When considering how to draw the line between whether an application is AI or not, I’m tempted to paraphrase U.S. Supreme Court Justice Potter Stewart:
I shall not today attempt further to define the kinds of applications I understand to be embraced within that shorthand description “artificial intelligence” , and perhaps I could never succeed in intelligibly doing so. But I know it when I see it.
A machine developing its own concept of what a “cat” is and learning to detect it in videos feels like AI to me. Although the more I read about it and understand it the more it just feels like a clever use of deep learning, creating a mathematical construct that is fit for purpose. It’s like magic: when the trick is revealed I can still be impressed, but it doesn’t feel like magic anymore.
Here’s a test: 50 years from now if someone looks back at this, do you think they’ll still call it artificial intelligence? Or by then will it be so intuitive and well understood – a trusted black box – that we’ll just call it automation?
Justice Stewart’s clever turn of phrase implies that on some issues it’s not possible to draw strict lines, the lines may move over time, and it’s in the eye of the beholder. To me, all of these apply to AI.
Think about “automation” for a moment as a contrast. Automation is simply using machines to do what humans would otherwise do. In the workplace it is associated with boring (but profitable) tasks such as scanning RF codes or macro scripts. Those examples clearly aren’t AI – they are very understandable. But they are very effective and have a much greater chance of changing (or stealing) your job than an AI within your working lifetime.
So there is a gray line between AI and automation that is defined by expectations. If a computer is doing amazing things one never thought it could do and you don’t even get how it’s doing it, that’s AI. If it’s doing non-amazing things that you understand (probably based on rules), it’s more in the realm of automation. If the computer is doing something a human would otherwise do by amazing means, it’s both AI and automation!
So what does the “AI” label mean? If I’m an exec and my goal is improving sales, do I care which side of the AI/automation line a system falls on? If my job is finding sales prospects and I lose it to a machine, do I really care whether it was done using machine learning in a black box performing unfathomable AI magic or automation in a CRM system? No. I just want to improve sales.
Don’t get hung up on whether a system counts as “AI” or not as long as it serves a useful purpose. If you are a CIO trying to make sure you’re always exploring the latest technology, then don’t trust “AI” labels on products and dig beneath the surface to see how the trick is being done – it still might be a useful trick.
Read Complimentary Relevant Research
100 Data and Analytics Predictions Through 2021
Over the next few years, data and analytics programs will become even more mission-critical throughout the business and across industries....
View Relevant Webinars
Digital Business Architecture: From Strategy to Guiding Execution
New techniques have emerged to help CIOs and EA practitioners leverage business architecture to guide investment and execution decisions,...
Comments or opinions expressed on this blog are those of the individual contributors only, and do not necessarily represent the views of Gartner, Inc. or its management. Readers may copy and redistribute blog postings on other blogs, or otherwise for private, non-commercial or journalistic purposes, with attribution to Gartner. This content may not be used for any other purposes in any other formats or media. The content on this blog is provided on an "as-is" basis. Gartner shall not be liable for any damages whatsoever arising out of the content or use of this blog.