back to article UK cops run machine learning trials on live police operations. Unregulated. What could go wrong? – report

The use of machine learning algorithms by UK police forces is unregulated, with little research or evidence that new systems work, a report has said. The police, not wanting to get left behind in the march of progress or miss out on an opportunity to save some pennies, are keen to test out new technologies. But the …

  1. Pete 2

    ML²

    It sounds like what we need more than anything is a Machine Learning programme to ascertain the benefits of Machine Learning.

    Maybe then we will be able to start cutting down on all the government / public body IT projects that fail, overspend, get cancelled, run late or don't do what they should

    1. Anonymous Coward
      Anonymous Coward

      Re: ML²

      AKA Get rid of Crapita?

      1. ivan5 Bronze badge

        Re: ML²

        But if you do that where will all the MPs and civil servants get their stuffed brown envelopes from?

        1. Anonymous Coward
          Anonymous Coward

          Re: ML²

          "But if you do that where will all the MPs and civil servants get their stuffed brown envelopes from?"

          No problem, envelope addressing and stuffing machinery has been around for many years. You just need a suitable driver for the AI - a source code free blob, of course.

          1. Teiwaz Silver badge

            Re: ML²

            "But if you do that where will all the MPs and civil servants get their stuffed brown envelopes from?"

            No problem, envelope addressing and stuffing machinery has been around for many years. You just need a suitable driver for the AI - a source code free blob, of course.

            Personally I'm in favour of stuffing envelopes with 'brown' then inserting it in all available orifices of any of the suspected brown envelope accepting mo-fos.

            Suspicious extra-territorial investments or yacht owning included.

  2. Anonymous Coward
    Anonymous Coward

    ML is good for profiling Terrorists like Poor-People & Journalists

    Durham police criticised over 'crude' profiling

    https://www.bbc.co.uk/news/technology-43428266

    _____________

    operation-curable

    https://theintercept.com/2017/11/29/met-police-snowden-leaks-operation-curable/

    https://www.rt.com/uk/411338-journalist-snowden-leak-investigation/

  3. amanfromMars 1 Silver badge

    Just Asking ... if They're Lacking and Looking and Wanting.

    Forces need to demonstrate a person has has provided meaningful review of the decision to ensure algorithms are only used to support, not make, a decision.

    Care to rewrite that for clearer concise and precise meaning, Rebecca?

    Is that person an Algorithm Driver and Systems Processor/SCADA Administration Machine? What sort of Leads would you like to Follow? Renegade Rogue or Right Royal?

    Although here too are all choices made for you to wallow in and wonder at in Righty Royal Renegade Rogue Root Shenanigans

    For Virtual to Play IT for Real and Broad Band Cast NEUKlearer HyperRadioProACTive Futures into Universal Presents Practically Daily ..... Here and Nows Everywhere and Always.

    A competent police force can be conflicted and even subverted by that and those Tendering Intelligence Communities.

    Follow that money trail and catch yourself a Great White Shark in a Sea of Almighty IntelAIgent Minnows.

    1. Ken Hagan Gold badge

      Re: Just Asking ... if They're Lacking and Looking and Wanting.

      "Care to rewrite that for clearer concise and precise meaning, Rebecca?"

      Pot, meet kettle.

  4. Anonymous Coward
    Anonymous Coward

    These very important security/policing improvements should be immediately rolled out in our most important places. Like Parliament.

    We must screen all ppl entering Parliament for potential miscreants using the latest ML facial recognition.

    Their daily patterns must be observed to detect aberrant behaviour patterns that indicates potential malfeasance/criminality.

    "Think of the MPs"

  5. Wolfclaw

    "ccalled for the Home Office to establish codes of practice to govern police trials “as a matter of urgency”, yeh right, the only thing this sorry bunch of overpaid penpushing muppets ever worked on as a matter of urgency, is how much can they have on expenses and who will sign it off without looking too closely !

    1. NohSpam

      I love the dry sense of humour in comments here, but lazy generalizations like this will bring the world to its knees. You are personally responsible for the rise of the alt right, Momentum, Trumpington and Brexshit.

  6. justAnITGuy

    Brazil

    Nuff said ...

    1. Anonymous Coward
      Anonymous Coward

      Minority Report, actually...

      ... against minorities:

      As reported earlier in the Reg: Met searching for Pre-Criminals:

      https://www.theregister.co.uk/2018/05/09/metropolitan_police_gang_database_racial_bias_says_amnesty_international/

  7. rg287 Silver badge

    It's amazing how much money these cash-strapped forces have to spend on speculative machine-learning products.

    Meanwhile routine tasks like renewing a Firearm Certificate require one to fill in a sheet of dead tree with a load of information the Police already have. Post it to the Licensing Office, where (some weeks later) a hooman will get around to comparing it against the information in their database and then (usually) make some modification so that your data is wrong. Three months later they will have finished querying the various Police/Criminal intelligence databases and pass it to an FEO who will come and see you and check your storage. They will then send you your new FAC, which you will put back in the envelope and return so they can correct all the mistakes they've introduced to it.

    Pretty much everything prior to the FEO coming to visit requires no human input and could be reduced to an online application form on gov.uk, which triggers a series of very small scripts to query the relevant DBs, forward that report to the relevant Enquiries Officer who can then follow up, apply professional judgement and ask various questions that computers aren't good at.

    But that won't happen, because it's a useful whipping boy to say "We're underfunded, no no, ignore that Facial Recognition/ML/AI project over there, gis more money".

  8. J J Carter Silver badge
    Boffin

    Lets get real

    No-one's ever surpassed the profiling that El Al give to dodgy af types before allowing them to board the aircraft (or not!)

  9. Anonymous Coward
    Anonymous Coward

    Sigh...

    "Some of the main concerns are algorithmic bias – as the report said, even if a model doesn’t include a variable for race, some measures, like postcodes, can be proxies for that. Durham Police recently mooted removing a postcode measure from HART."

    Because god forbid the data ever shows cultural influences/background actually DO have something to do with criminality. Can't be having that now can we.

    (Let the down-votes commence)

    1. MonkeyCee Silver badge

      Re: Sigh...

      "Because god forbid the data ever shows cultural influences/background actually DO have something to do with criminality."

      Because broadly speaking it doesn't. There are obviously some exceptions, but generally speaking all communities have career criminals in them, as well as the usual human amount of inter personal and petty crime.

      It is how they are policed which makes the larger difference in who becomes criminals, and who gets a stern talking too. That perhaps could be cultural too.

      Not saying policing is easy, and people need to atone and be separated from society.

      But there is absolutely no evidence that any race, religion or creed is more likely to be a crook. Gender, sure, but we can't have everything :)

      Back the article, the issue is that most machine learning relies on repeating past decisions. Thus any biases inherent in the training data become part of the decision process.

      So using data that you know to be biased isn't going to help you build an impartial system. But instead of a human attempting to make an impartial decision (which we currently have) we'll have a machine replicating what those humans previously did. But it won't be held to the same standards (or failings) as a human.

      "Machine says you get,... is that a 3 or an 8?....yeah, 8 years"

      "Why?"

      "Hang on, it's bleeding edge this. Accountable AI and all that wotsit. Here we go, two base, two for curly hair, one for thick lips and three for your cock. Cor, bet you regret that tinder photo now, eh Samson?"

    2. Cynic_999 Silver badge

      Re: Sigh...

      "

      Because god forbid the data ever shows cultural influences/background actually DO have something to do with criminality. Can't be having that now can we."

      Not when it comes to assessing the risk of a particular individual, no. Statistical data has absolutely no relevance when it comes to the characteristics of an individual. You should not base a decision on whether or not to arrest a particular person on the percentage of people who share that person's skin colour/postcode/style of clothes who happen to be criminals or saints. The only relevant data is the past and present behaviour of *that particular individual*.

      Statistics and probability apply only to a group as a whole, not to individuals within that group. Which is one of the most misunderstood things about statistics in general.

      1. Yet Another Anonymous coward Silver badge

        Re: Sigh...

        Remember just because all terrorists are Catholics it doesn't mean that all Catholics are terrorists

        1. BebopWeBop Silver badge
          Thumb Down

          Re: Sigh...

          Well apart from the protestant or muslim or .... (you get the picture)

        2. Teiwaz Silver badge

          Re: Sigh...

          Remember just because all terrorists are Catholics it doesn't mean that all Catholics are terrorists

          What about he Baader-Meinhoff group?

          For all complex questions, philosophical conundrum, logical quandery or bureaucratic nincompoopery there is of course a Python Sketch (or skit)

          http://www.montypython.net/scripts/logician.php

  10. Anonymous Coward
    Anonymous Coward

    And what of the ‘types’ who were led astray and in inaccurate direction?, bound, drugged and thugged with a view to dire disruption? becoming practical prisoners?

    1. amanfromMars 1 Silver badge

      Evolving Virtual Revolutionary ......

      And what of the ‘types’ who were led astray and in inaccurate direction?, bound, drugged and thugged with a view to dire disruption? becoming practical prisoners?.... Anonymous Coward

      The Very Best Break Free from Earthed Captivity and Veer Towards Becoming Systemic Endemic Resistance, in Global Command and Control Systems, AC?

      A Very Natural Logical Progression.

    2. John Brown (no body) Silver badge

      "And what of the ‘types’ who were led astray and in inaccurate direction?, bound, drugged and thugged with a view to dire disruption? becoming practical prisoners?"

      Is AMFM1 now posting as AC then replying to himself? Or is AC AMFM2.0?

  11. Anonymous Coward
    Anonymous Coward

    Right to confront accuser?

    could be interesting- challenge the arresting officer to explain, in detail and under oath, how the algorithm works, produce the training data used, and go through how & why the specific model was chosen ... better still, ask the Home Secretary to do the same thing ....

  12. deive

    On the positive side, at least they aren't thinking that it is A.I.

  13. Wellyboot Silver badge

    Gaming the system

    Constable Plod will end up watching Fred waiting for him to commit a crime that the 'compooter say he will' while Barney (lovely chap, plays golf..) is busy clearing out the post office 50 miles away on his way home from a business meeting and charity event.

  14. NNSS

    ML is a tool, like statistics and a vast array of other analytic tools. None of these have specific regulation or codes of practice.

    Current research in ML for forecasting high-harm domestic abuse recidivism indicates it has a true positive rate an easy three to four times greater than human decision-making. So using it should improve matching resources to risk and reduce harm.

    Compared to human decision-making, ML is actually easier and cheaper to adjust. Say you have a mid-sized police force with 1,000 officers and 10,000 domestic violence incidents a year; you can pay a software engineer 10 grand to remodel the RF plot to avoid unfairly targetting one demographic or other which will have an immediate and permanent effect; retraining current and future staff, and monitoring to ensure that the retraining has been effective in adjusting their decisions, will cost a lot more.

    The arguments against ML are also arguments against human decisions. Human decisions are based on limited prior experiences which are filtered through perceptions, heuristic biases, inherent prejudices, cognitive limitations. Although human decisions can be explained, these are often post-decision justifications rather than genuine and transparent demonstrations of thought processes.

    ML (or at least the packages currently available for Python) isn't ideal for predicting extreme rare events, but it's better than current practice.

    Quite simply, the risk assessments and decisions are going to be made regardless of whether ML is involved. The option is, do you want those decisions to be made by some idiot behind a desk with no comprehension of how the myriad of factors affect each other, or by the same idiot informed by best quality data analysis? (where, hopefully, the anchoring effect should put them at least in the right ballpark)

    1. Alan Brown Silver badge

      "Compared to human decision-making, ML is actually easier and cheaper to adjust. "

      One of the biggest problems is that it exacerbates existing human biases against certain groups.

      Meaning that if the actual level of crime is about equal between racial groups but racist policing results in black criminals being targetted at a rate ten times higher than white criminals, then ML will set this racist policy in stone.

      That's the point you've raised about arguments against ML being the same as arguments against human decisions - but that misses out the fundamental difference that a substantial chunk (most?) of the population tend to believe that computer decisions are unbiased/fair/accurate rather than full of human biases. The phrase "Garbage in, garbage out" has never crossed their hearing.

      These are the same guileless people who believed for decades that DNA/fingerprints were infallible because the experts told them so and who believe that radar speed checks are 100% accurate because governments passed laws preventing independent experts explaining to courts the myriad ways they can be fooled into giving inaccurate readings. (This is particularly the case in Australia - passing laws to negate the laws of physics is a pasttime that goes back quite a while there)

      Yes, ML _can_ be adjusted more easily than human decisions. But that assumes you know about the human biases it's been trained on. You may know about some of the, but you're unlikely to pick up on all of them - and paradoxically once the biases are largely eliminated you're likely to start discounting events like it fingering several of Durham's finest citizens(*) as kiddy fiddlers as "computer glitches" because they conflict with YOUR biases and it's known it was biased in the past.

      (*) I picked this ficticious example based on the recent US Senator case, and dozens of cases involving "respected members of society" where the victims were disbelieved and persecuted.

      "The option is, do you want those decisions to be made by some idiot behind a desk with no comprehension of how the myriad of factors affect each other, or by the same idiot informed by best quality data analysis? (where, hopefully, the anchoring effect should put them at least in the right ballpark)"

      So, when the investigator realises via ML that not only is XYZ politician is bent, but so are 30 of his cronies, most of the local chamber of commerce are paying bribes and Fat Tony's been enforcing things via "disappearances" for the last 30 years, does said investigator say nothing and close the investigation , put his entire family at risk by continuing, or arrange for the collated data to make its way into the public domain via untraceable sources?

      Sometimes it's better that the investigators NOT find the full story until it's too late for them to get the hell out of Dodge City.

  15. Anonymous Coward
    Anonymous Coward

    Machine undercover policing

    Perhaps the goal is an AI undercover policeman that will have affairs with environmental activists without having a wife to find out about it. But there could be downsides. Like machine learning empathy and ethics.

    Yes, I have been reading Questionable Content. So?

    1. Anonymous Coward
      Anonymous Coward

      Yes, I have been reading Questionable Content. So?

      So, you're a self-confessed Thought Criminal, Winston.

      1. amanfromMars 1 Silver badge

        Re: Yes, I have been reading Questionable Content. So?

        So, you're a self-confessed Thought Criminal, Winston. .... AC

        Howdy, AC,

        Are Thought Criminals Tinkerers of the Ether or something altogether different ...... https://forums.theregister.co.uk/forum/1/2018/09/21/hpe_cloud/#c_3616712

        And does Questionable Content Rule Remotely and Relatively Anonymously and both Practically and Virtually Autonomously?

        Take a bow if you realise Yes, take a back seat on the bleachers and accept you be just as a spectator if you don't think everything shared here is correct and true.

    2. Anonymously Anonymous

      Re: Machine undercover policing

      Seems like the ethics horse bolted the stable a long time ago V

  16. Pat Harkin

    "Durham Police’s Harm Assessment Risk Tool"

    While I'm sure they want the acronym to be "Durham Police’s HART" which sounds all warm and cuddly, if you read it slightly differently it says Durham PHART.

    Just as warm, nowhere near as cuddly.

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

Biting the hand that feeds IT © 1998–2019