Would you be offended if an online algorithm wrongly guessed your age from a photograph? Mine was off by about 15 years – and not in the good way. Behind the seemingly harmless fun involved in the How Old website from Microsoft, serious 'machine learning' is involved. The more photos uploaded for the machine to work on, the better it gets at face and gender recognition - and guessing ages. Read about How Machine Learning Works in the Economist, or delve deeper into their whole series on artificial intelligence.
Software algorithms work by poring through mountains of data, learning as they go. What if information about you gleaned from an algorithm was used to determine your loan application success? A rule of banking has always been to know your customer. By harvesting online data, or looking at how someone fills in a form, lenders now reckon “they can know borrowers as never before, and more accurately predict whether they will repay” than from a person’s credit history. So reports Steve Lohr in If Algorithms Know All, How Much Should Humans Help? (New York Times). But, according to Lohr, even the companies behind the algorithms think that a human sometimes needs to screen the results before they can be used in real life.
If machines and business are learning from big data, shouldn’t government and the public? Opening up the data that government collects and produces can be beneficial. In Evanston, Illinois, people can now conveniently access restaurant health inspection scores on Yelp, because of open data. The company Accela “worked with Evanston to automate the release of the City's restaurant inspection data, allowing citizen-facing websites and apps like Yelp to easily display the restaurant scores to diners as they choose where to eat. It even updates in real time as scores change”. In Open Data for California?, Maury Blackman of Accela describes the leaps that sections of the US government are making in terms of open data and real world applications.
And if everyone else is learning, shouldn’t you be too? A good place to start would be to learn how your colleagues think. A lot of people in the web design or web development world feel pressure to learn skills from other disciplines – the idea that designers should be able to write code, and coders should understand colour and layout is pervasive. While there’s a lot to be said for a wider knowledge base, instead of trying to train specialists from one discipline to become specialists in another, perhaps the two could just become better at communicating to each other? In Designers And Developers: No Longer A House Divided, Ivana McConnell writes about placing emphasis on learning teamwork and collaboration, not on learning each other's hard-earned specialities.
But what can we learn right now? Don't use "click here" as a link title. See what I did there? Stephanie Leary explains in an article called Why “click here” is a terrible link, and what to write instead.
Came accross any interesting articles lately? We'd love to see them! Just post them in the comments section.