Wednesday, March 10, 2010

The robot reporter: false hope or cautionary tale

[Guest post by karenatcmg.]

Could a robot do your job? A lab on the Northwestern University campus near Chicago already has a prototype artificial intelligence program that can report on baseball games and will soon develop programs to cover football and basketball.

“It’s the dream of every managing editor: a reporter who is cheap, works fast and isn’t moody,” writes Le Monde correspondent Yves Eudes, who recently visited the lab.

The “Stats Monkey,” as it is known, will also soon turn its digital brain to reporting on financial markets.

The inventors of "The Machine," which is the byline on the prototype reports, gush that the product reads the same as AP wire copy. But without typos! (The machines – no kidding – are able to reproduce the same pat sentences that every reporter rushed for time uses to express the same range of outcomes). The benevolent inventors say they are not out to replace humans with machines and put them out of work. Of course not. They say their program could relieve journalists of the boring, repetitive work to allow them time for the noble part of the calling: field reports, investigations and analysis. Besides, they add, the purpose of the program is to report on minor league and varsity games and the stock market performance of smaller companies, which don’t currently get coverage in the mainstream media.

By the way, down the hall from the Stats Monkey is “News at Seven,” an AI project that puts together an online newscast, complete with animated male and female co-anchors (Zoe and George!), based on the preferences of the viewer. It gathers and summarizes relevant reports from a series of news sites and then “voices” them.

Frankly, the AI machines seem a day late and a dollar short. What can they do that we don’t already do, compiling and relaying data within seconds on a wire desk or producing newscasts according to a formula, primarily using secondary sources?

Besides, if news organizations aren’t devoting resources to high-school leagues and small business stock performance today, why would they invest in machines to do it tomorrow?

And who is to say that, once they had them, our employers would rehumanize our work? After all, if they wanted the fulsome product of human brains – if that’s where they saw the quick buck – they could have it already,couldn’t they?

Obviously the machines will solve nothing. It’s not even clear that they will be more productive (ie. produce more at less cost) or that they will create fewer headaches than regular human journalists. I mean who has NOT worked with temperamental IT systems?

But news of them alone is perhaps enough to scare us into working that much faster, with that much more accuracy, to avoid being replaced by a reserve army of computer chips.

1 comment:

  1. Buuuurilliant. I like the analysis .
    Gathering news and information is evolving at the speed of the internet.
    But the good news is that humans need humans. Especially someone to filter all that news.
    The flood from YouTube is improved by human filtering, and that's what reporters and editors do.


Thanks for commenting! Just to let you know, I will remove comments that are unrelated to the post or include personal attacks on anyone.