The Financial Times has released a bot, She Said He Said, this month to encourage its journalists to feature more female experts in articles. The bot was developed after analysis revealed only 21% of quoted sources in Financial Times’ content are women. The bot uses an algorithm which analyzes first names and pronouns in a piece to predict a source’s gender. Data is then shared with section editors who can ensure a fair balance. The media outlet has said that in the future, writers will be alerted in real-time if they aren’t featuring a diverse range of sources.

There’s been enough coverage during 2018 to make clear that algorithms are only as good as the data they run on. Let AI loose on societies that are already biased and discriminatory, and you’ll get results that are equally biased and discriminatory (as Amazon found when it inadvertently created an AI recruiting tool that discriminated against female candidates).

However, this isn’t the whole story! At TrendWatching, we celebrate innovators that are making the world a better place. And just as ‘bad’ AI amplifies existing bias, ‘good’ AI will expose and correct bias. Indeed, that’s the topic of our SUPERHUMAN RESOURCES trend, one of our just-published 5 Trends for 2019!

Here are two quick thoughts if you’re deep in designing your AI strategy for 2019:

- Look in the mirror. Before you try and tackle unfairness in others, take a cue from the Financial Times and give yourself a good, hard look in the mirror first. Where might you be (consciously or subconsciously) be producing biased outcomes? Could AI help you mine your data in new ways in order to check? Or could it offer novel ways to help you avoid and eliminate bias going forward?

- Empower people. Winning organizations won’t use AI to eliminate jobs. They’ll deploy AI in ways that empowers their staff to deliver better outcomes. The Financial Times found that more representative content attracted a more diverse audience. Kronos and IBM created a chatbot career coach for hourly workers. How can you create an AI that benefits humans?