future foresight analytics

More thoughts on the future of foresight

Eric Garland Intelligence Analysis Leave a Comment

Today I find myself in San Francisco, CA for JPK Group Summits’ Business Forecasting and Analytics Summit. My keynote this morning was entitled The Future of Forecasting (slides below), and in it I explore the balance between our innate desires to predict the future, emerging technologies, and human behavior. It’s not the first time I’ve given this speech, but the challenging and sophisticated professionals here today definitely made me rethink how the future of foresight will unfold.

If there’s one technology that will impact the future of foresight, it’s artificial intelligence or machine learning. When working on Massive Data™ sets (I don’t have a trademark on this, but I feel like I should try) forecasting will benefit from algorithms able to derive insights from unusual sources. Right now, we use computers mainly to answer questions for us. The eventual use of artificial intelligence will be its ability to use our past queries and ask new questions that hadn’t even yet occurred to us.

I came to a new realization about this when a young researcher on machine learning described it as improving from mistakes. The more data you feed into it, the more questions you ask it, and the more you flag as wrong or unhelpful improves the software over time. This is fascinating, because we’re describing a dynamic that matches Nassim Nicholas Taleb’s concept of antifragility. According to Taleb, fragile things break, resilient things can take a beating, but antifragile things get better with misuse. For machine learning and analytics, the more often it’s wrong, the better it serves you in the future – perfectly antifragile.

There’s a lot left to research in this field, and forecasting and analytics strikes me that a field that won’t be replaced by machines any time soon, a fundamentally human activity.