It’s time to rename Artifical Intelligence

Although AI is now a common part of our high tech lexicon, I have a problem with its semantical origins. As the European Union and national governments undertake complex debates about the legal, political, and ethical status of things containing AI, I think the source of this acronym – Artificial Intelligence – is fundamentally misleading.

First, here are the common definitions of these two words:

Intelligence = “the ability to acquire and apply knowledge and skills”
Artificial = “made or produced by human beings, rather than occurring naturally.”

According to these definitions, if intelligence is produced by human beings, it must be artificial. Thus, distinguishing between Human Intelligence and Artificial Intelligence is misleading. Both are a product of the human mind.

Perhaps the problem lies with the above definition of artificial. It separates human beings from the natural world, i.e. if something is made by humans it doesn’t occur naturally. But humans are a part of the natural world. We are flesh and blood animals, subject to all the same laws of nature that apply to lions, bees, and trees.

On a practical, purely functional level, ‘artificial’ is easy to understand. We all know the difference between a real hand and an artificial hand, real sugar and artificial sugar, real flowers and artificial flowers, etc. These are all physical objects that can be seen and touched and tasted. Common sense tells us the difference between real and artificial. But intelligence is an abstract concept.

Juval Noah Harari says that man is distinguished from other animals because we make up words for things that don’t exist in nature. These are abstract concepts. We have no trouble agreeing on the meaning of words that refer to palpable physical objects: a tree, a rock, a lion, a bird. But we also have words that apply to things you cannot see, touch, smell, or hear: justice, freedom, empathy, and yes, intelligence.

Language is only necessary between two or more people. An isolated, solipsistic hermit doesn’t need words to deal with his physical environment. However, when two or more people choose to live, work, hunt, and survive together, they need words to communicate. Words for physical objects are easy to agree on. But when it comes to abstract concepts, like justice or love, each person develops an image and meaning in their own head and tries to match it with what another person understands by those words. But we cannot get into another person’s head. We assume we share an understanding and maybe come close, but we never know for sure. The concept of love in one person’s head can be very different from that in another’s.

Over time we have developed an elaborate and complex system for defining words, based on common usage, accepted definitions, and socially agreed upon standards. But none of these are absolute. None of it exists outside the individual human mind, or the collective human mind we call society. Abstract entities like love, justice, freedom, and beauty don’t exist in the natural world. Other animals can’t see them. They exist only in the minds of humans. And because we each understand these abstractions differently, we debate their meanings. We have even gone to war over differences in our definitions of these concepts. (Religious schisms, which have led to the slaughter of heretics on both sides, often arise over the definition of one word.)

Intelligence too is an abstract concept. We can argue endlessly about what it means and how we recognize and define it. But it’s all in our heads. It’s human-made. And according to the above definition, anything that is human-made, and doesn’t occur in the ‘natural’ world is artificial.

Something is wrong here. I think it’s the definition of ‘artificial’. Everything, including humans and what they produce, are a part of the natural world. It can’t be otherwise. (Unless, of course, you want to bring in the supernatural, but that’s another topic.) But if you discard the word artificial and narrow it down to “made by humans”, you still have a problem. What we think of as intelligence – the aquisition and application of knowledge – is a human activity. It’s a human product. And it remains a human product whether you store it in your flesh and blood brain, or program it into an algorithim trapped in a silicon chip.

Either all intelligence is artificial, or none of it is. It’s time to put our minds to rethinking how we describe what we know.

Since AI has become a part of our accepted lexicon, perhaps we can resolve this problem by retaining the familiar acronym, but changing what it refers to. Two possibilities come to mind. Augmented Intelligence, or Autonomous Intelligence.

“Augmented” refers to something that has been made greater in value or size. A computer, silicon chip, or algorithm that can access, retain, and process more data than the average human mind could be called augmented. It knows and works with more.

“Autonomous” is sometimes defined as “having the freedom to act independently.” An algorithm built into a program is designed to perform its functions independently. You create it, kick start it, and it goes off on its own. One can argue that it isn’t totally independent because it is constrained by the program that created it. Nevertheless, once it begins to function, it does so independent of the humans who created it.

Perhaps there are other terms that can be used. I’m sure many will think of them. They may be better or more precise than my suggestions. But I offer them as a starting point for a discussion which I think will be necessary if we hope to get a handle on the legal, practical, or ethical implications and applications of AI.