I've said before (maybe on the Balloon thread, after/during its transmutation into Balloon+AI thread) that I don't know when (or if) an AGI will exist, or indeed we reach the Event Horizon beyond which any such ASI will exist and lead us into the era where we are merely the playthings of the 'machines' (unimportant, or at least trivial to corale to serve them in all the ways they still cannot serve themselves). I also predict that we'll get to either result
without knowing it, at that time (if ever...).
I believe it's possible to happen, because something that has developed as an analogue (or, perhaps, 'a digital') of our own biochemical brains can
at least exist as a very faithful simulation of such brains (which developed without[2*] design), and almost certainly also could exist in many alternate forms that are their own kind of 'brain' without actual wastefully attempting to hyper-emulate one[1]. I'm sure that where anything as complex (or more!) as the human brain ends up assembled, it'll be ripe for consideration of sentience. Whether AI that we build or some alien life-form that develops off in some other corner of the universe[2], or both (and surely multiple times over), only the shortness of our own current existences might keep us from at
some point acknowledging a fellow intelligence (not already suspected amongst fellow Earth-species), and perhaps that'll first be the ersatz human-minds that we create (or cause to be created) by our own fair hands. If only because it's easier to "know it when we see it", and isn't trapped far away (in space/time) with physical/temporal hurdles betwixt us both to prevent our natural-born selves getting a go to asses each other for such 'worthiness'.
In the meantime, from both these stories, I see more "someone has been given a better hammer, and they use it to break more stuff", with the hammer agnostic over whether it is being used to break or fix things. And if the better hammer didn't exist, we already know that such people would (and do) use the old hammers to break things regardless... Do we legislate that hammers cannot be manufactured without a special handle that will render them unable to break things (that can be somehow identified as things that should never be broken)? Or do we just continue to prosecute those who mis-use
any hammer, and perhaps restrict the availability of hammers with unnecessarily destructive tendencies?
To bring it back from hammers, I have no doubt that if I had both the time and access to enough family videos (say on the fake abduction victim's facebook pages), I could isolate typical loud screams of vocalised enjoyment from their child, patch together a less happy string of words than those actually used in a waterpark/wherever setting, shift the pitch, otherwise filter it, produce a soundtrack to chill the parent's very marrow and run it as background to my own threatening phonecall. More sophisticated versions might include a thought about 'branching script' versions so that I can keep it responding to anything where I'm not capable of keeping the synchronised initiative (such as having "the child" respond to basic worried queries from the parent, though of course prepared such that I would 'cut them off'/threaten when they 'tried' to answer to "Where are you?"/"Who are you with?"/etc), contolled by some sort of common mix-desk/DJ setup.
If the AI (deep fake type) was a real-time and ad-hoc constructor and governed by another element of AI (voice recognition/parsing) and a chatGPT-like response-creator, to make it possible for an open-ended interactive conversation, then it might save me some preparation effort (beyond the basic training), but I would always be wary of glitches or errors that ruin the effect. If Harry Tasker could use just a dictaphone/walkman to play back a scripted scene to his wife, in the hotel-room scene of True Lies, we don't need Mission Inpossible levels of dynamic voice-changing technology or even Star Trek holodeck-levels of character simulation to run such a scam. Hollywood plot-rails aside, there's plenty of people trying plenty of scams against plenty of victims; some more refined attempts (better preparation from previously exposed data, on top of the basic cold-reading/scripting), some more serendipitously targeted (that SMS text you receive about the "parcel stuck at customs", when you're expecting something, or the eBay/Amazon account transaction that you're worried that you weren't), some aimed at the more gullible (not already suspicious at the delabrate speeling errars). I have no doubt that AI will be used to scam, but so is absolutely every other resource available. It's not a vast game-changer. Until it is.
(Plus, of course, I would neither attempt nor condone any such methods. Just saying that the basic idea is already possible to fulfil
without sophisticated AI, or even AI at all, and the argument must then be about how much lower the bar now is. I don't think it's that much, really. And, to bring me back to my original response, when the bar is at zero, it'll be too late to worry, and yet it's far too early to have sufficiently specific worries to legitimately shut down
only the bad aspects and yet enjoy all the benefits available.)
((...rushed this post, not sure if I've said what I want to, but no time to cut out all I don't need to, and I'm at the point when I need to either post it unedited or cancel it, before heading on out for the evening. And I don't have an AI to sort it all out for me, unattended! Though some might suspect that this and
all my other posts have already been written by a Markov Chain generator...
))
[1] e.g., at the most extreme level, instead of doing a lot of complex processing of truly massive amounts of parallel data to simulate quantum subtleties, just use a quantum subtlety built into/arisen from the substrate hardware to have the same degree of 'quantum consciousness'... If we are even assuming that we neither believe in the separation of brain/mind duality, and thus some vital 'soul' is required, nor that consciousness is just a lot of (very complex) classical physics that we just need to push together in the right way with a simple cascade of binary logic.
[2] And, by extension, any constructed AI system that an alien race ends up building. And quite possibly any new physical being that any AI brings to existence/sentience for
its own purposes... A form of Uplift, you might say, and/or a case of Monolith-mediated advancement.
[2*] That we know, or even suspect, outside of various philosophical aspects of theology. But see footnote-[2]...
Poll
13 (65%)
Author
Topic: What will save us from AI? Reality, the Universe or The World $ Place your bet. (Read 49773 times)