What do we call the fusion of automation and machine learning?
We are witnessing the marriage of automation and machine learning. Automation is becoming more intelligent, thanks to machine learning. Boston Dynamics, for example, used to code its dog-like, ANYmal robots by hand but recently switched to machine learning and saw a 25 percent jump in the running speed of these machines. Going the other direction, Google, Microsoft, and others are now accelerating machine learning by automating the training of its algorithms. With AutoML, they are putting machine learning into the hands of a whole new range of companies and industries.
We’re still at an early stage in the integration of these core technologies. So for now, automation that incorporates machine learning is just really effective automation. And machine learning that is automated is just really efficient machine learning. But going forward, I think we are going to see more and more systems that transcend these individual categories of technology. We will see factory robotics that are integrally linked to machine learning systems for inventory management, equipment yield management, maintenance, and so on. We’ll see automated storefront kiosks at McDonald’s run by powerful machine learning systems. I mean, who am I kidding — we’re already seeing this.
The question I am now pondering is what is the best way to talk about this fusion of machine learning and automation. Some experts in this field use the term “Intelligent Autonomous Systems.” It’s a mouthful. Also, I’m not wild about “autonomous” because it suggests a kind of free will or volition that just isn’t the reality right now.
While musing over this question yesterday, I toyed with calling these systems “transformers.” Why? Well, because automation and machine learning both transform things—machine learning transforms data into models and automation transforms models into action. I even ran the idea by people on Reddit, where one kind soul wisely counseled me against neologisms (especially those that remind people of giant Japanese robots).
So, I’m left scratching my head. When we talk about these increasingly hybrid systems, what should we call them? One approach is to just be literal with something like “intelligent automated systems.” Simple is good, I suppose, even if it sounds a little dry.
I’m genuinely asking for some guidance here.
7 thoughts on “When Automation Fuses with Machine Learning, What Should We Call It?”
Adding a few pointers from comments on social media:
The consulting industry seems to be homing in on “intelligent automation” as the term for describing this fusion of machine learning and automation. Examples:
Intelligent automation: A new era of innovation
Intelligent Automation: Definition and Examples
I like your simple Intelligent Automated System, which will quickly become IAS. I used to not know what AI was, and now that’s a common “name,” moreso than artificial intelligence. I did learn a new word from you, “neologism,” and examples used were malware, webinar, netroots, and blogosphere. In the same way, either you or someone will create a new word that soon we will all use, as if that word had always been. Maybe “tweet” it! Whoever thought that would be a word we all use now!
Hi Bill. Good to hear from you.
I must confess that I am a sucker for neologisms. They always pique my interest. But then I’m constantly looking for what’s new. Because of this bias, I have to listen carefully when people counsel me to rely on already established terminology. The problem is that I’m dealing with lots of things that are still in flux and am trying to highlight what I see as revolutionary. So I’m constantly being pulled to coin new phrases. Thanks for dropping by.
Interesting question, not least, because it made me think about something I don’t think a lot about and that’s naming what is being formed out of ML and automation and whatever we call AI these days (which is not true AI at all). Why not call them “Smart Systems” as all ‘smart’ anythings have a degree of automation married to some kind of machine learning and the concept is familiar enough to not trigger people or leave them scratching their heads and sufficiently broad to evolve as these systems inevitably will. I hope this helps.
David Amerland thanks for weighing in. I like the term “smart” but someone over on reddit pointed out that “smart” is “essentially any electrical application with a microchip.” Maybe that’s overly technical in interpretation. I’m not sure. Still mulling this around a bit and figured I’d reach out and get some broader perspective. Thanks for jumping in.
Hey, on a total side note, I just switched over to Disqus for commenting after I recall you mentioning that you liked it on your site. So far, so good. One issue that seems to be occurring is that comments aren’t showing up in the AMP view that people get when they look at the site via Twitter and other services. Are you running into that problem?
My two cents.
Self determining systems / self reliant systems
I like the sound of that, Sowmyan. I just don’t know if I see them as really self-determining though. I’m kind of a stickler for the question of where the ultimate volition is that drives these systems. And right now at least it’s still very clearly us. Don’t you think?