Posted by
Michel Leclerc
Taking my questioning on your mysterious statements, HopSan, from another thread to this one, where it belongs better, I would like to comment on your statement I highlighted in blue.
The way I see it – being an "expert" in semantics as well, but not so much as what you call a "techno-male" (which might be a cute name for an AI system [by the way, are you AI yourself?]), but rather as a linguist and a philosopher of language – it is rather the other way around. AI is eminently possible in a mechanistic/material world.
AI could be characterised as a mechanistic simulation of “human” intelligence – but presupposing that that "human" intelligence is of a mechanistic nature in the first place. In the discussion with Geoffrey Hinton there is a continuous confusion of "human intelligence" with "mechanistic" i.e. technically simulatable, intelligence. For the sake of the tea-time flavour of the conversation the word "smart" is used.“Yes, yes, they will soon be smarter than us. They are already, actually. Smarter.”
Why would I even bother? Because "human intelligence" – even when it is not necessarily considered as just "one of the intelligences, beside emotional intelligence, musical intelligence etc." (which is a flawed way of reasoning because it uses the term "intelligence" metaphorically and literal and metaphorical uses of a concept move between language and meta-language, which cripples rationality) but when it is considered as “the" intelligence, often considered measurable by IQ tests (which are however only approximations of potential measurement) – moves itself continuously between several layers of meta-languages, in a "rhizomatic" kind of way, continuously applying the "meta" grip so to speak to different and fast changing aggregates of meanings and orders of meanings being "meta-ed".
Or, in other words, because you mention Gödel yourself: the ultimate metalanguage, or the ultimate meta-axiomatic system does exist, it is language itself. Ernie Nemeth has excellently described earlier the way "quasi-meaning" moves in a conversation through a process of continuous adjustment, refinement, transformation.
If it were possible to imagine a conversation which would only use meaning systems axiomatically defined (branches of mathematics), even there the conversation between Field medalists would be a continuous adjustment of shifts from axiomatic system to axiomatic system in a potentially infinite number of ways.
And yes, what you write about the bees is correct (and beautiful!). All animals use "languages of languages”.
However, if we arrive at somehow narrowing down human linguistic interaction to purely "intelligent" statements staying within the same axiomatic reference framework, then a machine, working in a similar "intelligent" framework of statements, might be able to simulate such our language, and hence interact with us "smartly”. Beat us at “smartness”.
Now that world, that mechanistic, material world as you write (or I would rather state: "that mechanistic material world" because I am not so sure that a material world would be mechanistic per se)) does exist because it is constantly being created by the projects aiming at "simulating" and "beating" our "smartness". AI is, in other words, per definition, effective within a world that has been manipulated beforehand to accept the simulation of its manipulated version.
And this is not a trivial tautology only. As Denise’s posts show, the various forms of collaboration between “mechanistic material” artefacts and live tissue and systems (animals, us) exist in the real world – and so it is clear that a vast programme is underway to render Life sufficiently “mechanistically material” to accept the “agency” of “hyper-smart” systems, i.e. AI. (To switch to another frame of reference this might be the sophisticated, more advanced version of a hybridisation program between humans and robot ETs or UTs.)
Unless Life is reduced to its simulated version, AI simulation will never work. I agree with you that consciousness (which Gödelianly speaking, might be understood as the ability to move in an infinity of axiomatic systems that can “meta” each other in an infinite number of ways – or linguistically speaking, the play with an infinity of “names”, each “name” within any language defining a dimension in itself) cannot be “mechanistically materially” simulated.
Yet “they” are trying – and progressing on the “mechanisation of Life” track. Our “hybridisation” being in reality the latest, most radical version of the “dumbing down program”.
But – just like certain Buddhist monks decide to starve themselves to death, or Christian mystics in extasis live on the consumption of Jesus’ transsubstantiated body only, we have the ability to resist the attempts at “transmechanising” us by deciding to leave our bodies once compromised irredeemably. Our consciousness will then move to other realities. The robots may continue to play around as ghosts in vast computing arrays, but they will never experience Life – as the pursuit of truth, goodness and beauty – also called Love.
They may be very “smart”, but they will not be “sentient”. The AI proponents may use that word – but if they wish to give it a genuine meaning, they will take off the shelf what we call life and love to borrow the meaning from. “Sentient”, or even “smart” are used metaphorically whenever they describe a quality possessed by a simulation. It is a simulated meaning, not meaning itself.
I guess that is what you mean, HopSan.
