I want *programmers* to do the job _right_ to avoid problems and to produce the *best possible* AI. I’m thinking of TODAY leading to TOMORROW, not tomorrow as an isolated “what if” scenario.
Example: I have a team of programmers. We’re designing AI. We have to think 50, 100, 500 years in the future and we’d better do it _right_.
If we do it wrong, well, _then_ there may be problems with whatever AI produces later on.
Future AI will only write itself if we make CURRENT AI capable of writing itself properly in the future.