"Useful" is a loaded term. When people say that quantum computers are currently not useful, they mean that quantum computers are currently not maximizing shareholder value. IBM, Google, various startups, and others work on quantum computing because they hope that it will maximize shareholder value. The worst-case scenario for shareholder value maximizers is that quantum computing will only improve society indirectly. That is, quantum computing will only remain useful for basic research. Given that Post-quantum cryptography invalidates Shor's algorithm as a potential shareholder useful application and the overwhelming evidence that quantum computers will only be useful for doing quantum science, how should the quantum computing industry proceed?
The best path is to carve out a subfield in quantum computing. The purpose of this subfield is to leverage Feynman's insight that quantum computers are very good at simulating quantum physics in order to improve the applications of quantum mechanics. Shareholder useful applications of quantum mechanics such as atomic clocks, lasers and MRIs included. The subfield we are carving recognizes that quantum computing is a metatechnology: Technology used to design shareholder useful technologies that utilize quantum mechanics.
Let's call this subfield "QCAD" for "Quantum computer-aided design". It's a field that uses a quantum computer to design an extremely precise atomic clock or an accurate virus sensor. It's not classical computer-aided design but on a quantum computer. That is, it's not used to design, say, the ramsey cavity, glass rod or vacuum chamber of an atomic clock. It is however, something that allows quantum computing shareholders to create commercial value by licensing the designs that Quantum computer-aided designers generate to organizations that manufacture shareholder useful quantum technologies.
Discuss on Github