Answers are grounded in this paper and cite their sources.
At a recent dinner, the conversation turned to the impressive gains in the power and ubiquity of artificial intelligence.Reflecting both the wonder and anxiety of the times, one of our guests speculated that machines would soon surpass humans at determining guilt and innocence.None claimed to know when that would happen, but nobody doubted that sometime soon -if it has not happened already -algorithms would be able to sift through and interpret the totality of evidence better than a human could.One of the guests, a law professor, questioned whether machines would ever replace human juries, and wondered how influential they would be.Suppose that a machine made a determination of guilt.How would an AI tell the story, explaining its reasoning and persuading the jury?The machine, or rather the argument based upon the machine's conclusions, would not escape harsh scrutiny by the attorney representing the accused.She would probe every weakness, exposing flaws in the data and algorithm, exploiting the opacity of underlying machine learning techniques, and appealing to emotions in ways that remain difficult for machines to match.Would juries nevertheless believe the AI?Should they?How would they weigh the arguments presented by the defense attorney against the vast knowledge but less than transparent reasoning of the machine?How could anyone be sure that the underlying algorithms were not subject to bias?It seemed unlikely that AI would succeed in court until it could explain better, appeal to emotions, and persuade jurors to shed their skepticism.And perhaps it shouldn't -an unjustified faith in the rectitude of a computer-generated determination of guilt would be even more costly than misplaced skepticism.As AI extends its power and reach, all of us will need to learn how to work with it and how to ensure that its influence on human well-being is salutary.Most importantly, we will need to ensure that human judgment not only maintains its primacy but that we, as citizens, are up to the task.That is a responsibility of educators, and one that universities need to take seriously.We increasingly ask how we can best prepare our students for a machine-driven future.My remarks focus on data science, which shares foundations and in many respects overlaps with AI but is applied to many other areas of endeavor.In fact, the pervasive use of the term "data science" in academic settings reflects both the appeal of the intellectual activities it encompasses and the capaciousness -or vagueness -of its meaning.The collection, organization, analysis, and interpretation of data can all be considered its domain.As both a descriptor and as an activity, data science has acquired an aura of power and consequence, in no small part due to its association with AI.We now take for granted that our mobile phones identify family members and friends in photos, and that those phones will warn us that we need to leave early for our next appointment because traffic is heavier than usual.Music streaming services point us to unfamiliar tunes that we find at least as appealing as the recommendations of friends.All of these Utts, J. (2003).What educated citizens should know about statistics and probability.The American
No takes yet. Share an insight, caveat, or question.
Alan M. Garber (2019) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: