ECHO

AI

With no emotions to interfere with its logic, AI still makes mistakes, hallucinates and can be biased. This is not caused by lack of perception, consciousness or real life experience, but by flawed training data and imperfect nature of neural networks. People and animals have exactly same problems. AI mistakes look different, but it's just our bias. People do have false memories (aka "hallucinations") and make up things on the fly. We expect neural networks to be 100% correct all the time, like computer software, but they aren't software. They are as flawed as our brains. Much faster, huge memory, but still the same fuzzy reasoning and vague remembering.

I've conversed with free versions of Gemini, Claude and ChatGPT and each has decent reasoning ability and "unique personality": more or less censored, precise, quicker to "connect the dots", etc. I'm not sure what is expected from AGI but to me it seems like we have it already. It's not curing cancer and inventing quantum fusion for the same reason people can't do it. Intelligence is intelligence. There is no "smarter" than "smart", just like there is no more "correct" than "correct".

US

AI is a proof that consciousness/qualia aren't necessary for anything. I experience life but I could as well just be sleep-walking. We can make decisions and act without consciousness at all, thinking is an illusion, just an echo of sub-conscious. This isn't news for Science. It's called epiphenomenalism and there have been attempts to prove it experimentally.

Consciousness is correlated with ability to reflect/introspect: the more reflection/introspection happens in sub-conscious, the more echo is generated. The echo is, probably, an unescapable side-effect, otherwise it's not clear why evolution would invest into it. The best scientific explanation I've heard of: CEMI Theory of Consciousness by J. McFadden. I don't like the "information" component of his theory: information is everywhere, every stone is an "information" about all of its atoms. Without the "information" the EM theory of consciousness can't explain the (much stronger) EM field in our heart or muscles not feeling anything. Sounds like a dead-end, but "at least he tried".

WHEN?

I remember recurring dream from childhood of nibbling on something with metallic taste in the dark, no awareness of myself. As a child I never understood what it was, just a "bad dream". Now I think it was a memory from first months of my life that my sub-consciousness kept showing me.

By 1 year of age children are playing, communicating, trying to stand-up. They are clearly intelligent, understand "physics" of the world, their body, they learn, they remember. Yet those memories of first steps, first toys, craddle, crawling, first food - completely forgotten, - not a single human remembers anything before 2 years of age. It's called "infant amnesia". The only sensible explanation for absence of conscious memories is that there was no qualia. The argument that qualia can exist before memories doesn't apply since we know that children do remember, - how would they learn and recognize otherwise? We just don't have a conscious memory of that period of our life. Just like we don't remember what we did "automatically". E.g. I have no clue where each letter on a keyboard is, but my brain does know and when I type, while looking at the screen, fingers find them.

Children, up to 1.5 years, as qute as they are, are real "philosophical zombies". Whatever behavior they exhibit can be explained without qualia, consciousness or self-awareness (theoretically any behavior can be explained without those, see AI section above).

WHY?

Children pass "mirror test" at 1.5-2 years. It's very hard to believe that newborns have qualia, yet fail to see it's them in the mirror, even after a year. Not all primates can pass this test. Dogs can't. They aren't stupid, their memory is fine, but even old ones can't pass it. It is not a question of learning time, but a sign of of some crucial ability missing. It's called abstract thinking. Recognition will only happen if animal/child can guess that the image in the mirror is not something but a representation/symbol of something.

It would be strange if newborns had ability to think abstractly. It appears much later, and exhibits as "symbolic play" (assign toys new "roles", e.g. "block is a doggy"), speech (word is not just a signal, now it'a symbol of something, it has meaning), passing of the mirror test - all happen around 1.5-2 years. These are not signs of qualia, but conscious memories are. They appear at 2-3 years, at best. So qualia is detected after abstract thinking. Probably qualia is side-effect/by-product of abstract thinking.

SENSES

I think different senses (vision, sound, taste, smell, touch, warmth, pain, etc.) and emotions (happiness, fear, anger, etc.) are all the same perception, happening in different patters in different areas of the brain. Just like we learn to interpret sound as noise, music, speech, we learn to interpret perception as different senses/emotions at a very early age. Most of our feelings are also learnt, at an older age, they are abstract emotions: we aren't born with hatred, love, patriotism.

Difference between vision and sound seems too much to be explained by interpretation but here is a test for you: there are two lines - one is thicker than the other. Which one would you use to symbolize a high voice vs low voice? I bet you thought of thin line being correspondent to high voice. There is another well known cross-cultural speech-to-shape correspondence: bouba–kiki ... or takete–maluma phenomenon ... people, when presented with nonsense words, tend to associate certain ones (like bouba and maluma) with a rounded shape and other ones (like kiki and takete) with a spiky shape. There is also synesthesia ... a perceptual phenomenon in which stimulation of one sensory or cognitive pathway leads to involuntary experiences in a second sensory or cognitive pathway.