Everything is powered by artificial intelligence these days, if you took stock of manufacturers’ claims and general hype. Don’t, because there are a lot of misconceptions and misleading out there.
That question is a fundamental problem with coverage of AI. Many pundits, businesspeople, and journalists use the phrase without understanding its breadth and age.
Researchers began to lay the groundwork in the 1950s. By the 1970s, it was a regular subject of investigation by university computer science departments and large corporations.
AI isn’t a single concept but more like a wide space of areas that have broken out from different types of capabilities. That includes rule-based expert systems that determine what action to take based on predetermined criteria; natural language processing to let the machine respond to free spoken or types input; speech processing; vision processing; and machine learning that can pick up on patterns from previous examples of decisions and apply them going forward.
Forget about general intelligence that can do anything. AI in commercial use is almost inevitably of the so-called narrow type that solves a specific type of task. Even something like IBM’s Watson is a collection of different technologies that address different functions. In general, though, a piece of software that will do something like identify a license plate number on a car
What gets called AI?
Here is where things get complicated. Too many companies see AI as a sellable feature and work with marketing people who want to claim capabilities that may be irrelevant and, even if not, absent from the product. I’ve personally seen this happen, where technical information on some software didn’t mention AI but a marketing person insisted on claiming that it existed.
This is similar to what happened with the “paperless office,” “total quality management,” “supply chain management,” “mobile,” and “Internet of Things.” It leaves the audience to untangle claims that are often spurious.
Questions to ask
When trying to untangle one of these battles of marketing claims, here are some questions that can come in handy when talking to marketing and product people.
- How long as the product been in existence?
- What features existed before the addition of AI? What features didn’t?
- When was AI added?
- Exactly what type of AI technology was included? How was it implemented?
- Why was AI used?
- May I speak with someone on the technical staff who was involved with the implementation?
- Why specifically should users care about AI in this product? What does it add to their experiences?