Artificial Intelligence (AI) and robots were among the hot technologies at the recent CES, and while I understand the fascination with such a potentially revolutionary technology, AI’s vulnerability to hype is just so much greater than it is for other technologies. The reality is often more mundane. At this stage at least, the robots sometimes seem less likely to take over than they are to just… disappoint us.
I recently found a link to a document that summarizes some results of AI/robotics trials. A few highlights:
• Robotic arm trained to slide a block to a target position on a table achieves the goal by moving the table itself
• Game-playing agent accrues points by falsely inserting its name as the author of high-value items
• Creatures bred for speed grow really tall and generate high velocities by falling over
• In an artificial life simulation where survival required energy but giving birth had no energy cost, one species evolved a sedentary lifestyle that consisted mostly of mating to produce children who could be eaten (or used as mates to produce more edible children)
• Simulated pancake-making robot learned to throw the pancake as high in the air as possible to maximize time away from the ground
• Agent kills itself at the end of level 1 to avoid losing in level 2
• Self-driving car rewarded for speed learns to spin in circles
• Agent pauses the game indefinitely to avoid losing
I find some of these almost endearing: Poor robots! They’re neurotic and suffer from imposter syndrome just like the rest of us. They’d rather die than suffer the pain of failure. Or, they’re like smart but unruly children, petulantly literal when forced to follow instructions.
When it comes to enterprise communications, there’s reason to at least be prepared for AI causing as many problems as it solves, especially when it comes to management systems. At Enterprise Connect 2019, IP networking guru Terry Slattery, of NetCraftsmen, will present a session entitled, “Will Automation Break My Network Faster Than I Can Fix It?” From the abstract for that session: “Some enterprise managers worry that automation could actually exacerbate problems and cause them to spread more quickly than they could when configuration was more manual.” Terry will offer advice on how to understand when your network might be vulnerable to such problems, and how to head them off.
If you’re more inclined to look on the bright side, you’ll want to check out this year’s Enterprise Connect Innovation Showcase. The topic for this year’s edition of our annual spotlight on innovation is “Actionable Analytics.” The scenarios that showcase leader Dave Michels, of TalkingPointz, proposes as examples of Actionable Analytics include:
• Improvements via analysis of sentiment, productivity, retention, and more. These solutions can be used on internal and/or external communications.
• Capabilities that enable users outside the field of statistics and analytics to extract predictive and prescriptive insights from communications data and workflow.
Applying AI to communications data at the back end, to help you better manage your environment, is certainly among the major, though less touted, areas of potential for AI. Taken to extremes, of course, AI-driven analysis of workers’ communications data carries frightening implications when it comes to intrusions on employee privacy. So the “actionable” part of Actionable Analytics will need a strong governance component.
AI is likely to be a hot topic for a long time to come, and I expect it’ll be a foundation of many of our conversations at Enterprise Connect Orlando, which takes place at the Gaylord Palms hotel the week of March 18. We’d love to have you there to join the conversation.