Keith Gunderson (1966).

Therefore, one shouldn't say that such statements are analytically false. Claiming that the statement, "Machines can be conscious," is analytically false requires that one foretell every possible change of use of the words which comprise the statement--an impossibility. Therefore, one shouldn't say that such statements are analytically false.

The Gunderson argument

Keith Gunderson supports his claim with the following arguments:

"If we were to endow a subject with, say, conscious capacities, we would, necessarily, be dealing with a subject which was not a machine or a robot after all. In short, 'If S is a machine or robot, S cannot have certain (conscious) capacities'. It is simply an analytic (or logical) truth that machines and robots are not conscious. On the basis of an analysis of such words as 'machine', 'robot', and 'conscious' we can see this."

"But this approach disregards the difference between a word being able to acquire a new use in the language because of the meaning it has, and a word being arbitrarily given a new use. It should be asked whether such words as 'robot', 'conscious', and the like might acquire new uses (in conjunction) such that we would have to say that 'Robots can be conscious' is true without any change in the core of the meaning of either 'robot' of 'conscious'. If it is admitted that a word could acquire a new use (which would affect the truth-value of certain statements in which that word appeared) without acquiring a new meaning, then to be committed to the view that 'Robots can be conscious' is analytically false commends us to the task of anticipating every possible new use for the word 'robot', or 'conscious', which may have such an effect. but this task seems in practice, if not in principle, impossible" (K. Gunderson, 1966?, p. 410)

References

Gunderson, Keith. 1966?. Cybernetics and mind-body problems. Inquiry, No. 12. pp. 406-419.