Why I’m Not Buying Your Product’s AI Claims

Published Mar 23, 2018

Looking for some entertainment?

Type “AI” into Google News. Here’s a sample of the results from a recent peek into this particularly ridiculous corner of the web:

  • “A new test could tell us whether AI has common sense” (Most AI doesn’t even have sense yet, much less that of the common variety.)
  • “More Human Than Human makes the state of AI look ironically grim” (Huh?)
  • “Get ready, this year your next job interview may be with an AI robot” (I doubt it, but if so it would probably be better than many of the human-led interviews I’ve been part of.)
  • “Can an AI-powered bot help parents raise better humans?” (I have two little kids so I’m all for it!)

High on Hype

Yes, we find ourselves somewhere in the midst of the AI hype cycle, and as with any such cycle, there are both good and bad outcomes. The good? AI is giving us genuine promise and progress towards better outcomes for business and society. There are exciting uses of the technology to improve healthcare in the areas of diagnosis and treatment, enhance public safety through advancements such as autonomous vehicles, and democratize the knowledge required to participate in complex ecosystems such as financial markets.  

But with the good comes the bad. The clickbait noted above is undoubtedly part of the bad, despite being mildly entertaining. However, you know what really grinds my gears? It’s the pervasive co-opting of the AI term in product marketing. Like cloud-washing of yore, AI is suddenly everything and nothing at the same time.  Before we consider some examples, let’s first look at the Merriam-Webster definition of artificial intelligence:

  1. a branch of computer science dealing with the simulation of intelligent behavior in computers
  2. the capability of a machine to imitate intelligent human behavior

Admittedly both definitions leave room for interpretation, but let’s agree that the mimicking of human intelligence and all the complex cognitive discernment that comes with that is a core part of AI.

So for the sake of my own sanity, I’m building a BS filter that I apply to the AI marketing craze.

Can you really say you have AI?

  • If a product simply accepts voice input, and then uses that input as a search term – not AI. Yes, I understand that technically natural language processing (NLP) is a subdomain of AI, but it’s widespread and rapidly commoditizing technology. It’s a more convenient way to do search, but not a monumental advancement of AI.
  • If a product suggests the optimal time for of delivery of an email, in-app message, or notification based on rudimentary data points like prior timeframes of engagement, typical actions that precede engagement, or the like – not AI. That’s rudimentary data analysis.
  • If a product is a chatbot, it’s very likely not AI. Most chatbots are simple AI interfaces that accept user input, and then apply a rules engine to determine a response. Some may “learn” over time and do some degree of refinement, but usually it’s just about adding simple rules that get added to the evaluation chain. Interesting perhaps, but not AI.

Maybe you think I’m a bit cynical, but I’m going to keep building up my filter. Let’s agree to reserve the use of AI for those things that are pushing beyond input and search, data analysis, and complex rules engines. This is just doing work faster. Imitating human intelligence means dealing in the gray areas that our brains are adept at such as integrating human emotion into decision making, manipulating to produce desired behaviors, and more. I’m moving the AI bar up a notch and encourage you to do the same.