A Hero of a Different Stripe, page 21
“We all know how critical it is to have these skills,” Salazar told the anchor, but primarily addressing the audience that was watching and list-ening from around the country. “Reliance on search engine results and denigration of the value of expertise have significantly damaged public understanding of key issues. So many in our time, looking to confirm their biases and desires, find and latch on to sources that confirm what they already believe and hope to be true. Part of the problem is that, unlike other areas in which technology has improved our lives, we haven’t yet created machines or algorithms that will do a better job of undertaking research and reporting clearly and accurately on the state of our knowledge. I believe we can do better, and need to, urgently. That is the main reason why I am launching an appeal to computer programmers to find a solution.”
Salazar told the story of what happened with the robot’s software and plagiarism. If the robot was kicked out of seminary for academic dishonesty, sooner or later people would undoubtedly find out what had unfolded, or they would make up something worse to fill in the place of the unknown truth. It was better to be frank—and to explain the specific impetus for the software engineers and programmers to craft something that did precisely what Rev2B needed to be able to do. Salazar explained that he didn’t have the resources to offer a financial prize for the most effective solution. But he had something that was arguably better: an elevated press platform and extensive reach as a result of the publicity his church’s robots had gained thus far. The programmers who cracked this problem would be solving a societal issue, not just one connected with this particular robot. They would boost their own career through the media coverage their achievement would garner.
It only took a matter of days for the solutions to start pouring in. The challenge itself was straightforward enough: create software that could produce an essay that would get a passing grade writing about the same question that the robot had been attempting in its class, using reliable academic sources that were cited correctly. Yet Salazar hadn’t anticipated quite how many computer programmers, ranging from teenagers still in high school to well-established professionals at big companies, would respond to his challenge, nor from how many different parts of the world. In the end, he scrapped his plan to either vet the results himself or have the professor at the seminary who had caught the plagiarism do so. There were simply too many entries. So he set up a page on the church website and hosted the essays there, and crowdsourced judging them. He spread the word among seminary professors, students, and other clergy to make sure that such perspectives were among those providing an evaluation of submissions. It was crucial that the various kinds of issues that might lead one to fail on the assignment—including but not limited to plagiarism—were caught and flagged, disqualifying those entries.
Soon after word about the competition had begun to spread widely, a major edtech software company contacted Salazar and opened their service to him for the purpose of vetting the submissions. They told him that this would provide them with a good test of their software’s ability not only to detect plagiarism but to distinguish between human-produced and AI-produced work, something that many academics were already concerned about, and about which not a few had voiced strong opinions in response to Salazar’s challenge. Of course, it wasn’t as though the effort to create software that could undertake and report on research was an entirely new endeavor. Salazar had merely prompted a concentrated and consolidated effort in relation to a problem that numerous programmers and companies already had their sights on. And frankly, Salazar was relieved that the edtech side of things was now directly involved in what he was doing. He didn’t want to contribute only to the side of dishonest students in the ongoing software “arms race” between them and professors. With the in-volvement of these new collaborators in the effort, Salazar was able to feel ever more excited and exhilarated about the innovation that he was playing a small role in helping to foster.
In the end, there were multiple entries meeting the established criteria that came in before a successful entry was identified and the end of the competition announced. Salazar felt it was appropriate to showcase all of the leading software in some way, even though there was one particular entry that stood out from all the rest and represented a clear winner.
The winner of the competition was announced at a press conference. Keeping his audience around the world in suspense, Salazar started by surveying some of the losing entries, some of which were amusing. They served to highlight just how challenging a task this was for computer software (and its programmers) to accomplish. One piece of software had cited several porn websites over the course of its theological essay. Another focused not on Jesus of Nazareth but a confusing variety of people named Jesus who were alive today. Yet another cited only academic sources in its footnotes, but its text sounded like it was a campaign ad for Jesus running for office.
More than a few had been misled by the widespread presence of both atheist and Christian fundamentalist websites into arguing for views that were wildly at odds with the consensus of historians and other scholars in relevant fields. Many essays gave no sign of having understood the question. Given that the question was about the relationship between the historical Jesus and the motif of the Messianic secret in the Gospel of Mark, it was only fair to point out that many human seminary students have struggled to understand, never mind answer, a question about that topic. But that was what made this programming competition a challenge and its successful entries so impressive.
Having referred to successful entries in the plural, Salazar went on to present the programmers whose submissions had crafted acceptable essays, accentuating their achievement and quoting from their essays. However, these were not the winners of the competition, he added. Finally, Salazar offered an introduction to the winner. He shared some of her life story in the manner he was accustomed to when offering a sermon illustration, tugging at the heart strings of those who were now listening with rapt attention in front of screens of various sorts around the world. The winner had created a program that not only managed to identify really good sources and cite them correctly, but use them to produce essay content that was not merely lucid but downright eloquent in places. The winner was a woman in her early twenties living in India. Her name was Jayanti Khan, and she was studying computer science at American College in Madurai.
The essay that Khan had submitted was so impressive that initially Salazar had been suspicious, asking for a second and a third opinion about the essay before taking it seriously. Even then, he arranged for Khan to be flown in from India as part of the assessment process and asked for a demonstration of the software in his presence, with technical advisors present to try to make sure that he was not dealing with a modern essay-writing version of the Mechanical Turk, with a human being hiding behind the mask. However, the software was able to craft an essay in response to an unanticipated question that Salazar posed to it and did so in a much shorter time frame than any human being could have accomplished it. Indeed, the writing alone would have taken longer, never mind the research. There was no way that this could be the work of some human being waiting for remote prompting. It was technically impossible for cheating to be involved, for multiple reasons. It was clear that the software was simply that powerful.
The eloquence of the writing, Khan explained, was the result of a machine-learning project she had already been working on prior to the competition. The English spoken in India has some distinctive characteristics that can seem odd to readers and hearers outside the country. After she submitted articles to multiple computer science journals and received rejections from all of them, she had wanted to make sure it wasn’t the language in which she expressed herself that was at fault. So she had trained a program to compare phrasing in her completed drafts of her own articles against the data set of the most cited academic articles in the field of English. The machine had no comprehension of what it was reading, of course. But skillful use of language (unless you are writing poetry that bends and breaks the rules) doesn’t involve innovation in grammar or phrasing, but merely a following of best practices when expressing your own ideas. A machine can learn that—and help a human being to learn to do it better, too.
As Salazar provided this brief bio of the winner, he felt it important to mention something else. Khan wasn’t a Christian, the minister remarked as he finally invited her to join him in front of the cameras at the press conference. Yet she would be enabling a robot to pursue Christian ministry as a seminary student. She wasn’t American, yet she created software that would benefit us and not only herself or her nation. The end result of the project, Salazar said, was a testament to international and inter-religious cooperation, and not just progress in software engineering and artificial intelligence.
With the new software installed, Rev2B continued its studies at the seminary. No more plagiarism was detected, and its grades saw an immediate improvement. By its second year of study, public and media interest had largely dissipated. Several congregations had undertaken efforts to incorporate robots into their congregational life. One example that received significant publicity featured robot musicians offering very different genres of music. A megachurch in Chicago had a robot drummer with six arms whose performances were indeed phenomenal.
However, no one seemed to be in a hurry to invest in sending another robot to seminary. If they did so, they would simply look like copycats at this stage. If Rev2B ended up failing to graduate, their robot was more than likely to meet the same end as well. There was thus no benefit to doing anything other than waiting to see what the results of this experiment would turn out to be. By this point, there was no way of snatching away the honor that would reward Salazar and his church if things turned out well. And things could still go horribly wrong. That was why everyone was apparently willing to let the risk be shouldered entirely by Salazar and Rev2B.
When the robot was in its final semester at the seminary, and successfully applied for permission to walk and receive its diploma at the next graduation ceremony, Rev. Salazar sent Jayanti Khan an email to invite her to the graduation. He had made sure to follow her career path so that he would be able to contact her when the time came. She had graduated from college two years earlier, and while she had received countless offers from some of the biggest software companies in the world, she had chosen to start her own business instead, and did so even while she was still a student. The company seemed to be doing impressively well, at least as far as he could tell as an outsider to that industry. It was only twenty minutes after he sent the email that his phone rang, displaying a longer than usual number indicating the call came from somewhere overseas. Salazar answered it quickly.
“Hello?”
“Rev. Salazar? This is Jayanti calling you from India. I was so delighted to receive your invitation to Rev2B’s graduation and would love to attend. I had been hoping to hear from you once the graduation was confirmed.” Khan paused before continuing. “I have a proposition for you and want to say before I go any further that, if you agree, I intend to show my appreciation by covering all of Rev2B’s tuition and fees.”
“That’s very kind of you to offer,” Salazar replied. “But it really isn’t necessary. The robot’s media appearances and my own book deal have already taken care of those expenses.”
“Oh, so sorry for not being clear,” Khan replied. “I meant future tuition.”
Salazar paused for a moment before asking, “What future tuition?”
“That’s what I wanted to talk to you about. I would like to know if you are open to Rev2B pursuing a doctorate.” Khan went on to explain that, with the success of her software that could synthesize existing information in the way one expects in an undergraduate assignment, the next big challenge was whether an AI could do original cutting-edge research. She and her employees had been working on this problem and felt they were making significant progress. But the real test of success this time would be whether a machine-written research article could pass peer review and be accepted by a major journal. As of this moment, no robot other than Rev2B had a legitimate institutional affiliation and an academic supervisor that would allow this to be pursued in a way that could be documented and studied.
Salazar wasn’t sure what the ramifications of pursuing this might be. But he knew that Khan had turned their robot student in danger of expulsion into one that made the dean’s list each academic year. If he tried, he could probably think of some reasons that he could legitimately refuse, excuses he might offer that might have some validity. But he knew immediately that he did not want to refuse.
“How can I refuse, after everything your software has done for us?” Salazar asked, pausing for only a second before continuing. “But if you don’t mind, I have long wanted to ask you something. I know you’re not a Christian. If the robot goes on to graduate school, it will presumably be in theology or something related to that. Goodness, it would probably be simpler if it could do work in chemistry, wouldn’t it? Research in the natural sciences, or perhaps mathematics, would seem more up a robot’s alley. But I suspect that you’re actually more interested in it trying to write a thesis in the humanities—that’s the real challenge for programmers now, isn’t it? But anyway, as I started to say, with you not being a Christian, I have long wondered how you feel about your software turning a robot into a Presbyterian minister.”
There was a long enough pause that Salazar began to worry that either they had been cut off or that he might have offended her. But when Khan spoke, she sounded as though she were speaking from a source of personal conviction, and with some enthusiasm for the topic.
“My father is a Muslim, and my mother is Hindu. They brought me up to respect all beliefs, and in both of their traditions Jesus is a respected figure. On a professional level, I would have been happy if my path to success had come from just about anywhere. But on a personal one, I am so happy for the way this all came together. You said it yourself at the press conference where you announced that I had won, although at that point you hadn’t asked me about my perspective on the matter. This project thus far, and what I am proposing to carry it further, doesn’t just illustrate the possibility of international and interfaith cooperation. It illustrates the connections between computer science and the humanities, between faith and reason. I’m not even thinking of your religious faith when I say that. Every step of the way, as I have heard the story told, you have stepped out into the unknown in a way that didn’t expect miracles but did depend on diverse others rising to the challenge of the situation with hope, trust, willingness to risk, and goodwill. You didn’t act irrationally, but you also dared to set out in uncharted directions over and over again. That’s what I want to see continue and enhanced in the next stage of things. What do you say?”
Salazar agreed, and after chatting some more, they ended their conversation. As Salazar set his phone down, he scanned the books that filled the shelves around his office walls until his eyes alighted on a volume on interreligious dialog and interfaith cooperation. He got up, pulled it from the shelf, and turned to the index. As he suspected, there was no entry on robots, none on computers, not even one on technology. They were indeed in uncharted territory—and that was not just because the book was now more than a decade old. For all our penchant for innovation and creativity, for progress, human beings are still a highly risk-averse species.
Salazar sat down in a different chair on the other side of the office, leaned back, and looked out the window at the barren branches of the old oak tree nearby. They stood as though poised, waiting for a signal to indicate that these last few weeks of early spring chilliness had at long last passed, and now they can start to bud. People have argued that technology divides people and that it brings people together, Salazar mused silently. People have had the same arguments about religion. The answer was plainer to him now than it had ever been before.
Technology and religion don’t do either one of those things—not on their own, at any rate. A phone doesn’t connect or divide people while lying neglected on a desk. But it can do either, depending on how it is used. The same is true of robots, and religion, and just about anything else for that matter. But when we don’t want to change our habits as a society and we find ourselves disconnected from or in conflict with one another, we blame our tools for shortcomings that reside within ourselves.
By the time the first robot in history walked across the stage of the seminary’s auditorium, it had already been accepted into the DMin program. Its proposed thesis topic was to research the history and prospects for the future of the congregation where it had been serving as assistant minister for the past four years. The foundation of the research project would utilize significant amounts of sociological as well as historical data, allowing the robot to draw on its computational strengths to crunch numbers and trawl enormous swaths of data that no human being could ever hope to get a handle on. Upon that foundation, the robot would then face the challenge of engaging in a more humanistic undertaking. Could it not only detect neglected patterns in this tiny slice of religious history but also interpret them in a manner that would allow it to make a recommendation about what the congregation ought to do in the future?
By the end of the first year, the robot had indeed identified correlations that no one had previously been fully aware of. Patterns in the tiniest fluctuations in Sunday morning attendance that consistently presaged larger shifts soon to follow. Correlations between church membership numbers, how much people gave to the church, and demographic shifts in the surrounding neighborhood, city, and nation. The impact of the arrival of a new minister, the addition of staff members such as ministerial assistants, and other kinds of events, both positive and negative, in the life of the church.
