AI is my Co-Pilot: Should we feel threatened?

Back around 2010 or 2011 my youngest son Michael was 12 years old.  I was telling him about the work around autonomous driving.  I was doing some research then around IoT (Internet of Things) and a larger presentation I would put together on an overview of Artificial Intelligence, Machine Learning and Deep Learning.  Back then it seemed like we were right around the corner from self-driving cars, allowing us to relax, study or work while being driven (transported) to our desired location.

My son, who was, even at twelve, greatly looking forward to driving asked if he’d be able to drive when he reached 16.  I assured him, he’d probably get in a few good years.  Well, he’s 25 and has been driving since he was 15 and half (with the learner’s permit).  He lives in Southern California and probably, at times, wishes I (or the experts) were correct on self-driving cars. No doubt technology has advanced greatly but I don’t believe any of us are ready to just give the car over to the technology just yet.

To minimize the problem there are only 4 options: turn right, turn left, speed up or slow down.  Of course, slowing down can mean stomping on the brakes, while swerving to the right.  Still, I contend there really are only 4 things to do.  The difficulty is that there is a lot of space, physics, motion and stuff around the car, not to mention signage that affects the 4 options.  Plus, you have the reliability of the sensors and interpretation of the results.  So fully autonomous, self-driving cars are still a little way into the future.

A new concept that is making a lot of headway is something called copilot.  We are seeing a lot of interesting usages for an AI copilot, especially around coding but in various areas the idea is blossoming. Microsoft with development partner OpenAI launched GitHub Copilot in 2021.  In fact, I’d say, Microsoft has been leading the copilot trend. They updated their copilot offerings at the Build conference this past month. Additionally, they are embedding Copilot within their Office 365 products. It’s now possible for Copilot to record every action that you take on your PC.  Good news; maybe bad news. Overall, I see the term ‘copilot’ as much less threatening than the vision of AI overlords destroying the human race.

I have long believed in AI was augmenting, not replacing humans.  By way of example, I like to use a chess example. There are twice as many Grand Masters now than when Deep Blue (IBM) beat Kasparov in 1997.  Kasparov pioneered man plus machine matches called Centaur (man plus machine). In Freestyle battles pure AI engines won 42 games while Centaurs won 53 games. The current best chess player is “Intagrand” a team of several humans and several chess programs.  Also, Magnus Carlsen, the top ranked chess player, trained with AI and has the highest human Grand Master rating of all time.  I believe in retrospect; one could term the Centaur as a chess player plus copilot.

Keith Moore (Distinguished Technologist) and I have been discussing potential AI use cases around NonStop. I think most are aware of my interest in using AI to thwart human trafficking within the financial services area utilizing money-laundering patterns but another idea which fits the copilot model better would be to feed NonStop manuals, softdocs, GNSC logs and anything else we could find into a LLM (Large Language Model) to create a NonStop copilot which could hopefully answer system management, programming and database questions that might arise when running a NonStop.

It could be a partial answer to managed services where a copilot could be created to assist with most requirements.  It’s still an idea in incubation but who knows, perhaps it will arrive arrive before self-driving cars 😊