew Scriver

        Re: When AI Will Be Ready

        Seems to me that there's a shift from talking about AI to ML (Machine Learning). Same with Autonomous Vehicles shifting to Assistive Technology.

        For each, the former isn't quite ready and there are growing doubts about the feasibility.

    2. Hans 1
      Windows

      Re: When AI Will Be Ready

      Fusion power is 100 years away, we do not even have a working prototype, and the industrial version will have to be much bigger -> the materials to use for the enclosure are not yet known for the prototypes we have, that last mere seconds,....

    3. Ken Hagan Gold badge

      Re: When AI Will Be Ready

      Obviously I can't tell you when AI will be ready but we can surely agree that it won't happen before we even have a definition of the I.

      1. I.Geller Bronze badge

        Re: When AI Will Be Ready

        Money is the only measure, do you agree? Google, from 2004, has made 100% of its money on my earliest version of AI. IBM Watson make billions right now, on the development for off-line.

  5. PhilipN Silver badge

    I wish

    "Microsoft president Brad Smith has revealed that the company turned down an order from California cops for any of its software over human rights concerns".

    1. Tigra 07

      Re: I wish

      The human right concern being that humans should not have to suffer Windows 10

  6. SVV

    we won't craft killer robots

    Well, at least we now know when Microsoft will unveil the Windows Killer Robot Edition from their underground bunker : just after the surface of the Earth is a piie of smouldering rubble.

    1. Mark 85

      Re: we won't craft killer robots

      Test targets will probably be anyone using Win7 or older.

  7. Anonymous Coward
    Anonymous Coward

    At least Microsoft had some sense to understand the limits/problems of their technology

    Google would happily go along with it and only reluctantly pull out when it became public and their employees revolted.

    Facebook would be all-in with Zuckerberg blaming everyone but himself when things went horribly wrong and the software flagged an innocent man as a criminal leading to cops rolling up on him with a hair trigger leading to a fatal result.

    1. NATTtrash
      Big Brother

      Re: At least Microsoft had some sense to understand the limits/problems of their technology

      You know, in some kind of way I do understand companies seeing an opportunity to make more money here. After all, they are companies, and that's what companies do. What is more questionable is public servants contemplating to face scan (all) their citizens, no matter what, just because they can and feel like it. It wouldn't surprise me if, at some point, they come up with a law that it illegal to cover your face...

      ...

      ...

      Oh dear...

    2. Chris G

      Re: At least Microsoft had some sense to understand the limits/problems of their technology

      That leads me to wonder what research the Zuck conducts on the billions of photos that have passed through his organs?

      Aside from passing them tp NSA silos.

    3. This post has been deleted by its author

  8. Anonymous Coward
    Anonymous Coward

    Two questions

    In which location could facial recognition most likely be used without negative PR impact, California, or China?

    In which location could facial recognition be sold with profits sizeable enough to matter to Microsoft, California, or China?

  9. Anonymous Coward
    Anonymous Coward

    Diversity

    "tech companies need staff who represent the world - women, African-Americans"

    How about some AFRICANS. A bunch of different coloured wealthy educated Americans does not create diversity it creates a monoculture.

    1. Ken Hagan Gold badge

      Re: Diversity

      In the context of an article on facial recognition software, I would have thought that "different coloured Americans" was indeed sufficient. In the wider context of AI, your point stands though.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like