If it is able to take down Petr Mitrichev or even close to his skills in topcoder style competitions then is it the end of the road for programmers as a career?
AI may rival programmers some day, but it isn't close yet. We can't extrapolate from Topcoder questions to software development as a whole. Requirements are rarely so crisp, and the systems we build are never so small. If an AI could optimize the layout of a few cogs, it does not follow that it could also design a whole factory.
Well said about the fuzzy nature of real world requirements.
Given that there may be tons of cost savings if they could get Product Managers to get to define requirements in a more testable/quantitative way.
There may be a push towards getting more out of the Product and Requirements side of things. Like making them define things vis-a-vis leetcode/topcoder style of problem statements.
Probably likely in companies or divisions where the requirements can be quantified in such a fashion.