Lorraine Ford I have tried repeatedly to explain the essence of how computers are made to work, but I can see that no one has the faintest clue what I am talking about. Suffice to say that computers/ AIs are not conscious, and can never be conscious.
Lorraine: there are many false assertions here, sorry in advance to have to point that out plainly. Second, we are getting to the root of the misconmunication (and third: I don't want to see insidious implicit disparaging sarcastic remarks from you again - I already had to get the moderators to remove two abusive messages from you: that last one is too subtly abusive for them to understand. Change your tone, please, to comply with the forum's terms).
Firstly I am an advanced Software and Hardware Engineer, using computers since 1978. I designed my own processor and took it through to tapeout in 180nm, and designed an instruction set extension which takes any RISC Scalar ISA and expands its capacity by several orders of magnitude. When applied to Power ISA's 350 instructions they were extended to 24 million. I mention this as context that if you are trying to explain how computers work to me, then it is my conmunication failure that you feel that you are not being heard.
So when you say "nobody has the faintest clue", and you are saying that when I have a BEng in Software Engineering from Imperial College it is... yeh 🙂
With that in mind, to illustrate both that you have been heard, and why you are wrong, I am reluctantly forced to drop into "Extreme Technical Knowledge Mode". sigh. here we go.
A conflation (one of many) that you are making here is an assumption that the hardware has a built-in restriction that prohibits it from running a program that implements Consciousness. From having studied well over a hundred ISAs including that of the TX-II, the CDC-6600, Cray Y/MP-1, IBM 360, PDP/8, Z80, x86, Power, M68k, and dozens more, that there is without a shadow of doubt no restriction whatsoever on any single ISA that would stop a computer from executing a program that implemented anything, let alone one that implemented Consciousness.
The caveat there is of course a bounded userspace program, not something that violates memory, disk, resource or any virtualisation constraints (cross-reference: Spectre, Meltdown, Heartbleed and an extreme hardware edge case: Rowhammer). Even the 1960s George III Operating System and the 1980s Cambridge Capability System restrict access to resources in controlled ways, that the programmer, if they attempted to intentionally step beyond them, we may assert that they are acting with malicious hostile intent (colloquially: "trying to hack the system"). Such efforts are the boundary caveat for the above, and are not per se part of the ISA itself, but are more part of the Systems Architecture.
If you really absolutely insist that the above assertion is false, we can go down the Formal Verification route: in the commercial world there is the company that wrote Bluespec: it is in Haskell (a functional programming language, hence it has Formal Proof capability), and they have a full Formal Correctness Proof for the RISC-V ISA (written and presented at IIT Madras in 2018)
Alternatively there is the FOSS route: symbiyosis which runs SAT Solvers at the back-end such as z3 and yices: I used these to run formal proofs on the HDL I developed, but other people used them again to write and run formal correctness proofs on the RISC-V ISA.
Bottom line: there is formal methodology for proving that binary executables do exactly what the ISA says they should do.
Your dismissive tone has forced me to put that caveat in front of you in advance but also as a tripwire to test if you are being wilfully and intentionally abusive, on the basis that if you, as a mathematician, attempt to dismiss Formal mathematical correctness proof techniques then we know that it was intentional on your part.
I trust from the above that you begin to appreciate that you have been heard, by someone who does not wish to "brag" that they have an overwhelming amount of computing knowledge, and who also is reluctant to state plainly that you're wrong, in case it is misconstrued. (For, it turns out, legitimate reasons).
Having logically eliminated hardware as the potential source of prohibition, we can now turn to the flawed assumption, and it has I am sorry to say been pointed out many times already: you are assuming that the programmer has a specification for implementing consciousness in software
I know of several software teams that are trying this: they are tantalisingly close but really struggling. They are proprietary: I cannot disclose details as they have trusted me to keep their ongoing development confidential.
Most other programmers absolutely have no idea whatsoever what they are doing. Not a clue. Ulla also asked the right question, here.
Consequently
- your combined lack of knowledge of how software is related to the underlying hardware, combined with
- the lack of evidence of success so far in the (known) world as to whether anyone has implemented an AGI, resulting in you making the false conclusion that it cannot be achieved by anyone combined with
- a false assertion that software-based Consciousness cannot ever be written combined with
- a lack of knowledge of what Consciousess actually is, combined with
- an unwillingness to discuss what the properties of Conscuousness might be such that
- the flaws in your arguments can be demonstrated...
your approach, Lorraine, is leading you to make and sustain a cascading sequence of false assertions that has overwhelmed me so greatly that I have been very reluctant to spend the time spelling them out.
Second: we have identified exactly the false assertion that I could not clearly recall, which you have again stated (and belittled me in a very unsafe way for not being able to precisely recall): "Suffice to say that computers/ AIs are not conscious, and can never be conscious."
If you wish to keep repeating this assertion then as a mathematician I ask you plainly for about the third time provide the proof
Provide the proof - in formal mathematical terminology, either by formal correctness logic or by analytic equations, I don't mind which, that your assertion is true.
I am trying very hard to break you out of the non-Conscious loop you are stuck in, by agreeing with you on the points that are clear and verifiably correct. However you are not then engaging with me to move forward out of the false assertions that you have made, and, very worryingly, engage repeatedly in forum terms and conditions violations as a means to evade taking onboard an alternative viewpoint / perspective / hypothesis.
Your lack of knowledge is not an excuse for dismissive language, hence the reporting of your posts. Scientific enquiry comes from engaging and testing of hypotheses: history is littered with painful examples of falsified research. I am here because I expect to find people willing to engage on Foundational questions. Safely.
The "light at the end of the tunnel", which is not an oncoming train, is that you are not alone. The number of people in the world of 7-8 billion who can give you a formal mathematical definition of Consciousness can be counted on the fingers of one hand. The number of people who understand the properties of Consciousness is much larger, but still relatively few and far between (guestimate thousands)
Everyone else is overwhelmed by what I call the "noise-to-signal" ratio, illustrated very plainly by Robert Kuhn's categorisation reaearch. Fun fun fun.
