The AI Ethics and Philosophy Forum at SAIS Europe conducted its first poll during the Club Fair. We have taken our time in drawing out findings, and we are excited to share the results.

Our initial goal was to settle the question of whether a machine can make art. In fact, we brought to light two conclusions which I believe are much more interesting. First, that SAIS students trust themselves to make normative decisions, and second, that they cannot be trusted with returning borrowed pens.

Every great inquiry begins with a modest instrument. Galileo had his telescope. Darwin had his finches. The AI Ethics and Philosophy Forum had forty respondents, one question, and three pens. What follows is our attempt to wring every possible conclusion from those instruments, including several they cannot support.

Methodology

Respondents were asked a single question: Is AI-produced art appropriately defined as art? The sample (n = 40) was drawn using the time-honored technique of polling whoever happened to be in the room. Critics will call this a convenience sample. We prefer to think of it as a sample of convenience.

Three pens were made available for the duration of the experiment. No pen-retention protocol was in place.

Findings

The results were decisive. 30 respondents (75%) said no, AI-produced art is not true art. 9 (22.5%) said yes. A single respondent (2.5%) refused this weighty philosophical question, preferring to let the AI answer the question on its own behalf.

Yet, the most striking finding is not the split between yes and no. The most potent finding is the near-unanimity beneath. Presented with a normative question, 39 of 40 respondents (97.5%) trusted their intuition. Whether they embraced AI art or rejected it, a clear majority of students did so on their own authority.

This, in its own way, is  proof of the Enlightenment values that give merit to the academic process. Madison’s Bill of Rights presumes a citizen capable of judgment; Paine’s Common Sense presumes that judgment is, well, common. Our survey is proud to stand alongside these two intellectual giants.

The lone deferrer deserves a word of sympathy. Some will see surrender. The respondent identified the one party with a direct stake in the outcome and invited it to testify. Whether this reflects humility or a conflict of interest, we leave to the peer reviewers.

The Pen Problem

The experiment began with three pens. It ended with one.

Here the analysis takes a darker turn. Some forty people were given access to our pens, and two pens disappeared. Assuming each thief took exactly one pen, two of forty participants are thieves. Extrapolating to the broader student body, we can deduce that some 5% of SAIS students are thieves.

This logic is not without merit, yet it rests on assumptions. A naive confidence interval, a simple statistical tool for determining the true value of our findings, of around 5% produces estimates from roughly 12% down to -2%. The upper bound is alarming. The lower bound is more alarming still, for it implies that SAIS may harbor negative thieves: students who quietly leave pens behind, restocking the commons out of pure civic virtue. We have not yet observed such a student, but we remain hopeful.

Alternative hypotheses must be considered. A single respondent may have taken both pens, which would halve the theft rate while doubling the individual’s ambition. The pens may have been borrowed with every intention of return. Finally, we cannot rule out that the respondent

who deferred to the machine on art also asked the pen whether the pens were free to take. We have no evidence for this. We include it because no faux-scientific paper is complete without one unfalsifiable theory.

Conclusion

What have we learned? First, when asked whether a machine can make art, nearly everyone prefers to decide for themselves. Second, when handed a free pen, a smaller, statistically ambiguous share of people prefer to decide on its ownership for themselves about that, too.

The two findings are correlated. Both reflect a deep faith in individual judgment, the same faith that animates the academic enterprise we are all taking part in. Madison understood the stakes: “If men were angels, no government would be necessary.” Our data suggest a corollary: if graduate students were angels, no one would need to count the pens.

The truth of the matter is clear. Human judgment remains sovereign, at least for now. The same cannot be said of our pens.