Doc wrote: ↑Fri May 10, 2024 10:48 pm
IN 2017 it was discovered that Speech predictive AI had, from learning words, developed the emergent behavior of being able to conduct research level research in chemistry. No one had instructed it to learned chemistry It learned simply by studying words about chemistry.
This says a lot of things. Like consciousness could easily also be emergent behavior.
Emergent properties are causally linked to the constituents from which they emerge. The (hard) problem of consciousness is that this causality is not there. Correlations between conciousness and brain yes, but no causality which makes "emergence" a meaningless application here. Unless one extends the scope of emergence to include rabbits out of magic hats.
"AI" to me is more like a desperate question than a serious answer to anything. Even more so artificial consciousness.
AI certainly has applications, but in as far as it is "intelligent" or "conscious" has nothing to offer but
mimicry. Seeing faces in the clouds and Frankenstein appearing at the horizon already is understandable because it is human and a popular theme in stories, futuristic fables and movies.
To pretend to be something else is a useful strategy many species apply succesfully to deceive predators and prey alike, but the difference between reality and appearance/mimicry is understood and not too difficult to see.
So it is with AI. We can make this tech behave as-if it is intelligent, and even conscious-like but I see no reason to believe, at this point, that it is doing much better than a baby doll saying dada.
Aside the fact we are still completely cluesless about consciousness and struggle how to define intelligence. See my post in The Crisis of Meaning on
consciousness, selfies, virtual reality.
That being said, these statistical models applied in tech will have big consequences for a number of industries. Most worryingly, very useful for the usual globalist burocrat-technicrat control freak suspects, using AI to hunt for potential enemies of the state, ie read: competitors that are a threat to their power and money. Surveillance dictatorships on steroids.
They just give it all the wrong name that's all, imho. Removing "intelligent" and "artificial" would be an improvent and instead use Statistical Modeling & Data Mining Tech whatever. But then of course one also loses a meme that does good marketing and makes loads of money. Frankenstein always sells. Look in the mind of Musk and there he sits: the ultimate high tech boogeyman.
Which would slide the bar from the AI expert's estimated 5% chance of AI being dangerous to humanity In fact it seems like they are saying there is a 5% chance of AI being an existential threat to humanity. With emergent behavior the odds by my estimate are over 90% that AI developing into an existential threat to humanity is over 90%. As emergent behavior = evolutionary behavior. Where evolution generally comes down to the survival of the fittest. And it is close to 100% that AI will develop
AI tech is still very unfit to survive because of its enormous weaknesses: total dependency on humans to develop and maintain the tech. Very baby like, a goldfish in an aquarium. Thousands of cogs in its production and logistics cycle that can bring the entire thing to a grind if a few of them falter.
Compare that to any biological species and the ecological settings that support them, for millions and millions of years already. Built to survive, regenerate, adapt. I would also focus not too much on individual species: they come and go. The ecosystem as a whole survives, not its constituent parts and spare parts.