The Marketing Abuse of Artificial Intelligence
For years, “artificial intelligence” meant processing information like the human brain. Not any more.
Isaac Asimov’s book, “I, Robot,” began as a series of short stories published in the 1940s. His robots, with their positronic brains, thought more or less like humans. Dr. Susan Calvin was a main character in the stories. She was a robopsychologist.
Decades later, artificial intelligence, or “AI,” has become a marketing term, and it means nothing more than “a cool way to process information using our nifty computer program.” Marketing specialists co-opted the term when they needed a way to describe two different things: machine learning, and the processing of big data.
Big data created a specific new problem. We had plenty of information about what people were purchasing, and plenty of information about the traits of the purchasers, but we didn’t have a way to put all of that information together in meaningful ways. Spreadsheets didn’t help, because they were too big to be read by one person. Databases didn’t help, because the tables that held the data were similarly too big. The solution, from a programmer’s perspective, was obvious: write software to parse and combine all of that information about the purchasers and what they purchased, and see if any useful connections jump out.
The computers – the machines – came up with some interesting connections, and they told us about them. We learned what time of day a particular advertisement resulted in the most sales. We learned that simple changes, like changing the word “rapid” to “speedy,” would cause more clicks. Using machines to learn about big data wasn’t just interesting, it was downright profitable.
And the marketers sold these new programs, and the powerful computers that ran them, as “artificial intelligence.” The thing was, they didn’t resemble human thought processing any more closely than the programs that came before them. Oh, these powerful new programs could do things that humans couldn’t do – but, by definition, that means that they’re not human, or even “human-like.”
Artificial intelligence is now used to mean, “capable of doing things a person can’t,” instead of, “thinking like a person thinks.” Which is silly, because computers have always been able to do things a person can’t. All the way back in the mid-1990s, my IBM 286 running MS-DOS could multiply two four digit numbers faster than I could. But we didn’t call that capability artificial intelligence. That would be like calling a car an artificial horse. We measure an engine’s power with a standardized unit of measure called horsepower, but we don’t expect cars to be horse-like. But Isaac Asimov did expect the artificial intelligence in a positronic brain to be human-like. We all did.
To remedy the problem of the evolved meaning of artificial intelligence, I propose breaking the term into two sub-categories.
Artificial Machine Intelligence (AMI)
Artificial Machine Intelligence, or AMI, is analogous to cars and horsepower. The computer can process larger quantities of information, and it can do it faster, than a person. We can even develop some standards around how much a typical person can remember, and how quickly an average person can solve a fixed-length problem, and describe a computer and its software as having a thousand peoplepower (I’m joking).
Artificial Human Intelligence (AHI)
Artificial Human Intelligence, or AHI, is the older, traditional meaning of artificial intelligence. This is the type of intelligence that Asimov was describing as a characteristic of the positronic brain when he wrote the Three Laws of Robotics. This type of artificial intelligence can be used for decision making that involves non-mathematical values, such as ethical and moral problems. Moral values will play a role in the decisions made by driverless cars. For example, to avoid a collision, should the driverless car turn toward the building on the left, or the group of children playing on the right? What if there are people in both escape paths? Should the car prefer to collide with another car, on the chance that the car will shield its occupants?
Summary
Artificial intelligence used to refer to computer processing that mimicked human thought processes, and included human value propositions in its decision making. Today, this definition of artificial intelligence is still valid, but it’s no longer the only meaning. The term is also used to describe information processing that is beyond human capability: “more powerful than human intelligence.” Because artificial intelligence is a term that is now used in two different ways, it’s time to break the concept into sub-categories: Artificial Machine Intelligence (AMI) and Artificial Human Intelligence (AHI). Marketers will be better able to communicate what their systems can do, and purchasers of modern computing systems will be better able to understand what they need, and what they’re paying for.