Ram Allampalli

For non-verifiable domains, the only way you can improve AI performance at this time is via curating more annotated training data, which is expensive and only yields logarithmic improvements. And here's the thing: nearly all jobs have non-verifiable elements. There's virtually no job that's end-to-end verifiable. Even the job of a mathematician is not end-to-end verifiable. Sofware engineering involves many verifiable tasks, but it isn't end-to-end verifiable. For this reason the gap between "AI can automate most of these tasks" and "AI can fully replace this job" will remain for a very long time, across nearly all jobs. Like with self-driving, working 99% of the time is not nearly good enough to remove the human. So even when we get superhuman Automated Theorem Provers, mathematicians will still have jobs. We might even end up with more of them (scary I know)