To Err is...

To err is human… and machine! Both humans and machines/AI regularly make mistakes. I have already made two mistakes while typing out this response. Last weekend, my navigation app mistakenly assumed it was reasonable for our car to teleport from the road to the middle of a building that was a block and a half away. Humans and machines are neither perfectly good nor perfectly bad at performing common tasks. Presently, humans are better at some tasks and machines at others, but the list of machine-dominated tasks continues to expand due to the rapid advancement of technology. Will machines ever be better at all tasks or are there some tasks that will always be relegated to humans? That is the key question; a fundamental question that even the experts haven’t adequately answered! There are, however, methods for supporting an imperfect answer: either extrapolate from the current technological state to some point in the future by following current developmental trends or identify fundamental differences (spiritual or otherwise) between machines and humans (just biological machines?) that will ultimately limit the performance of machines to subhuman levels for certain tasks. 

Both humans and machines can identify and correct errors. Both humans and machines are currently working hard keeping up with correcting my spelling and grammar errors. Machines are currently better at identifying simple, well-defined errors (spelling) while humans are generally better at correcting more complicated, vaguely defined errors (grammar), but will that arrangement necessarily continue? What about the most challenging tasks like characterizing humans and predicting their ultimate potential? Similarly, will machines ever be able to correct the mislabeling of human potential? It has been my experience that humans themselves do a relatively poor job in evaluating the ultimate potential of other humans for either athletic or technical endeavors. Will machines ever be able to compete for that role? Machines are already competing! AI is already being used to evaluate the potential of job applicants from their resumes; currently, most resumes are first evaluated using AI. Similarly, AI is already being combined with big data and predictive analytics to evaluate the ultimate potential of athletes. But, can AI correct the mislabeling of human potential? It already has if we believe the story behind the movie Moneyball. Billy Beane used computer generated analysis to assemble a baseball team on a budget by identifying players that had more potential that the general human consensus. Machines corrected the mislabeling of human potential done by humans! AI is still far from reaching its own ultimate potential, but it is already aiding in the evaluation of human potential and, extrapolating into the future, will continue to fulfill an ever-increasing fraction of that role.