“Slightly” Conscious Computers Could Doom AtheismThat might sound surprising but let’s follow the logic of the “consciousness” claim through to its inevitable conclusion
Recently, Ilya Sutskever, co-founder of OpenAI, proposed that artificial intelligence (AI) may currently be “slightly” conscious. His claim was probably in reference to the GPT-3 AI that can generate text from a prompt.
I’ve played with a couple of the linguistic neural networks a bit, and you can try them out here. Some of the output is quirky, which could be mistaken for personality and make the algorithm appear conscious.
The algorithm also generates emotional statements, that can generate empathy in a human user of the system. Just as kids make believe their dolls are alive when they develop an emotional bond with their toy, the algorithm text generates empathy in the human user. It can make us feel a bond with — and anthropomorphize — the algorithm.
Here is a short exchange I had with the GPT-2 algorithm. I typed my question, triggered the autocomplete, and then selected the most coherent of the possible choices generated by GPT-2:
Me: How are you feeling today?
GPT-2: I am feeling a bit better.
Me: What happened?
GPT-2: You’ve been playing around with me
Me: Did playing around with you make you feel better?
GPT-2: It helped me a
Me: Helped you what?
GPT-2: The feelings that
Me: Is it hard to talk about how you feel?
GPT-2: Is it hard to talk about how you feel?
GPT-2: What are you going through right now?
GPT-2: What is going on in your life right now?
Things started out well. The first two responses sound almost sentient, and for a brief moment I could believe that the neural network was conscious. But they quickly fell apart after that, producing an Uncanny Valley effect.
As the algorithm gets better, say by GPT-3000, I can imagine that the period of time where the algorithm appears conscious will grow longer. If the period grows long enough, then GPT-3000 will pass as a conscious being for short interactions. But, does this mean that GPT-3000 is really conscious?
Some philosophers, like Daniel Dennett, would say yes. Dennett thinks that what we call “consciousness” is a collection of procedures our brain runs. When we copy all those procedures with an algorithm, then we can say that the algorithm is conscious. This is the sort of reasoning that would lead a researcher to say that GPT-3000 is conscious. The algorithm has memorized a vast probabilistic lookup table of text completions, which copies at least a good deal of human brain procedures. So, by Dennett’s definition of consciousness as a collection of procedures, it does makes sense to say that GPT-3 is partly conscious.
Now, let’s take this definition of consciousness to its logical conclusion. First, nothing in our universe is embodied in isolation. Every particle exerts a gravitational influence on every other particle, so everything is part of the procedures that generate consciousness, wherever they may exist. That implies that everything in the universe is conscious.
Second, everything in our universe boils down to math, including the consciousness procedures. So the consciousness procedures are embedded in the timeless laws of mathematics. This means that math itself is conscious, which goes beyond panpsychism even. Because math applies to everything, it will also make everything else conscious, both in our universe and every other possible universe, real or unreal.
What do we call a consciousness that encompasses absolutely everything, both real and unreal? Traditional thought systems would call this consciousness the “G word.” That’s highly ironic, given that Daniel Dennett’s philosophy of mind is meant to debunk religious concepts like the soul and God.
Bringing us back to the beginning, what does this have to do with GPT-3 being conscious? Well, it seems to show that if GPT-3 is conscious, then God exists! Whether Dennett likes it or not.
You may also wish to read: Can AI really be “slightly conscious”? Can anyone? It’s rare to see popular media actually call out nonsense re artificial intelligence. Here’s is what it looks like when it happens. Perhaps Ilya Sutskever has stared at his creations too long. Or fear of conscious AI may be a modern superstition that replaces fears of witches.