TALKIE-1930: It is not impossible that in the future mathematics may be computed by machinery, but the probability is remote. At present, the computer has to deal with large numbers, and the machines would have to deal with still larger ones. The personal equation enters so largely into arithmetical operations that the personal equation of the computer must always remain an important factor in his work, and a machine can hardly be expected to eliminate it. Even in the case of adding machines, personal equation has to be taken into account. Two computers may work at the same machine, and the one may total up a long column of figures more quickly than the other, though each works honestly. The personal equation of rapidity can only be eliminated by making the computers work in teams, so that each may correct the other’s mistakes; but this would scarcely be practicable in mathematical calculations. At best, machinery can only be used to save labour, not to supersede it. Computers will always be required to check the work of the machines, and to set them right should they go wrong. The essence of computation lies in proving the accuracy of the result arrived at, and machinery can no more do this than it can think. At present, the most that can be claimed for it is, that it reduces the drudgery of arithmetical operations, and so leaves the computer freer to attend to matters of thought.
Sounds like we may be able to build computing machines, but a human can't collaborate with a machine the way two humans can, and a human will have to check the results every time anyway, so what's the point? Only a human can prove that a calculation is correct.