Arthur C. Clark once wrote, “Any sufficiently advanced technology is indistinguishable from magic.” This observation certainly applies to Artificial Intelligence.  

Unfortunately, there are federal agencies that aren’t quite enthralled with “magic”, and they do require some information on how AI derives its conclusions.  Kind of like your high school math teacher asking you to show your work on that last answer.

Today, we have an accomplished practitioner of AI giving listeners an idea of what understanding AI is. The interview is based on a recent article Patrick Elder wrote called, “Explainable AI.”

The challenge is obvious – AI is based on bringing in massive amounts of data, it could be in the form of words, code, or images.  This is all well and good if you are a high school student and want some help with writing a paper on, for example, Arthur C. Clarke. 

The federal government is challenged with storing sensitive information and not all of it is permitted for collection to render AI effective.

Patrick Elder details three approaches: white, black, and glass box.  The black box approach gives results and humans don’t know how they derive conclusions.  The white box is transparent about how it gets conclusions.  These are both contrasted with a model called the glass box.   

During the interview, Patrick provided examples of explainable AI. If you would like to dig deeper, you can read his article, “Explainable AI: How XAI Puts the End User Back in the Driver’s Seat."

Follow John Gilroy on Twitter @RayGilray

Follow John Gilroy on LinkedIn  https://www.linkedin.com/in/john-gilroy/

Listen to past episodes of Federal Tech Podcast  www.federaltechpodcast.com