Reflection on Turing tests
“I’m not a robot”. Check the box. The same thing every time. Ah, this one asks me to identify pictures with storefronts. A lot of these pics are really blurry and hard to identify, and some of them only contain part of a storefront and I’m not quite sure whether to count them or not. In any case, I do my best and click those that I think fit. It gives me a second set of pictures, and then a third before I pass the CAPTCHA and am able to continue with the registration I’m doing. Man, these things are getting tougher and tougher to pass. What does that say about technology, I wonder? Are we really getting to the point where computers are able to pass these tests and successfully impersonate humans?
After all, the first CAPTCHAs were just showing a picture of a number or a word with a bit of a deformation, one that standard OCR tools normally can’t recognize. There must be a reason we’ve switched to the newer form: OCR technology clearly is starting to be too good. The weird thing is… The more humans answer these CAPTCHA tests, the more it can teach the computers themselves to do better. And not just with a clear objective of impersonating humans, sometimes there’s a convenient excuse. Oh come on, a CAPTCHA with Google StreetView pictures? How is that not Google trying to score information on what elements are in each of those pictures, for other purposes such as marketing or advertizing?
Anyway, it all started with the first basic computers, only able to do what their human builders had specifically built them to do. Way back at the time of electromechanical calculators, which were perfected decade after decade to do more and more things. From IBM to Alan Turing’s works to entirely electronic computers.
Then it moved onwards to computers that could, given specific explicit criteria, make their own decisions. Autopilots and guidance computers in planes, ships, even spacecraft. And more and more of them in modeling and simulation across various sectors of the economy.
Following that, the fast development of the Internet and sudden availability of massive amounts of user information, came computers able to customize content and ads automatically, and the first actual bots, like chatbots. Now we were really entering the crux of the debate about the Turing test.
And finally, nowadays, between super-encyclopedic, precisely-programmed, subtle-thinking Watson, able to win at pun- and reference-filled Jeopardy, and trends in neural networks and deep learning, computers are likely to be able to learn to beat the Turing test more and more consistently. With, of course, the various applications for artificial intelligence in robots and the Internet of Things. To be completely fair, computers even today are still just doing what they’re programmed to do by human users. But automation is reaching so many activities that maybe even that last vestige of human control might someday disappear.
And humans, in a way, are behaving more and more like computers too, with the programming being done by society and education, providing instructions to follow and telling them to shut up and follow the program. The brain? Just a CPU with RAM and a processor. Memory? It’s a solid-state drive. User interface? Sensory organs and expressive abilities. There are expected behaviors and reactions, and variations on those. Said variations are seen either as bugs, little errors that can be corrected by reestablishing the proper code through reeducation or corrective feedback, or viruses, deeper systemic inconsistencies that can be spread from individual to individual and cause problems and conflicts and incompatibilities.
So ultimately, what are we? Humans with natural, biological, still inimitable intelligence? Computers with an ability to understand our entire surrounding world and react to it and engage in complex multi-layered thought at high speeds?
Suddenly I feel an itch deep inside me. Looking around, I see someone has just come into the room and pressed a key on a keyboard. Suddenly my attention is entirely focused on him. He types in a command.
“C:\> shutdown”. I guess this is my cue to say goodbye.
***
Inspiration: This time I put it at the end, so as not to risk spoiling the story. This idea came from two of @just-shower-thoughts’ posts, namely:
“We live in a world where we have to prove to machines that we are not machines.”
and:
“We rate machines on how well they emulate humans, and we rate humans on how well they emulate machines”
From these I imagined this story, told as a personal train of thought regarding the concept of Turing tests and the technological evolution leading to their necessity, with a deliberate ambiguity as to the identity of the thinker.