We’ve been tinkering with a new product that uses data collection and machine learning to redefine how we read news, and it got me thinking. Can we teach computers not just to find, index and filter stories, but to write them?

Will there come a moment when machines can create what we read? Would financial markets be more rational if computers reported on the data instead of humans? It’s a frequently proven fact that reporters are apt to provide color and skew a story one way or another (intentionally or not); or, as often the case, people simply make errors. Sure, the news might be less entertaining, but could computers make it more reliable? And even harder to imagine, could a computer become the next Dostoevsky or Joyce or Shakespeare?

For news, there are of course a number of challenges to solve, for example:

  • Assignment; how a computer knows what’s a story, what isn’t and where to look
  • Sources; which are reliable or not
  • Fact checking, and of course
  • How to pull facts, and “write” an article in natural language

    • All monumental tasks, and all requiring an artful human touch to get computers thinking straight — which data sets to target, fact checking methodology, algorithms for language disambiguation etc. — but seemingly none insurmountable.

    After a quick Google search (All Hail the Great Googs!) I found a couple of curious and amazing products that begin to suggest feasibility:

  • A gorgeous iPad app created by David Benqué that takes aim at computer generated fiction called The Infinite Adventure Machine, which launches this week during Paris Design Week. David’s app uses the 31 functions of folktales identified by the philosopher Vladimir Propp, to help users imagine their own stories. As David says on his blog “TIAM aims to question the limitations and implications of attempts at programming language and narrative.”

  • A company called Narrative Science that claims to have in fact solved the problem for news, and uses computers to generate long and short from articles from structured data. Started by Kris Hammond and Larry Birnbaum, co-directors of the Intelligent Information Laboratory at Northwestern University, the company has been picking-up steam and ink since landing $6M in investment around June of this year, with articles written about them by the New York Times and PC Magazine.

    • Methinks we are on our way.

    By Jordan