Passing the Turing Test A Bumpy Road Ahead

Can a digital computer ever pass the Turing Test? In this paper, we argue that the metaphysical possibility of such a program is not the only concern. Even if we assume a generous variant of the Church-Turing thesis, practical concerns dominate two of today’s most promising approaches: human brain simulation and large-scale cognitive algorithms. We report on recent trends in the field of high-performance computing (HPC), suggesting that hardware failures and energy usage are limiting factors for tomorrow’s supercomputers. In addition, we observe that scaling up digital computers to handle large, unstructured problems appears to necessitate more “brain-like” hardware and software.