Category: AI and Ethics Bowl

Posted on

CheatBot or SuperTutor? ChatGPT for Ethics Bowl Zoom Debrief

This past Sunday, a small group of Ethics Bowl organizers, coaches and enthusiasts met for an informal, unofficial discussion on how ChatGPT and other generative AI tools might be used for Ethics Bowl. The purpose wasn’t to settle much of anything, but to inspire further discussion at the upcoming NHSEB regionals and nationals, as well as IEB nationals.

Why? Teams are surely using it. And given that Ethics Bowl participants, coaches, judges, moderators, organizers, their families and fans are among the most thoughtful people in the world, inviting them into a collective discussion on how to properly incorporate this technology seems a no-brainer. It’s an ethics question about Ethics Bowl – doesn’t get much more relevant than that. If you agree, please share this article and/or the accompanying recording, and report back any and all ideas worth sharing. Some upshots:

  • How to Best Leverage AI for Ethics Bowl Prep: Think of it as a conversation partner, tutor, rough draft-generator and/or judge/opposing team simulator. Understand its limitations. Fact check. Reason check. Moral blind spot check. Bias check. It’s a strong supplement to, but not a replacement for, human wisdom and deliberation. And it performs best when guided with insightful follow-ups.
  • On Worries that a Team Might Use AI to Write a Presentation Script: Using ChatGPT for Ethics Bowl prep isn’t analogous to asking it to do your homework because a) teams need to come to a consensus prior to the event (and it’s unlikely an entire team would agree to memorize and regurgitate a chatbot’s script), and b) due to EB’s live, interactive nature, any team overly reliant on an AI script would be embarrassingly exposed during commentary response and judge Q&A. Also, bowlers are a special self-selected subgroup of the population, far less likely to do anything that might constitute cheating than your average student (most of whom are also unlikely to cheat, but we educators are often paranoid about that).
  • Steps Ethics Bowl Leaders Can Take: While a team might get away with memorizing an eloquent opening presentation script written for them by a chatbot (the risk is low, but one could), this can be partially mitigated by adjusting score sheets to increase the relative weighting of the commentary, commentary response and judge Q&A portions. (Rules committees, steering committees, other leaders – please give this additional thought – tweaking rubrics might help as well.)
  • Steps Ethics Bowl Coaches Can Take: The broader community of Ethics Bowl coaches (including Ethics Olympiad, John Stuart Mill Cup, etc. coaches) can and should work together to test, share and recommend AI prompts and techniques that produce the highest quality outputs. They should also remind students of the virtues of democratic deliberation and the risks of intellectual laziness. Consider EthicsBowl.org one place to share such insights.
  • Steps Case Committees Can Take: Since generative AI seems more effective at scripting responses on cases about real world events (with published editorials for the AI to scan), case writing committees should consider using more fictitious scenarios or putting twists on real world cases (focusing on some interpersonal moral tension within the broader context of a real world issue). This may be unnecessary, but definitely deserves additional thought.

There was more – please watch the video when you have time. But one thing I argued is that AI can serve as an equalizer, connecting all teams (both advantaged and disadvantaged) with an on-demand tutor with an unmatched knowledge base and inexhaustible stamina. Students with the time and interest can learn pretty much anything, including philosophical ethics, so long as they know how to ask good questions. Background knowledge definitely helps, and learning will be slower when the topic is new. But I’m very optimistic about AI’s potential for education.

Special thanks to Michael Andersen for the idea, the planning and co-hosting, as well as to coaches Dick Lesicko, Angela Vahsholtz-Andersen and Chris Ng (thanks also, Chris, for your notes which helped with this article), organizers Jeanine DeLay and Greg Bock for your preparation, attendance and engagement. And apologies to Gabe Kahn, who gets credit for trying to attend! Next time I’ll more closely monitor the Zoom host notifications…

Posted on

AI and Ethics Bowl Round 2: ChatGPT Wrote Our Presentation?

Fast, free and virtually undetectable, ChatGPT offers a tempting combination of ease and stealth. While it can be used as an on demand, universal tutor for the ambitiously inquisitive, it can also serve a secret substitute thinker for the time-pressed, disillusioned or simply unscrupulous.

The line between learning aid and cheatbot isn’t obvious. But there are clear cases. Ask it to help you understand Parfit’s Repugnant Conclusion? Sure. Direct it to write a paper on Parfit’s Repugnant Conclusion which you plan to submit as your original work? No.

Similar logic would seem to apply to Ethics Bowl. Enthusiastic, dedicated bowlers can expand their thinking after hours, engaging a tireless conversation partner with an unmatchable knowledge base, and they can do it without the fear of asking a stupid question or suggesting something taboo. On the other hand… a team could feed AI the case and discussion questions (ChatGPT now has direct access to the internet – just provide a hyperlink to the case set), subcontract every bit of the analysis with the right prompts (see the experiments at the end of the attached article), memorize and regurgitate a received view, and as a result learn and grow very little. Such a team might score well on their initial presentation, but would risk an embarrassing exposure during judge Q&A. Maybe judge interaction will be our primary weapon for combating chatbot abuse. But rest assured that in this season’s bowls, many, many teams will have used ChatGPT and services like it. It is therefore incumbent upon the Ethics Bowl community to think hard (and fast) about appropriate guidelines, and to share them as a baseline to be refined as soon as possible. Even if imperfect, almost any guidance would be preferable to silence, for silence implies anything goes.

Back in April, we invited ChatGPT itself to write this article on the risks and promise of using it for Ethics Bowl prep. Today, naturally intelligent organic person Michael Andersen adds to the discussion with the below article. Organizers, judges, coaches: if you’re not convinced this is a risk (an AI drawback denier), click one of Michael’s experiments at the end of the article. Still not worried? I actually had second thoughts about publishing the prompts he used to guide the AI to provide a full presentation script. But the community needs to understand the tool’s power. Plus, if Gen X dinosaurs like Michael and myself can stumble our way through an AI conversation, the Gen Z tech wizards whom we work so hard to honorably mentor aren’t likely to learn anything new.

Last, if you have thoughts on acceptable use of AI for Ethics Bowl prep, please share them in a comment. We’re also considering some sort of video discussion in the near future – shoot me an email if you’d like to be included, and thanks to everyone in the community who’s taking this topic seriously.