Ever since Microsoft's chatbot Tay started spouting racist commentary after 24 hours of interacting with humans on Twitter, it has been obvious that our AI creations can fall prey to human prejudice. Now a group of researchers has figured out one reason why that happens.

Last Wednesday, with relatively little fanfare, Google introduced a new technology called Google Knowledge Graph. Type in “François Hollande,” and you are offered a capsule history (with links) to his children, partner, birthday, education, and so forth. In the short-term, Knowledge Graph will not make a big difference in your world—you might get much the same information by visiting Hollande’s Wikipedia page, and a lot of people might still prefer to ask their friends. But what’s under the hood represents a significant change in engineering for the world’s largest search-engine company. And more than that, in a decade or two, scientists and journalists may well look back at this moment as the dividing line between machines that dredged massive amounts of data—with no clue what that data meant—and machines that started to think, just a little bit, like people.

With a large enough sample, any outrageous thing is likely to happen. The point is that truly rare events, say events that occur only once in a million [as the mathematician Littlewood (1953) required for an event to be surprising] are bound to be plentiful in a population of 250 million people. If a coincidence occurs to one person in a million each day, then we expect 250 occurrences a day and close to 100,000 such occurrences a year.

Going from a year to a lifetime and from the population of the United States to that of the world (5 billion at this writing), we can be absolutely sure that we will see incredibly remarkable events. When such events occur, they are often noted and recorded. If they happen to us or someone we know, it is hard to escape that spooky feeling.

Infographic on compulsive hoarding

Ever wonder how Google manages to search the entire web and return results in half a second? Well, RobertvH from Munich did, and Mountain View’s head of web-spam, Matt Cutts, talks you through it in the above YouTube video. The short answer? Lots of backend firepower and, you know, a few years in the search game. If you remember the Google dance, Cutts explains what caused that, before going on to give a good idea about how today’s version of the site does what it does. If you’re thinking this all sounds a bit too much like SEO 101, you’d be half-way right, but as Matt’s delivery is so engaging, we’re def hoping there’ll be a follow up.

The differences, across a number of different brands of automated essay scoring software (AES) and essay types, were minute. “The results demonstrated that over all, automated essay scoring was capable of producing scores similar to human scores for extended-response writing items,” the Akron researchers write, “with equal performance for both source-based and traditional writing genre.”

“In terms of being able to replicate the mean [ratings] and standard deviation of human readers, the automated scoring engines did remarkably well,” Mark D. Shermis, the dean of the college of education at Akron and the study’s lead author, said in an interview.

Can physicists produce insights about language that have eluded linguists and English professors? That possibility was put to the test this week when a team of physicists published a paper drawing on Google's massive collection of scanned books. They claim to have identified universal laws governing the birth, life course and death of words.

The paper marks an advance in a new field dubbed "Culturomics": the application of data-crunching to subjects typically considered part of the humanities. Last year a group of social scientists and evolutionary theorists, plus the Google Books team, showed off the kinds of things that could be done with Google's data, which include the contents of five-million-plus books, dating back to 1800.

One of the things I instinctively hated about my “ancestral culture,” that of Bangladesh, is that there wasn’t that great of an emphasis on individual independent thought. Why, for example, was it important never to drink water while you were eating, as opposed to after you were done? The response was simple: that’s the rule. Even if there was a functional rationale, there wasn’t even any pretense at offering a reasoned explanation for why a custom was a custom. It’s just how it was.

Traditionally, human rights work has been more akin to investigative reporting, but Ball is the most influential of a handful of people around the world who see that world not in terms of words, but of figures. His specialty is applying quantitative analysis to mountains of anecdotes, finding the correlations that coax out a story that cannot easily be dismissed.

Could the movements of refugees have been random? No, Ball said. He had also plotted killings of Kosovars and found that both phenomena occurred at the same times and in the same places -- flight and death, hand in hand. "I remember well the moment of astonishment that I felt when I saw the killing graph for the first time," Ball replied to Milosevic. "I assumed I had made an error, because the correlation was so close."

Something had caused both phenomena, and Ball examined three possibilities. First, the surges in killings and flight did not happen during or shortly after NATO bombings. Nor were they consistent with the pattern of attacks by Albanian guerrilla groups. They were consistent, however, with the third hypothesis, that Serb forces conducted a systematic campaign of killing and expulsions.

This mental approach to causality is often effective, which is why it’s so deeply embedded in the brain. However, those same shortcuts get us into serious trouble in the modern world when we use our perceptual habits to explain events that we can’t perceive or easily understand. Rather than accept the complexity of a situation—say, that snarl of causal interactions in the cholesterol pathway—we persist in pretending that we’re staring at a blue ball and a red ball bouncing off each other. There’s a fundamental mismatch between how the world works and how we think about the world.

Stephen Pinker argues that we are becoming less violent. Nonsense, says John Gray

Statistical analysis of OKCupid profiles exposes some sexually fascinating revelations:

• Herbivores like giving oral more than omnivores
• Twitter users are more likely to masturbate today
• Christians and Atheists are just as likely to claim they have never masturbated
• The correlation between men who prefer gentle sex & use of the word 'boating'

I f**king love statistics

What will the digital world look like in ten years?  The trends are already clear.

Capacities in bandwidth and storage will continue on their exponential path.  The explosion in the volume of information and number of devices will persist.  Our data will be linked and most likely be processed in qubits rather than bits.

However, trends tell us very little.  It’s discontinuities that drive history.  Everything seems fine and then boom!  E-commerce comes along, then search engines, social media, smart phones and on and on.  Much like the flood that set Noah on his journey, such events, although driven by trends, take us in completely new directions and create new orders.

We used to have massive mainframes, which were housed in a basement somewhere. Users could log on and, if they had booked some time, could use the computer and the output would show up on their screen or get sent to a printer.  Then came the PC revolution and you could do it all by yourself.  

Most of the time Google does not actually have the answers. When people say, “I looked it up on Google,” they are committing a solecism. When they try to erase their embarrassing personal histories “on Google,” they are barking up the wrong tree. It is seldom right to say that anything is true “according to Google.” Google is the oracle of redirection. Go there for “hamadryad,” and it points you to Wikipedia. Or the Free Online Dictionary. Or the Official Hamadryad Web Site (it’s a rock band, too, wouldn’t you know). Google defines its mission as “to organize the world’s information,” not to possess it or accumulate it. Then again, a substantial portion of the world’s printed books have now been copied onto the company’s servers, where they share space with millions of hours of video and detailed multilevel imagery of the entire globe, from satellites and from its squadrons of roving street-level cameras. 

Gleick makes his case in a sweeping survey that covers the five millenniums of humanity’s engagement with information, from the invention of writing in Sumer to the elevation of information to a first principle in the sciences over the last half-century or so. It’s a grand narrative if ever there was one, but its key moment can be pinpointed to 1948, when Claude Shannon, a young mathematician with a background in cryptography and telephony, published a paper called “A Mathematical Theory of Communication” in a Bell Labs technical journal. For Shannon, communication was purely a matter of sending a message over a noisy channel so that someone else could recover it. Whether the message was meaningful, he said, was “irrelevant to the engineering problem.” Think of a game of Wheel of Fortune, where each card that’s turned over narrows the set of possible answers, except that here the answer could be anything: a common English phrase, a Polish surname, or just a set of license plate numbers

In case you haven’t heard, the newest champion of "Jeopardy!," the popular TV game show, is a computer. Watson, an enormous computer developed by researchers at IBM, was pitted against the two previous human champions, Brad Rutter and Ken Jennings. At the end of the first round, aired on Valentine’s Day, Jennings and Watson were tied for first place. But Watson trounced both humans in the next round, despite making some odd mistakes. And he won the second game, aired on February 16, suggesting the first victory was more than just beginner’s luck. When the IBM computer Deep Blue beat chess champion Garry Kasparov in 1997, it was not doing anything qualitatively different from an ordinary calculator. It was just calculating really quickly—running through all the possible chess moves in response to the previous move by Kasparov and picking the one most likely to succeed. That’s just the sort of problem that a fast-enough computer running the right algorithm was bound to solve.

Yet again, Wikipedia is about to break new ground. The website that has become one of the biggest open repositories of knowledge is due – within the next week or so – to hit the mark of 3m articles in English. It's all a very long way from January 2001, when Wikipedia launched. Its first million articles took five years to put together, but the second was achieved by 2007. It was not just the number of articles that grew, but also the number of people involved in creating them. During Wikipedia's first burst of activity between 2004 and 2007, the number of active users on the site rocketed from just a few thousand to more than 300,000. However, statistics released by the site's analytics team suggest Wikipedia's explosive growth is all but finished. The quickening pace that helped the site reach the 2m article milestone just 17 months after breaking the 1m barrier suddenly evaporated: adding the next million has taken nearly two years. While the encyclopedia is still growing overall, t

What does it mean to read in a digital age? Researchers are just beginning to explore the question, and educators are engaged in passionate debate about how reading may be changing on the Internet. It is impossible to write about any one piece of research at great length, so for those interested in more in-depth information, here are links to some studies, speeches, reading tests — old and new — and other resources.