Philosophy Now Forum

how would you program them to interpret meaning correctly?

When you consider that a military guard’s primary responsibility is to ascertain reality according to conditions and then respond appropriately:

Could a robot ever do that, while armed with lethal force?

video links

I suspect that consciousness - that is, a "what it's like" - comes first, then meaning/intelligence can arise.

Reason is that anything I imagine has a "what it's like" to it - e.g. a bright shade of blue spanning across a plane. A computer can output this to a screen, but that doesn't entail it's actually blue to the computer. It was just calculations that created what, to a human, looks blue.

Actually I doubt a computer even performs calculations - it's just a machine following the laws of physics, and to a human interpreter, is "as-if" making calculations.

I recall John Searle argued along the lines above.

So it may be that we need the right chemical/biological material, constructed in the right way, before we can get consciousness and intelligence.

Perhaps this view is too dualist however. Perhaps a better conception is that the computer is an extension of human nervous systems. When you bing search, for example, the computer's activity has meaning because it has meaning for you.

we'll it would never be actual meaning to an so program. You would have to simulate meaning by proper and accurate association.

unique characteristics somehow extrapolated by the device and matched to language terms to come up with the correct interpretation.

osgart wrote:can we simulate meaning consciousness in artificial intelligence using language and program proper association

Yes - so long as you use the term - simulate.

I feel creating an "AI" capable of having a conversation to the degree of the Turing test is not overly difficult, however any bod suggesting that if it passed this test would render it 'conscious' is an idiot.

attofishpi wrote: ↑Fri May 26, 2017 10:06 pm

osgart wrote:can we simulate meaning consciousness in artificial intelligence using language and program proper association

Yes - so long as you use the term - simulate.

I feel creating an "AI" capable of having a conversation to the degree of the Turing test is not overly difficult, however any bod suggesting that if it passed this test would render it 'conscious' is an idiot.

Exactly. We cannot begin to imagine replicating consciousness within something else if we lack an understanding of that of ourselves. We may be able to make an AI seem conscious, but it actually having its own consciousness is an entirely separate matter.
We'd need to gain a base understanding of our own first, which has been made clear to be no easy task.