Trust & AI in the Supply Chain

Fake it 'till you make it might be the traditional wisdom in Silicon Valley, but it may not be the most effective strategy for AI. As I have thought since I wrote my MBA thesis (you can imagine how long ago that was...), a critical risk in computer based supply chain management is that the decisions the computer makes are not credible to the users, so they ignore it and do what they think is right, based on their limited experience and bias-riddled thinking.

I am not sure which is better, but if we could get the two to work together, wouldn't it be great? That conundrum is at the core of AI's "explain-ability crisis," described in this article in Fortune: https://links.newsletter.fortune.com/e/evib?_t=5c2d888702774d17aa3d0350287b6d73&_m=25475478e582423b82e298def81fa6c7&_e=LNvQ5DpgIjTR9Kni-ExrQO8eG0NR-mq-sztlqhs2aXWWnAHj0vtezKGsoAAwpG50bsEAvKwxNBVxkU4pXu7frM4xrYDCYMrCs02V0IXxcXxG8AWRN2kc0XQJ-cR7pA41.

How on earth did the computer come to that conclusion? This is not a new problem, but early solutions outlined in the article suggest we are still a long way from a viable solution.

While it is a matter of life and death in Ukraine debunking deep-fakes and directing loitering lethal weapons, it is also foundational to the use of AI in critical operations. AI can be used today in areas of limited impact, like chat bots and spend coding, but what about re-routing a shipment that is vital to making the plan? Can you trust a computer you do not really understand?

I wouldn't.

And that is why this is an important issue to crack. What do you think?