Skeptical of AI? You Should Be

Artificial Intelligence (AI) is a very broad topic, and the superset for concepts like machine learning. The hype around artificial intelligence has become extraordinarily deafening. Developers, vendors, end users, and the analyst community are routinely making assumption about AI that tests the arrogance and ignorance of everyone involved.

What I’ve found is that, outside a small set of seasoned university professors, select vendors, and patent holders for artificial intelligence algorithms, no one is really an expert in this emerging field. AI is a buzzword carelessly bandied about to evoke “cutting-edge”, but the reality of the claims often belong in science fiction.

There are broad assumptions on what is possible with AI, what actually works, and, most importantly, what the outcomes will actually look like. AI has a place in modern computing and next-generation technologies, but it is A. & I. (arrogance and ignorance)—or just plain folly—to think it can solve all the problems for every organization. That is just overhype and marketing gone off-the-rails. Consider the following:

• On March 5, 2019, The Verge reported that, based on an MMC Ventures Whitepaper, “Forty percent of AI Startups in Europe don’t really use AI”. The paper itself is quite illuminating and, if you are looking to shed your own A. & I. misperceptions, it is certainly a good overview and place to start.
• Kolibree developed a product called “Ara” that is marketed as the first toothbrush with AI to help keep your teeth clean. This is perhaps the most ridiculous case of overzealous marketing taking advantage of consumer hype.
• According to Roger Schank, a cognitive computing expert, capabilities of IBM’s Watson that are claimed to be AI, are, in fact, not truly AI. While this may come as a shock to many readers, Schank’s blog on the topic addresses the bias in marketing versus the true coding and capabilities of Watson.
• Whether the claims of AI are to detect fake news, or to stop fraudulent insurance claims or theft, the hype needs to be tempered by the limitations of the actual product and use cases they solve. For example, analyzing text for patterns using machine learning is only a subset of AI that really just amounts to pattern matching and not actually intelligent behavior.
• And, if you think any of this AI stuff is funny, consider how this CEO feels. There is probably nothing like getting ridiculed by white hats at a major conference about over clams of how revolutionary your AI is compared to the rest of the industry. I won’t spoil this one, you will have to check out this PCMag article to see what I mean. Now, it’s your turn to be the judge of AI or A. & I. on this one.

If you believe all the artificial intelligence claims presented by vendors and auctioneers, you are falling into the A. & I. trap. There are no laws, governance, or even basic oversight to stop some of these outrageous claims. Anyone has the right to call their tech “AI”, just like we all have the right to call them out on it (or swallow the claim—hook, line, and sinker).

Therefore, if you are genuinely seeking out AI expertise for your next-generation technology (which, by the way, is another widely misused and often truth-stretching term), I strongly recommend asking the tough questions about the algorithms, patents, and underlying technology that forms the basis for their claims. If the AI is legitimate, they should be able to provide these proofs—never settle for a Jedi hand wave. Avoid the marketing hype and find professionals that understand the difference between basic mathematical statistics and those applications that can truly imitate human learning and behavior. There is a massive chasm between the two, and true AI does not necessarily fit in every application and business use case.

For a deeper take on this subject, you can check out my podcast, Is Artificial Intelligence a Massive Con? for the New Futurist here.