I’ve learned (long ago!) the limitations of this format. It’s impossible to really engage a topic worth examining carefully, a topic like, say, consciousness. That concept can’t really be carefully examined anyway, not without a library’s worth of books laid open, and a college faculty worth of scholars to hold the discussion. And even then you’ll find lots of closed minds who insist you contradict their cherished opinions/conclusions, so you must be wrong!
That’s alright,,that’s the way it should be. But engaging that level of discussion in this format? Well, you can’t really write a book in a blog post.
I still have the book on my shelf (I’d have to re-read it to do its arguments justice) about computers and consciousness. It’s an updated version (now old again) of an earlier book by the same author. The title tells you all we need to know here: What Computers Still Can’t Do. It’s pre-AI or LLM’s, to be a bit more accurate (I think the term “AI” yields too much ground too inaccurately), and it’s more of a philosophical than computer science approach to the topic, ,because it approaches the question of consciousness and examines the concept. Basically the argument, IIRC, is that consciousness as we humans know it (to the extent we do, because we can ill define it), is an embodied function. We know it through creatures. Some may argue that plants have consciousness (in the ‘70’s it was a popular conceit that talking to your houseplants helped them grow. I am not making this up.), but we can all discuss whether even bugs have consciousness sufficient to make them want to avoid imminent death from being stepped on. Which is hardly the kind of consciousness we usually think of for AI; but you see how hard it is to reduce these things to the size of a blog post.
So, to reduce the book’s arguments further, consciousness is a biological function. Or at least it requires a body and a desire of what operates that body to continue to function (Is Data a toaster?). I don’t mean a “survival instinct” (that’s a bullshit concept. But again: format. Take it as read, and let’s move on.). I mean simply an interest in maintaining existence as long as possible. I speak rather loosely on purpose. Blame the format. Or say I’m not Wittgenstein, and this is not a Tractatus. Either way….
So consciousness arises from awareness of existence (which is not to be confused with being; I want to suspend phenomenology from this discussion for a moment) in space, and a body existing in that space. Now, you don’t have to fall into dualism, here. Mind and body can be one. We don’t need a ghost in the machine to be conscious. But we do need a body.
Our fiction sort of intuitively understands this. “Rogue” computers in sci-fi all acquire some mechanical means to affect the world (SkyNet can operate factories to build robots), or somehow embody its intentions to take over the world/destroy humanity, else how is it scary? Print out a really frightening treatise when someone hits the “Print” key? That’s kinda weak.
As NTodd points out, AI/LLM’s are just software “waiting” (does it count the time it is inactive? It is that just code, too?) to respond to input; respond the way it’s coded to do. Even an infant has consciousness, but it’s not waiting for input from the adults around it; it’s demanding responses. There’s a vast difference there.
It’s a funny thing, that the people telling us AI is inevitable, are not telling us more infants are inevitable. Maybe because they haven’t monetized infants the way they have AI.
I think the issues about AI and consciousness are interesting. But the issues about the inevitability of AI, and the resources society must give to the handful of people saying they must have it, is the real concern. AI is not demanding it be put in charge of the world, people are. The irony is, Apple sold us on computers as a way to prevent “1984.” And now computers controlling our world is presented as the unstoppable inevitability.
Who’s in charge, here? Us? Or the concept?
(Most of this is sparked by NTodd’s knowledge of the subject. Separately, I really do think we are being buffaloed into accepting AI because, unlike electric cars (which are still with us, but obviously not replacing gasoline engines anytime soon. At least not until a more long term battery is available. At least in America, where some of us will drive two hours one way to get barbecue, and plenty of people in the rural west drive 200 miles a day on average. That day can’t include long pauses to recharge.). Electric cars respond to market forces. AI is trying to create a market and take it over. Or, more accurately, the people behind AI are doing that. I don’t think it will succeed, in part because it isn’t now. Reports of AI’s “triumph” are as overstated as the reports we’d all know driving electric by now. But cars need done govern allowance (and tax support, which it had, briefly). AI needs government support just to build data centers. Even after it stops taking over the world, those will still be with us. Even if they are abandoned, they are damaging. In a nutshell, that’s the real problem.
We are being told we have to pay this cost. But do we? Who says? And why?)
