At some point in time during the evolution, because of some factors, some beings first started to become conscious of themselves and their surroundings. That conscious experience is beyond some mere sensory reflexive actions trained. Can that be possible with AI?

  • 2
    $\begingroup$ Aren't these questions more suitable for philosphystackexchange? These are very abstract questions in which people have very less knowledge and people will have diverse opinions $\endgroup$
    – user9947
    Jan 27 '18 at 12:52

Current limitations in our knowledge mean that the question is not directly answerable:

  • There is no scientific consensus on what consciousness is. Therefore any device designed to "be conscious" is necessarily going to be built on the premise of unsupported, maybe fringe, theory.

  • There is no robust measure of consciousness. If any AI system was built in order to exhibit conscious behaviour, there would be no way to prove it is conscious. There is no general agreement or theory on whether any particular animal species is conscious for example. This is often limited by communication. Of the few animals smart enough to be trained in communication with humans, there appears to be conscious behaviour. Researcher opinion ranges from "all non-humans do not possess consciousness" to "all animals have some degree of consciousness".

  • There is incomplete understanding of what the components of consciousness are. A bottom-up build of a conscious machine requires a baseline theory of what those components are.

  • We may be able to ignore lack of knowledge and take a very high level of abstraction, such as A-life or evolutionary approach where nothing is assumed and the hope is that consciousness would spontaneously emerge from a complex enough simulation (as we assume it has done with organic life in the real world). However, this would seem to require many orders of magnitude more computing power than is currently possible.

To answer the question as written:

Can the first emergence of consciousness in the Evolution be replicated in AI?

Despite the many books, articles and posts written on this subject over many years, the answer is two-fold:

  • We do not know of any fundamental reason why AI could not be conscious.

  • We have no theory or experimental proof that AI can replicate consciousness.

I would go further than this, and say that anyone who tells you otherwise on these two points has already subscribed to some unproven theory of consciousness.

As well as well-thought-out peer-reviewed theories and experiments by scientists and researchers, there is a lot of pseudo-scientific junk published on the internet on this subject. So take care if researching reading material.

  • $\begingroup$ what if consciousness and intelligence is not solely the property and function of physical evolution surely then this leads only to simulation and not replication. $\endgroup$
    – Bobs
    Feb 7 '18 at 1:46
  • $\begingroup$ @Bobs: Potentially. However, that is already a statement from "some unproven theory of consciousness". There is absolutely no evidence that consciousness cannot be achieved physically via evolution. In that regards, all we have is a big question: How did life on Earth result in conscious systems? The answer to date: We don't know. $\endgroup$ Feb 7 '18 at 8:15
  • $\begingroup$ yes we don't have a concise theory of consiouness else we wouldn't be having this conversation I was merely pointing out my dualistic experience of my consciousness that leads me to believe that it's replication through construction may be more difficult because the physical and intangible experience I have and there separation. And at the moment there is no method to prove any other individual than me having a similar consciousness. Which means it's difficult to judge when another has it as well as a constructed machine. $\endgroup$
    – Bobs
    Feb 8 '18 at 23:26

It partly depends on the framing of the question, in terms of how you are defining consciousness.

Neil Slater's answer is comprehensive, and his warnings about pseudo-science and junk publication should be heeded.

However, since you frame this in the context of "Can the first emergence of consciousness in evolution be replicated in AI", I feel like I can provide an answer.

  • If we define rudimentary consciousness as simple awareness of the environment, distinct from higher functions such as self-consciousness, then yes.

Under this definition, any algorithm that takes input is "conscious". This in no way represents human-level consciousness, or even the consciousness of higher animals, but is more akin to simple organisms such as microbes.

  • $\begingroup$ under that definition it reduces response purely down mechanistic cause and effect from the algorithm and not a determination of response through a tendency to choose the reaction and not simply an equivocal reply $\endgroup$
    – Bobs
    Feb 10 '18 at 16:31
  • $\begingroup$ I personally have no doubt human reactions are largely mechanistic, and would surely be deterministic if quantum theory didn't open the door to true randomness in nature (validity of free will tbd.) But thank you for recognizing my reduction of the question. This is a complex problem, so starting small is likely the way to go. $\endgroup$
    – DukeZhou
    Feb 14 '18 at 20:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.