Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Math Programming Software The Internet Technology

How Algorithms May Affect You (phys.org) 85

New submitter Muckluck shares an excerpt from a report via Phys.Org that provides "an interesting look at how algorithms may be shaping your life": When you browse online for a new pair of shoes, pick a movie to stream on Netflix or apply for a car loan, an algorithm likely has its word to say on the outcome. The complex mathematical formulas are playing a growing role in all walks of life: from detecting skin cancers to suggesting new Facebook friends, deciding who gets a job, how police resources are deployed, who gets insurance at what cost, or who is on a "no fly" list. Algorithms are being used -- experimentally -- to write news articles from raw data, while Donald Trump's presidential campaign was helped by behavioral marketers who used an algorithm to locate the highest concentrations of "persuadable voters." But while such automated tools can inject a measure of objectivity into erstwhile subjective decisions, fears are rising over the lack of transparency algorithms can entail, with pressure growing to apply standards of ethics or "accountability." Data scientist Cathy O'Neil cautions about "blindly trusting" formulas to determine a fair outcome. "Algorithms are not inherently fair, because the person who builds the model defines success," she said. Phys.Org cites O'Neil's 2016 book, "Weapons of Math Destruction," which provides some "troubling examples in the United States" of "nefarious" algorithms. "Her findings were echoed in a White House report last year warning that algorithmic systems 'are not infallible -- they rely on the imperfect inputs, logic, probability, and people who design them,'" reports Phys.Org. "The report noted that data systems can ideally help weed out human bias but warned against algorithms 'systematically disadvantaging certain groups.'"
This discussion has been archived. No new comments can be posted.

How Algorithms May Affect You

Comments Filter:
  • by JustAnotherOldGuy ( 4145623 ) on Tuesday February 14, 2017 @10:36PM (#53870865) Journal

    It's the end of intelligence as we know it.

    • Maybe just the opposite, the beginning of intelligence. The problems are garbage data, conflation, risk analysis with random failures, entropy, and just ignoring facts-- among so many other problems.

      How far does an algorithm take bias until it's actually discriminating based on such things as gender, race, etc? We're in the very early stages of "big data" and we're doing a bad job of it. The problem is this: we'll continue doing a bad job until we have more transparency, IMHO.

  • by Waffle Iron ( 339739 ) on Tuesday February 14, 2017 @10:48PM (#53870951)

    In my day, we had a simple and effective way to judge algorithms:

    O(n log(n)) or faster: good
    O(n^2) or slower: bad

  • Because the processes were so transparent to begin with. At least the algorithms have the possibility of being looked at. Maybe that should have been the story.
    • by Anonymous Coward

      First of all, the problem is these systems increasingly take away power from the judicial and other democratic systems, which actually were somewhat transparent. You can see the law that affects you if you want, it's public. With algorithms, you can't. An example are the 'risk of recidivism scores' that increasingly influence judges' sentencing. Those, it turned out, disproportionally called black people risky.
      https://www.propublica.org/ser... [propublica.org]

      Secondly, those new algorithms are rarely transparant, for a numb

      • The Volkswagon Diesel scandal was because the decisions Volkswagon executives made weren't transparent, not algorithms. Despite laws being readable, when it comes to enforcing and interpreting them, you have to rely on nontransparent judges and juries. People aren't transparent and it has always gone back to people, no matter how much you want to throw shade on the new stuff, or claim that this or that aspect of the old stuff somehow made things transparent.
    • We should be demanding access to the data and algorithms used to generate pricing for mandatory services like the ACA, home insurance, and automobile insurance. We should never be required to buy anything without even knowing the basis for the charges.

      • We should be demanding access to the data and algorithms used to generate pricing for mandatory services like the ACA, home insurance, and automobile insurance. We should never be required to buy anything without even knowing the basis for the charges.

        When greed and corruption pervert capitalism, one doesn't have to look far to understand pricing.

        Let's also not pretend insurance is a business concept that has ever struggled to survive. They collect a few billion, and then hire an army of lobbyists to ensure their flavor greed is mandatory.

        How this corrupt process works isn't some kind of mystery to solve.

    • But even if the algorithms are 100% open and transparent, that means nothing if the data feed into them is poor. If the bank uses an algorithm to determine if it want to lend money to you, how is the data about you collected? Who decided to classify you as a say medium risk person? What cirterias did he/she/they use for that? How thorogh were he/she/they in gathering decition material? What did he/she/they miss/ignore/misunderstood?

      Unless there is full and complete transparency and accountability for dat

    • ... At least the algorithms have the possibility of being looked at. Maybe that should have been the story.

      Algorithms that determine what goods and services a person has access to, the quality of said goods and services, and how much he or she pays for them, need more than the "possibility" of being looked at. They need legislation that requires them to be publicly released in their entirety before being put into use, and every modification, bug-fix, etc. needs to be published as soon as, or before, implementation takes place. And there need to be really onerous and consistently enforced penalties for not relea

  • What groups?
    Some federal database might buy a state database and find lots of illegal migrants getting free city or state services?

    That a person is a religious covert? Does their faith or cult have issues? A person buying products or searching for topics that get reported and tracked?
    A person looking to travel? Most of the US online tracking is looking for any trace of radicalization and mobilization. Is a person of interest looking up interesting things?
    Get some US gov/mil work? Need a polygraph?
    • All that a government might want to do something about in the future, don't just think about the us where the gov already have their people's asses
  • by Anonymous Coward

    The author seems to think that the alternative to "algorithms" is people making good decisions. That's simply not true. The alternative is people attempting, or not, to follow some agreed on some ill-defined process.

    Note that algorithms make it harder to ignore that we have to make tradeoffs. To take one of the examples from the article, we may have to choose between "well-respected teachers" and those who actually help students significantly. (Of course, we could reveal performance information and let p

    • by Sique ( 173459 ) on Wednesday February 15, 2017 @02:46AM (#53871721) Homepage
      No, the author makes the point that algorithms don't exculpate anyone from making bad decision. "The computer said: No." is no excuse for mishandling someone. We had the example on Slashdot of the algorithm that tries to predict recidivism and thus recommend probation or prison. A deeper analysis showed that it was biased against black people because it predicted higher recidivism for them than they had in reality, and it was biased pro whites as it predicted lower recidivism rates than real. And it was not even factoring in the skin color of the people in question. But the way it weighed the socio-economic factors seems to be the problem. It was scoring high on recidivism when many socio-economic risk factors were slightly up, but gave low scores when only one or two risk factors were high, but all others were low. Thus it was overestimating the recidivism rates of poor people with a weak family background, but completely missing the recidivism risk of well off people from a stable family, but deep personal problems.

      But because the program was actually used in judiary decisions in several States, it unnecessarily sent people to prison, while it recommended to set high risk people free on probation, and it did it with a strong racial bias that was contradicted by reality.

      • Culpable or not people find it easy to offload their morals when there is a system to blame. Take Nazi Germany as example...
        • by Sique ( 173459 )
          Yes, it's Milgram [wikipedia.org] all over again. The seemingly knowledgable expert is replaced by the seemingly neutral algorithm, and thus people blindly follow their directions.
  • "undereducated voters"

    • by Ihlosi ( 895663 )
      It is a mistake to assume that anyone who is susceptible to rhetoric is 'uneducated'. Rhetoric and persuasion exploit the time gap which it takes for knowledge and reason to kick in, for one thing.
      • ... mistake to assume that anyone who is susceptible to rhetoric is 'uneducated' ...

        Think that over a bit:

        In the 2016 election, a wide gap in presidential preferences emerged between those with and without a college degree. College graduates backed Clinton by a 9-point margin (52%-43%), while those without a college degree backed Trump 52%-44% [pewresearch.org].

        • I read the phrase "think that over" as meaning you disagree with the quoted text. However, your link does not contradicting that statement, unless you also believe Clinton employed less rhetoric during the campaign. Wikipedia defines rhetoric as "the art of discourse, wherein a writer or speaker strives to inform, persuade or motivate particular audiences in specific situations". Even if we use the less formal definition ("excessively flowery or emotional, often meaningless, speech") can you truly say Clin
  • Al Gore [wikipedia.org] rhythm [wikipedia.org]

  • no [youtube.com]
  • by mbeckman ( 645148 ) on Tuesday February 14, 2017 @11:33PM (#53871163)
    Is this what passes for intelligent discourse at Phys.org? So many errors, so little time. First, an algorithm is NOT involved when I pick a movie to stream on Netflix. I pick the movie, Netflix streams it. Unless you want to count the code necessary to display a web page and process a click. Second, algorithms are NOT "complex mathematical formulas." A formula is a specification for a single computational step (or a series of similar steps, in the case of calculus). An algorithm is a non-mathematical procedure, with memory, decision making, input and output from and to various sources and sinks, and, well, formulas. Algorithms contain formulas, but formulas don't contain algorithms. And phys.org does not contain the sense God gave raisins.
  • Agent Smith (Score:4, Insightful)

    by Zobeid ( 314469 ) on Wednesday February 15, 2017 @12:14AM (#53871307)

    "Which is why the Matrix was redesigned to this, the peak of your civilization. I say your civilization, because as soon as we started thinking for you, it really became our civilization, which is of course what this is all about."

  • "A federal appeals court decisively struck down North Carolina’s voter identification law on Friday, saying its provisions deliberately “target African-Americans with almost surgical precision” in an effort to depress black turnout at the polls."-July 29, 2016
    https://www.nytimes.com/2016/0... [nytimes.com]

  • Black Box Society (Score:2, Insightful)

    by Anonymous Coward

    A recent report on this in the Netherlands summarised it as "playtime has to be over". Big data and the algorithms that work on top it are getting a serious amount of power over our lives. Any little scrap of data is starting to influence your chances of getting a job, a cheap loan, or even a date.

    If you want to know how scary this gets, check out this presentation by Alexander Nix on how he used this type of data to influence the elections.
    https://www.youtube.com/watch?... [youtube.com]

    Or have a look at the new "Social

  • Take the deployment of Police resources. They try to predict in what area crimes are more likely to happen at a time in point. Area has higher crime rate, police forces are deployed in the area. Result: even more crimes recorded in that area.
    Model says people are not the right candidate to get the job. Result: you don't get the job and other people who look good to the model get the job. Again, more data that "proves" those people should not get that job and the other people do.
    • Few people count the results of random searches in the crime rate. Crime rate depends on crime reported, not recorded.

    • Before models, managers would interview potential new hires, and would generally hire or not hire based on the impression they got of the candidate. That's not necessarily better than a faulty algorithm. The algorithm can be examined and changed.

  • ... garbage in, garbage out.

  • The fact is that SJWs cannot seem to comprehend that inequality in result isn't itself proof of some bias, PARTICULARLY if the bias-factor isn't even part of the algorithm.

    Further, the fear is that simple objective analysis will occur without human intervention, and thus lack someone to call racist, sexist etc (in essence, so they're pre-labeling the author of algorithms as racist, sexist etc.).

    For example
    Your algorithm shows that people below a certain income level fail to repay loans at the normal rate, s

  • The first movie I ever watched on netflix was Inside Out. At the end, Netflix's first recommendation was that if I liked Inside Out, I should watched Inside Out in french.

    Great algorithm there. Oh the complexity. What's next? The spanish version?

    Probably the worst suggestion any person could have ever made to anyone outside of a french class.

    The algorithm must have been so happy. Think about it. It found a movie, where every word spoken is totally different, but there's a 100% match on the title! Woo

  • Doesn't work on us. We can think at the same level as the creators of said algorithms. It's not a "War of Math Destruction" it's a "War of Meta Thinking". It is akin to playing a game of chess and trying to guess based on previous experience from playing your opponent what you think they will do and make a move to counter it. But if you're opponent is thinking in the same manner and is thinking you might think in this way and anticipates you arriving at that conclusion, he/she can counter your counter.

    T

So you think that money is the root of all evil. Have you ever asked what is the root of money? -- Ayn Rand

Working...