Until the "AI Slave" is able to formally identify that it has produced a correct solution, which requires the ability to identify and correct non-trivial mistak

Until the "AI Slave" is able to formally identify that it has produced a correct solution, which requires the ability to identify and correct non-trivial mistakes including at the "conceptual" level, and probably to provide the "rationale" for its "thinking", it is unlikely to affect this particular area too much.