Why the Vaunted “Artificial Intelligence Boom” is a Bubble

 

The Goddess of Computing

About three times a month I go to Delaware to avoid sales tax on certain items. When the trip falls on a Saturday, there’s no ‘Cluck & Buck Show’ to keep me awake on the open highway, so I listen to Kim Comando on WILM radio. She’s impressive. So impressive I should probably have included her on my NIMH list of significant female voices. She really knows her stuff about the complicated world of phone and desktop applications, communications, and security products, as well as the broader business world in which they operate. I’ve heard her give would-be entrepreneurs outline business plans for computer-based services that could be started and run from home. Off the top of her head, mind. She likes to help. I’m getting somewhere with this, believe me.

Last time I heard her she did a segment on her show about ways to earn money on your computer without getting a job in an office. She mentioned one opportunity in particular, at Amazon, whereby it’s possible to make a living income doing things a computer can’t do. Huh? You do it and convey the result by computer. What might that be. The example she cited was recognizing what color things are… You can get paid for looking at things Amazon wants to use in AI development projects and identifying colors of things. Because computers can’t do that…

*******************

Here’s what happened since I wrote that intro. I spent a couple of very intense hours writing a post about the real issues associated with developing proficient Artificial Intelligence systems. The result was about 2,000 words long and went deeper into the subject than I’d gone in several other posts about AI. I finished the post and went to grab a couple links to the older essays. When I returned, everything I’d written after the introduction above was gone. I was petty upset about it. Losing work to the anomalies of computer dysfunction is nothing new, and I usually just rewrite what is lost, but I’m taking a different approach this time.

Lately, I’ve been in the mode of trusting the universe more. In small ways and big ways, I’ve been paying closer attention to what seems to occur or pop up randomly, even in my choice of entertainments to watch in background as I work on other stuff. Even losses like this one, which hurt. I really liked what I had written but… But what? It wasn’t good enough? I’d missed something? What was the universe (or more palatable to many of you, my subconscious mind) trying to tell me? Was there something in the stream of questions I routinely send into the ether that was in fact being answered with a two-by-four of sorts? Maybe. May be…

What had I written in the missing post? 

Overall I was suggesting that what the world is calling Artificial Intelligence is a misnomer. The body of applications and programming design strategies being peddled to us is better described as “Imitation Intelligence,” which is a very different thing. 

The specific impetus for my essay was Kim Comando’s nugget of info about color, which offered a proof of the difference that makes it more visible and definable.

A computer can perform calculations about color, specifically exact shades of color as defined by a quantitative identification system called RGB.

In other words, the computer can follow instructions about color in great detail. What it can’t do is actually recognize a color. You can’t point at something and ask the computer, “What color is that?” It has no basis for answering that simple question.

It can’t see. It can’t hear. It can’t taste. It can’t feel sensation or emotion. It can’t experience the sensation of living through time. Because it’s not physical and it’s not alive.

My post began from this point and made reference to an academic philosophical article I’d encountered years ago, which was all about the problems posed by the human question “What is Red?” This wasn’t a technological treatment of the question but a far deeper plunge into the nature of consciousness itself. 

The philosopher was at length to make distinctions between the words “red” and “redness,” and terms like “things that are red,” “the essence of redness,” and what he called, finally, “the sense of redness.”

You’ll have to trust me when I tell you the article’ author did not really solve the problems of definition he was attaching labels to. The net effect for me as a reader was observing evidence of what had killed philosophy as an academic discipline many decades ago, the collapse of words and language alone as a means of addressing existential questions. Why I had given up such readings after unpleasant bouts with Immanuel Kant as a youth. He destroys the meanings of words and then pretends that he has better words to fill in the blanks he’s created. Which he doesn’t.

The article on redness, however, did have one takeaway worth remembering. It asked readers to examine what came to mind when the word red is thought of. Not as a thing but a concept. Generally there is a shapeless smear in some shade of red, not a memory, not an answer, just a symbolic image made up to meet a mind/brain request. This raises the consciousness problem, which is that mental experience is not strictly physical but metaphysical. It functions by giving us a transcendent version of the physical experience of the body. Not the imitative version of reality cobbled together by digital algorithms but something higher than the mental, yes, even conscious experience of great apes, whales, elephants, and dogs. Human consciousness includes the “sense of redness” but also much more than that.

My next step in the post text was to reference a memorable nugget from one of Rex Stout’s novels about his iconic armchair detective Nero Wolfe. In the course of hunting down some troublesome detail in a murder investigation, Wolfe asks a young woman, “Do you know the difference between Invention and Imagination?” (I forget the specific context…) She cocked her pretty head and replied, “I can’t define it for you, but I know which is which when I see it.” Wolfe liked her answer. (Critics came to call this woman Wolfe’s Irene Adler, one of many comparisons to the Holmes oeuvre,) 

“Imagination” is the great leap ahead from “Redness” that exposes the problem with “Artificial Intelligence.” Through long practice (deliberate?) we have used the words “intelligent” and “smart” almost interchangeably when, in fact, “smart” is only a subset of the real meaning of intelligence, which is “understanding.” Understanding is not something IQ tests measure, for example, except in the narrowest and shallowest of ways. Reading comprehension is not synonymous, for example, with understanding what has been read. It is in the testing universe confined to remembering the common definitions of words used in a sentence and what the logic of the sentence might be. The questions Why? and Who?, for example, never appear in IQ tests, which measure skills and cultural data sets by means of timed, context-free exercises with clearly identifiable “right” and “wrong” answers. An IQ score is not a measure of any one mind’s powers of understanding, only their “smarts” in terms that make sense to decision makers looking to winnow out the dull and unacceptably ignorant.

No one anywhere has ever demonstrated that a computer program or system has powers of Imagination that transcend mere mechanical Invention. By mechanical, I made clear in my essay that in using the word “mechanical” I was not talking about physical cogs and gears but the step by step 1-to-‘n’ progressions we see in literal and figurative assembly lines. Best, quickest way to see what I’m talking about? The Rube Goldberg Machine:

No computer can produce anything like this image in response to the 

instruction

“Build a complicated machine to do something simple. And make it funny.”

My post went on to describe the process by which the word “Intelligence” wormed its way into the name of a new branch of programming experiments. The idea of linking large databases with algorithms that could perform sorting and compiling and formatting of whole disciplines of learning and information categories was initially labeled “Expert Systems.” Nothing wrong with that. Another set of pioneers began applying their programming skills to model behaviors in various animate realms, like the flocking behaviors of birds, and the natural course of micro-evolution and complexification observed in microscopic organisms. These were grouped a bit ambitiously under the term “Artificial Life”because its first milestone was enabling code to edit and/or rewrite itself in response to changing environmental conditions. The results looked something like what we observe in nature, with no acknowledgment that what was different in the programmer rules set was that unlike the Darwinians, the Artificial Life systems had a designer ‘god,’ who wrote the program that gave rise to them. 

Out of all this interesting work came a confusing mingling of the experimental arenas into a field that came to be called, largely by default, “Artificial Intelligence.” Was real imagination involved? Yes. By the programmers, not by the systems, which were still programs as they have always been designed, absolutely uncomprehending digitally defined algorithms that execute tasks exactly as defined by the code they’re made of. If you doubt this, make one typo in a password you use every day and every day and every day and realize that even if the correct password generates a friendly female “Hello, honey” response, the computer does not know who or what you are, or what anything is, except the next one or zero in the string of instructions. Everything that looks like intelligence is an illusion created by the blistering light speed of digital processing. If I can do something dumb at mind-blowing speed enough times to give you the answer you’ve specified, then I look smart. 

But I don’t even like fhe label “Smart Systems.” They’re not smart, even if they can improve their time to results by detecting and correcting inefficiencies in their own code, which definitions have to be built into the system by programmers in advance. [Note that as I write this, I just typoed “programmers” into “p rogrammers,” which was corrected by my smart AutoCorrect software to “p ro grammars” by a logic(?) I defy anyone to comprehend. Nothing here on this computer has any idea what I’m writing, why, or what any of it may mean.] 

And yet.. and yet… we have supposedly brilliant tech experts warning us about the Singularity that will occur in a decade, a month, or tomorrow when computers finally surpass human intelligence and start taking over.

My post continued by referencing my own recent personal experience with AI apps in the kinds of work I do, which do involve the use of Imagination. Mine, not the software’s. I have used some of the new hot licks apps read my writing out loud into a sound file, turn actual photographs into cartoon avatars, provide me with what look like photographs of faces that do not exist, and to turn song lyrics I typed onscreen into a complete song sung by a human female sounding voice to music played by what sounded like instruments. I can work with what I got, but I can also detect the gaps between what it says it does and what it really does. It’s all just programming, done at speeds once impossible, that combine data files which already exist in combinations and against backgrounds that make them look unique, though in reality they are merely collages arranged by uncomprehending instruction sets. I could prove this by example but I won’t. I can summarize by reporting that I have achieved outcomes I accept but they are all the result of imaginative decisions by me, step by step by step, sometimes for hours, and never the result of an instruction by me to any app for a “sexy older woman with troubled eyes, a bad boyfriend, and a tight budget.”

People are inevitably going to be disappointed with what they get from the miraculous new AI products they’ve been promised. There was a dot com bubble, a subprime mortgages bubble, and various other investment bubbles that have popped and taken down the whole economy with them in the past. AI is likely to do that again in the not-too distant future. Look out…

“IT’S NOT ALIVE!,,.” It’s just more junky software written by illiterates and sold by conmen. Don’t say I didn’t warn you.

*******************

That’s pretty much where I got to in my missing post. Why did the universe force me to write a do-over?

Because of all the stuff I left out that makes the Artificial Intelligence” problem more sinister than the post indicated. I’ve said nothing about Chaos Theory, Complexity Theory, the hoary Turing Test, Quantum Mechanics and the quantum implications of consciousness, (so-called) Quantum Computing, String Theory, and very importantly, the deep-seated institutional corruptions that have compromised our technologies, sciences, and academic vetting processes for a great many years now. The keywords that must undergird meaningful discussions of these additional topics have been introduced here though. Consciousness, Imagination, Intelligence, and the Senses that combine to make human experience exponentially more powerful than anything any computer ever dreamt of is capable of.

I also have an obligation to explain more about the personal operating mode of accepting input from the universe. It’s a real thing. And my last big project as a writer is to prove it to those who are interested. Stay tuned.

P.S.. A quick word about my own early exploration of AI back in the late 1970s.