A while back, I postulated that the most of our pop-science (and sci-fi) methods of testing for "consciousness" were, at best, tests of processing speed and efficiency.
Yes, including the one in Blade Runner.
However, because the — or one of the — functions of consciousness seems to be resolving conflict, including incorporating new paradigms when necessary, I also predicted that an intelligent entity that was not conscious could not comprehend the material passing through it (akin to the "Chinese Room" thought experiment). That implies that the ability to actually reason and incorporate new data would be a possible way to test for consciousness. {1}
AI/ML Can’t Reason
Apple researchers have demonstrated that "generative AI" is quite incapable of reasoning.
"We hypothesize that this decline [in accuracy] is due to the fact that current LLMs are not capable of genuine logical reasoning; instead, they attempt to replicate the reasoning steps observed in their training data."
Gary Marcus breaks down what that research article means on his Substack; for me, this is as close to a natural experiment testing my hypothesis as we’re going to get, and the results match what I would predict from my hypothesis. (Yes, I have plenty of anecdata to support it, but systematic research is better.)
This does not mean that generative AI is not intelligent; despite humanity’s conflation of the two, consciousness is not a prerequisite for being intelligent. It does imply pretty strongly that we have to be very aware of the limitations of this class of tool. It’s a shiny, fancy, cool new tool, but that doesn’t mean it’s useful at all times and in all situations. That new power drill may be cool, and may be able to do other things than drill a hole, but it’s still a pretty crappy hammer, and there’s no reason to incorporate it into your phone.
What About People?
What this research implies about humans who exhibit this kind of behavior and how "conscious" they are… is both a lot and not a lot. {2}
That’s because — at least for humans — consciousness is not a quality, it is a state. From sleepwalking/talking to the autonomic nervous system to "driving on automatic," there are times that humans act intelligently but not consciously. That implies that a hypothetical philosophical zombie "is" a p-zombie in the same way that someone "is" asleep. Therefore, if my "crazy conspiracy theory" is correct, it is possible to kick that consciousness back "on", the way that a cat running out into the street will jolt you into full awareness when you were unconsciously driving home. {3}
It’s Not About Ideology
Yes, I know what it sounds like, but I am NOT talking about "redpilling" or "woke mind virus" stupidity. My hypothesis is both testable and ideology-agnostic. The fools spouting nonsense about a "woke mind virus" are making a judgment based on whether or not you agree with them. (Ironically, that’s the sort of misunderstanding of the concept that you would expect from, say, a LLM or a talking p-zombie).
My hypothesis is about the ability (or lack thereof) to incorporate a new idea or paradigm and to resolve by integration conflicting data. You can find the kinds of behavior I’m describing in groups of all political persuasions.
Further, this is just a descriptive model. Even if I’m completely correct, I have no clue as to the actual mechanisms by which this would happen. I’m very aware that I’m using the transitive property a lot with a lot of thought experiments. That’s a dangerous thing to do: thought experiments like Schrödinger’s cat are often misunderstood {4} and can lead to wrong conclusions when taken too literally.
Consciousness By Accident
That said, it does serve as a reminder that we have to be careful and pay close attention to what we are creating.
If consciousness is an emergent state, not a quality on our fleshy "hardware" for reasons we do not understand, then we must acknowledge the possibility that the same thing could happen with an inorganic system. Further, that state may not be due to the actual code or hardware, just as our consciousness doesn’t turn on and off only because of our physiological state.
But we have to remember: Until we know what the necessary components are for consciousness to arise, we simply do not know if we are creating the conditions for that state to emerge elsewhere.
And that’s a degree of responsibility that deserves more respect than it’s getting.
{1} This may be a second.
{2} At least one person’s called it a "crazy conspiracy theory" to my face.
{3} Or do what Sarasti did, but, y’know, that’s not advised unless you’re a fictional vampire in space.
{4} It was originally meant to show how nonsensical Schrödinger though quantum theory was, for example.