Google gives $1M grant to Press Association to develop robot journalists

New consumers of the future could be reading stories pieced together by advanced data-analyzing robots rather than human journalists, if Google has its way.

On Thursday, the Press Association, the U.K.’s national news agency, announced that it received €706,000 from the tech giant for its Reporters and Data an Robots (RADAR)     1    .

A collaboration between the Press Association and data-driven news start-up Urbs Media, RADAR aims to set up an artificial intelligene-fueled, news service that will generate tens of thousands of news stories a month using publicly     2     data.

Everyone from big-name news organizations to hyper-local outlets and bloggers could     3     from the program. Press Association editor-in-chief Peter Clifton claimed, “this is a hugely exciting development, and we believe our     4     with Urbs Media can be a genuine game-changer for media outlets across the U. K. and Ireland.”

According to the Press Association, RADAR won’t     5     mark the end of the flesh-and-blood journalists, but will rather enable the AI to produce a volume of stories that would be impossible to match manually.

The envisioned work-flow would begin with human journalists identifying open data sets and “creating detailed story templates across a(n)     6     of topics including crime, health and employment.” The robotic reporter would then take over and scan the data, use language generation software to craft together story text and automatically locate relevant photos and video. Press Association clients would then be able to use a special distribution platform to identify news stories of     7     to their audience.

Content automation isn’t a totally     8     concept in the news industry however. AP estimated that it can free up 20 per cent of journalists’ time, allowing them to focus on more complex, qualitative tasks.

Shockingly, not all journalists are sold on the AI infiltration. A study found that journalists from leading news organizations had several reservations when shown an automatically     9     sports story. “I would never, ever, ever have written a story like that,” one BBC journalist said, while a CNN reporter thought the story was repetitive and lacked variation.

Nevertheless, it looks like AI in journalism is here to stay. At a time when many media outlets are     10     commercial pressure, RADAR will provide the news ecosystem with a cost-effective way to provide insightful local stories.

Computer programs recognize white men better than black women

Software that recognizes faces has bounded ahead in recent years, aroused by a boom in a form of artificial intelligence called deep learning. Several firms now offer face recognition as a     1     service, via their respective clouds. The ability to recognize in faces such things as an individual’s sex has improved too, and this is also commercially     2    .

The algorithms(算法)     3     have, however, long been suspected of bias. Specifically, they are declared to be better at     4     white faces than those of other people. Until now, that suspicion has been unsupported by evidence. But next week, at Fairness, Accountability and Transparency, a conference in New York, Joy Buolamwini of the Massachusetts Institute of Technology will     5     work which suggests it is true.

Ms Buolamwini and her colleague Timnit Gebru looked at three sex-recognition systems, those of IBM, Microsoft and Facebook. They tested these on a set of 1,270 photographs of parliamentarians(国会议员) from around the world and found that all three classified lighter faces more     6     than darker ones. All also classified males more accurately than females. IBM’s algorithm, for example, got light male faces wrong just 0.3% of the time. That compared with 34.7% of the time for dark female faces. The other two systems had similar gulfs in their performances. Probably, this bias     7     from the sets of data the firms concerned used to train their software. Ms Buolamwini and Ms Gebru could not, however, test this because those data sets are     8     guarded.

IBM has     9     quickly. It said it had retrained its system on a new data set for the past year, and that this had greatly improved its accuracy . When testing the new system on an updated version of the set of politicians Ms Buolamwini and Ms Gebru had used, the firm said it now achieved an error rate of 3.46% on dark-skinned female faces—a tenth of that the two researchers had found using the     10     system. For light-skinned males the error rate also fell, to 0.25%.