Yes, that is easily enough corrected.
The larger problem is that there is a limited range of responses to inputs.
That's inherently different from a human evaluator, who can adjust his/her tone/response/expectations according to the student's individual needs from day to day.
We've lived this for seven years now, and trust me when I say that even the most well-engineered instructional space lags horribly behind what live, Socratic instruction can accomplish.
There is no way for a computer to grade an open-ended response to anything in anything like an adequate (nevermind "good") manner.
Even MOOC's have learned that there simply isn't any substitute for that part of things. You can't program a computer to understand a nuanced response to a query which requires synthesis and extension of students, because the possible output from those students is simply too varied and unpredictable. It depends upon not only their understanding of what they've seen in the current instructional module, but upon all that came before, and no two students are exactly alike in that particular regard.
Gifted students suffer disproportionately from computerized/rubric-driven grading done by automata, quite honestly. One reason for this is that they tend to accumulate experience/information/understanding and naturally apply it broadly more than most students. Another reason is that they, more than most students, tend to see nuance and ambiguity where none was intended, and carefully examine information, often resulting in creative, innovative, or just divergent responses/methods. In other words, the higher the LOG, the less predictable such students are, on average.
Computerized assessment/teaching is fully predicated on the predictability of students as learners.
Flipped classrooms? EXCELLENT idea. But that means that there is still a classroom (e.g. an environment where students and instructors come together to discuss and share).
Automated assessment? Horrible in practice, most especially for GT learners, who all too often find that EVERY T/F question is "well, sometimes/maybe" rather than either one.
