show image

Racist bots, imaginary flu pandemics: The downsides of Big Data

Big Data is currently touted as one of the biggest technology advances around but it’s not infallible, as an article on GCN points out.

Analysing data is leading to a number of false positives, it appears – including recent predictions of flu outbreaks when Google searches for “flu” peaked but weren’t backed up by incidents.

This was referred to in a report on public health in the US. The other example highlighted by GCN was Microsoft‘s recent foray into artificial intelligence, with the Tay avatar on Twitter, which picked up what was being said around it and ended up (for example) coming out with an affection for Hitler and a hatred for feminists. (Hint: if a few winder-uppers start to feed your artificial intelligence machine poor data, it will rapidly start to come out with junk.)

The obvious answer is to keep awaiting better analytics and to take Big Data, no matter how persuasive, in context: people concerning themselves about flu because it’s winter might well not translate into people catching the virus, for example.

Applications of common sense, anyone?