Human simulation and human essence (an addendum)
In 1950, Alan Turing argued that if a machine had the ability to perfectly simulate human conversation, it would be reasonable to say that “the machine can think”. We say that the machine that can do a perfect simulation of a human being has “passed the Turing Test”.
Turing noticed that the only way you can tell if other people can “think” is by noticing how they act. He wrote “instead of arguing continually over this point, it is usual to have a polite convention that everyone thinks”. I call this bullet point “Turing’s Polite Convention“:
- If a machine acts like it’s thinking, then the machine is actually thinking, at least as far as we’re concerned.
This is somewhat similar to the deep-seated assumption in our culture I called “The Tell”, but it’s not the same thing.
I don’ t think Turing believed in the Tell; he was too smart for that. I’m sure Turing never intended the test to measure “intelligence” or “consciousness” or any of the “essential” things that we’re talking about here. His purpose was to give us a way to talk about the philosophy of AI that didn’t get involved in a lot of, well, philosophy. His goal was to show that most of the common objections to the proposal “A Machine Can Think” were, at best, poorly framed and, at worst, incoherent, and thus it is perfectly reasonable to suppose that “A Machine Can At Least Act Like It’s Thinking”. And that was enough. It helped justify starting the whole enterprise of AI research.
Unfortunately, Turing’s modest goals were ignored by many writers in the years after his paper was published. In futurology and popular articles about AI, the test was talked about as if it was a kind of gold standard for “true” artificial intelligence, and this would occasionally drift off into discussion about machines being “persons”. For example:
“[O]nce an AI does indeed pass the Turing Test, we may very well consider ‘it’ to be a ‘he’ or a ‘she.’ … it is also my view that once nonbiological intelligence does achieve a fully human level of intelligence, such that it can pass the Turing test, humans will treat such entities as if they were conscious.” Ray Kurzwill, “A Wager On The Turing Test: Why I Think it Will Win”, 2002
Here’s what I think happened: many commentators were unable to resist the pull of the Human Essence Assumption. They believed, deeply and unconsciously, that there was something special about human beings that was important, so different that it would show. (An assumption I call “The Tell“). That the only way to simulate a human being perfectly was to actually find a way to capture the human essence in a program.
Typically, this would naturally segue into the prediction that machines would suddenly become more powerful and dangerous, at least a dangerous as a human being. Passing the Turing Test was described as an historical, world-changing moment, like the discovery of fire or the wheel. This is a version of the assumption I’m calling Nishmet — that the debut of “true” artificial intelligence will be sudden and dramatic and have world changing consequences.
We intuitively feel that there must be a particular “level” of AI that is more significant and important, a level closely tied to the abilities of human beings, so we intuitively feel that there must be a name for this level. We called it “Passing the Turing Test”, but we weren’t really thinking about what we were saying.
“There are those who believe that passing the Turing test means that human-level intelligence will have been achieved by machines. The direct consequence of this, as pointed out by Kurzweil and others, is that the singularity will be upon us, thereby resulting in the demise of the human race.” Kevin Warwick and Hunan Shah, “Passing the Turing Test Does Not Mean the End of Humanity”, 2015
All that was nonsense, of course, and most of the academic writing and a good part of the popular writing about the Test agrees that it was nonsense. It’s not hard to see that it doesn’t measure “consciousness” or “intelligence” or “personhood”, and that passing it wouldn’t be a world-changing event. If you are really thinking about it, it’s obvious that:
- It only measures a behavior which can’t tell you if something is conscious
- It only measures human behavior which can’t tell you if it’s intelligent (The Economist called the test “Artificial Stupidity”. You got to love that.),
- Personhood is too difficult to define to be measured in any case.
The criticism took a good deal of different forms, from common-sense dismissals in the Economist or the New York Times to the dense impenetrable jargon of contemporary philosophy of mind.
I could be wrong about this, but I can’t help thinking that the sheer complexity of the philosophy on this is motivated reasoning. They spent way too much time trying to prove the obvious. I can’t help thinking that they began their investigation with an unexamined intuition about personhood or the future and then worked to defend it without ever saying it out loud. In fact, in their disagreements, they occasionally break down and accuse the other guy as being “under the grip of the ideology” (as John Searle wrote). I can’t prove this impression — it would take a lifetime to wade through all of the literature and catch them in the act. They’re too smart about it and there’s just too much of it.
_________________________________
Requiem for the Turing Test: The AI Effect
Chat-GPT 4.5 passed the Turing test easily and definitively in 2025. However, now that it has actually happened, people still aren’t comfortable saying that “Chat-GPT can think in the same sense that people can think.”
This is yet another example of the “AI Effect”. When presented with a real machine in the real world that passes the test, we change our story overnight. Now we “all agree” that “obviously” Chat-GPT doesn’t really think.
In fact, Google researcher Blake Lemoine was fired for saying as much.
It seems ironic to me that this is where we are at now. There were people who, for decades, had been arguing against the Turing test, and some of them were ridiculed them as Luddites by people who never took the time to actually understand what they were saying.
Faith in the Turing test was once a kind of secular technophilic religion for some people, believing that Turing’s argument implied that superintelligent machines were possible, that they could “forge the Gods“.
But religion adapts. It always has. Today, the Nishmet, the moment that we Forge the Gods, is called “the debut of AGI“. Never mind that “artificial general intelligence” originally meant “the ability to solve a variety of problems, as opposed to just one problem”. We need a word for this — and “passing the Turing Test” won’t work any more. Call it AGI. Why not?
CHARLES GILLINGHAM 