Is AI a CxOs dream or an Iceberg about to be hit
Blog: Brian's Blog - Enterprise Strategy, Architecture and Management
This morning’s ponderings. This latest iteration of AI, Machine Learning and the like seems much like the hype of the 70s AI with the exception of it operates faster. But does that mean it operates better? If one follow the middle page news, AI still needs a lot of work. Yes for optimization it works, but standard OR worked too. What is interesting are the side effects that make the news: An AI Blog app learns to become racist, a app turns off critical systems to conserve power, and other no so minor effects that have not seen the light of day due to the sensitivity of impact.
Bugs and side effects are normal in any new and even old technologies. A any student of dynamic systems will tell you that side effects that you didn’t planned for are common. These can be from your decisions to change a complex system or someone else’s decisions to do the same. The interdependencies –some hidden from our awareness—in systems creates this complex web of unintended consequences.
If I was a CEO would I trust AI to be making major decisions for me? Given SOX and other standards place accountability on my shoulders I can’t push it off to “my AI system made the decision”.
This opens up two other and interdependent courses of advanced research for us at Elyon Strategies: Governance and Dynamic Systems. While we have existing practices that utilize excellent methods for both I’m seeing both potential breakthroughs linking the two as well as a new area of concern when it comes to AI.