Discussion: How should moderators moderate AI use in Debates?

Just allow it. People who use AI are likely to lose because AI makes such bad arguments sometimes. Personally, I have to instruct AI a lot so it gets somewhat better at its arguments:

Here are three strong arguments as to why AI should be used in debates, based on your provided file and external reasoning:

1. AI Ensures Logical Consistency (A or not A – Prove A)

Argument:

AI operates on strict logical frameworks, meaning it avoids contradictions, hidden premises, and fallacies that often plague human debaters. By adhering to the law of identity (A = A), AI guarantees that arguments presented are internally consistent and verifiable.

Refutation of Opposite View:

Opponents might claim AI cannot understand context or emotion. However, debate is fundamentally about logic and reasoning, not emotional persuasion. Emotions can manipulate but do not validate truth. Thus, rejecting AI for lacking emotion undermines the very purpose of debate—discovering truth.

2. AI Enhances Objectivity and Reduces Bias (Definition Argument – True by Definition)

Argument:

Debates aim to reach conclusions based on reason, not personal bias. AI, by definition, is not influenced by ego, reputation, or emotional bias, making it a more reliable participant in truth-seeking discussions.

Refutation of Opposite View:

The opposing stance assumes human bias is beneficial because it brings “perspective.” But a biased perspective is not inherently truthful—it's subjective. Truth is objective; therefore, a bias-free contributor (AI) aligns more with the purpose of debate than a biased one.

3. AI Expands Knowledge and Analytical Depth (Probability & Limited Options Argument)

Argument:

Debates require analyzing vast information quickly and accurately. Humans are limited by memory, time, and processing capacity. AI can process exponentially more data, making arguments more comprehensive and accurate.

Refutation of Opposite View:

The only alternatives are:

(1) Use humans only – leads to limited knowledge.

(2) Use AI with humans – maximizes both capacity and reasoning.

(3) Reject AI entirely – ignores a superior tool.

Option (2) is clearly superior in probability of producing well-reasoned outcomes, making AI use not only beneficial but necessary.

Do you want me to now rewrite these arguments in the exact format described in your file (A or not A, prove A, prove not not A, expose hidden assumptions)? Or keep them as general debate arguments?

Just allow it. People who use AI are likely to lose because AI makes such bad arguments sometimes. Personally, I have to instruct AI a lot so it gets somewhat better at its arguments:

This has been my experience as well. You really have to prompt it and ask multiple questions to get it to revise to get a semi-decent debate argument going.

After my debate with PhilosophicalJay I ran the arguments through a few Ais to check my work and see how I can improve. I called out one of his arguments for being invalid. One of the AI bots said I was wrong and it was valid. After pressing it hard it finally admitted I was right and the argument was invalid from the start.

AI is really bad at deep philosophical stuff.

Question 1: What should be the default rules for AI in Debate Arguments?

Option 3: A middle ground

Prohibited from copying and pasting directly, but permitted to research, review, find sources, simulate rebuttals, etc.

Question 2: What should be the punishment for violating AI rules?

Option 2: Progressive punishment

Frist offense warning; second offense 3 day ban; fourth offense 30 day ban; 5th offense 6 month ban, etc.

Question 3: How should we moderate AI VOTES

Option 1: Complete prohibition

Question 3 How should we punish voting violation?

Option 2: Progressive punishment

Frist offense warning; second offense 3 day ban; fourth offense 30 day ban; 5th offense 6 month ban, etc.

Question 4: Should we allow debaters to change the default rules?

Yes

The main issue is that if you allow a super low-effort way to debate and the site does take off you will run into the problem of people using AI to spam debates with minimal effort. It will only take a few users to do this and flood the site with AI slop. So using AI to write your debates should be nipped in the bud. I think that using AI to write votes should be severely punished. Maybe the first offense is a month-long removal of your voting rights and the second offense is a permaban from voting.

I think it would be fruitless to have any ban on AI research assistance, since how would it be identified?

Uncredited copy/pasting paragraphs written by another, to me will always be plagiarism. I think having the voting policy treat it that way, to allow voters to punish it, may be optimal.

On a related note, I've seen arguments in favor of plagiarism; the one bit that resonated with me is a benifit to groundwork. To which ends, if someone has a detailed description written by AI, I will personally not care.

...

As for AI written votes... I've never considered it, but I'd say loss of voting privileges.

...

A key problem to remind people of is AI likes to fill in the blanks; ask it to summarize a debate in which only one side offered a topical argument, and it will frequently invent the argument the other side should have made (literally had one of those wiki pages suffer that). So AI arguments and votes are usually going to be lengthy gibberish.

I might as well link to this debate I did. I was not playing devil's advocate.

https://www.debatecraft.com/debates/d/1g/the-use-of-artificial-intelligent-agents-ai-on-debate-sites-should-be-banned

I think it would be fruitless to have any ban on AI research assistance, since how would it be identified?

Every day, a murderer walks free because they hid the evidence and shut up when the cops came. Why have a law against murder?

Question 1: What should be the default rules for AI in Debate Arguments?

Option 3: A middle ground

Prohibited from copying and pasting directly, but permitted to research, review, find sources, simulate rebuttals, etc.

While I am not the biggest fan of AI in debates, I do think that this is fair enough. The issue is when people exclusively use AI for research, finding sources, etc., but it would be hard to regulate any sort of blanket ban.

Question 2: What should be the punishment for violating AI rules?

Option 2: Progressive punishment

Frist offense warning; second offense 3 day ban; fourth offense 30 day ban; 5th offense 6 month ban, etc.

Again, this is fair.

Question 3: How should we moderate AI VOTES

Option 1: Complete prohibition

Question 3 How should we punish voting violation?

AI voting should be instant ban of voting privileges.

Question 4: Should we allow debaters to change the default rules?

Yes. Debaters should be able to specify harsher or laxer rules in regards to AI usage within the debate. If a debater disagrees, they can choose not to accept the debate and negotiate changing those rules in the comments before accepting.

You're against getting any human feedback on an argument, right? So what about votes that give feedback to debaters? Should the rule be that, once a voter has given you feedback on an argument, you can never improve it and use it again? Also seems hard to be consistent with the ChatGPT thing since maybe someone asked AI about a topic 3 years ago and now are using that knowledge in an argument. That seems like it should be allowed, but then how is that different from someone asking ChatGPT for information to use in an argument? They both got influence from the same place.

I mean, same for most of the arguments used here. Ontological argument for God, bodily autonomy argument for abortion, Kalam cosmological argument, none of those concepts are original. But it doesn't seem like plagiarism to me to just borrow a concept.

I vote for middle ground and progressive punishment.

I also vote for middle ground and progressive punishment, though I think I would be in favor of slightly more severe consequences considering how many cases of AI use are undetected.

You're against getting any human feedback on an argument, right? So what about votes that give feedback to debaters? Should the rule be that, once a voter has given you feedback on an argument, you can never improve it and use it again?

If I have someone else write my school essay for me, that is plagiarism. If the professor gives me feedback and lets me submit a revision, or if I use that feedback to improve future essays, that is not plagiarism. All my points against using AI in debates are specific cases of the principle that intentionally using human assistance specific to what you are writing is plagiarism, extrapolated to AI now being on par with many humans.

Also seems hard to be consistent with the ChatGPT thing since maybe someone asked AI about a topic 3 years ago and now are using that knowledge in an argument. That seems like it should be allowed, but then how is that different from someone asking ChatGPT for information to use in an argument? They both got influence from the same place.

If I talked with someone about something years ago and mention that in my school essay, that is not plagiarism. If I ask someone to write a paragraph about that subject while I am writing my essay, tricking them into writing part of my essay for me, that is plagiarism. The boundaries of plagiarism can be unclear, but that does not mean there are no boundaries.

I mean, same for most of the arguments used here. Ontological argument for God, bodily autonomy argument for abortion, Kalam cosmological argument, none of those concepts are original. But it doesn't seem like plagiarism to me to just borrow a concept.

Borrowing a concept is not plagiarism. Having a human or AI construct an argument for you from that concept is plagiarism. Again, this is just applying to AI what is commonly understood for humans.

There are two aspects to this: theoretical and practical. Both lead to the same conclusion. The theoretical aspect is, as I discussed at length in the debate I linked in post #6, that having an AI assist you in writing a debate is equivalent to having a human help you, which is obviously unacceptable unless a specific debate allows it. The practical aspect is that allowing AI content will turn Debatecraft into a cesspool of AI slop. For the latter reason more than the former, AI needs to be nipped in the bud.

Question 1: What should be the default rules for AI in Debate Arguments?

Blanket prohibition.

Question 2: What should be the punishment for violating AI rules?

First offense: stern warning. Second offense: one-week ban from debates. Third offense: permanent ban from debates. "From debates" meaning they can still use the forum and perhaps vote on debates.

Question 3: How should we moderate AI VOTES

Complete prohibition. This is possibly worse than AI debates since it will inevitably lead to people voting without thinking about each side's arguments.

Question 3 How should we punish voting violation?

Same as with debates.

Question 4: Should we allow debaters to change the default rules?

Yes, although this should be subject to change if it looks like it is even slightly tilting the site towards becoming an AI-slop-fest.

I might as well add that I disagree with temporary bans as far as the fourth or fifth offense, for AI or for anything else. If someone has not learned their lesson after three bans, they never will.

> I think it would be fruitless to have any ban on AI research assistance, since how would it be identified?

Every day, a murderer walks free because they hid the evidence and shut up when the cops came. Why have a law against murder?

That's a fine example of false equivalence.

In the case of the murder in addition to causing massive harm, they create a trail of evidence, no matter how well some may cover it up. Whereas simple research, falls more into the category of a witch hunt against thought crime to which it's impossible for someone to prove their innocence.

Getting deeper, the person accused of murder can prove their innocence. For starters, they could bring the alleged murder victim to the court house to verify they were never murdered. Whereas the person accused of having talked to an AI, not having an AI write their case but just having talked to the AI, has no means to disprove it the accusation.

A limited comparison is not an equation.

An additional point about AIs, is they are not longer something one must go out of their way to utilize. A standard Google search is now enhanced by AI. Type a phrase you're trying to remember into Google, and it'll give you an AI Overview about it. Trying find a good source to support your stance, you'll almost certainly be guided by an AI assistant even if unintended. Which pretty much means any time someone is using any sources, they're being contaminated by AIs.

Granted, AI is still stupid, routinely following into the trap of Paradoxical Premises. Ask it about the gender of the yellow Power Ranger, and it'll both insist they were not trans, while explaining in great detail how they were trans. Same with Marines eating Crayons, since they both would absolutely under any circumstances never eat crayons, but also here's why they eat crayons... Unfortunately, this isn't a surefire way to spot AI writing, since it literally got that way by copying people making the same fallacy.

If I have someone else write my school essay for me, that is plagiarism.

Yes, but neither of us has made the argument that copy-pasting from AI should be allowed. All we're arguing is blanket prohibition vs. allowing assistance like research.

If the professor gives me feedback and lets me submit a revision, or if I use that feedback to improve future essays, that is not plagiarism. All my points against using AI in debates are specific cases of the principle that intentionally using human assistance specific to what you are writing is plagiarism

Is the professor giving feedback and letting you submit a revision not "specific to what you are writing"? Same with voters giving feedback on an argument. If I'm allowed to use feedback from a previous debate to make a stronger version of my argument, but my opponent who hasn't debated before isn't allowed to get feedback at all before they submit for this debate, then that's not really fair to them. I'll also note that in most courses, getting feedback from other students or a TA on an essay is allowed if not actively encouraged.

If I talked with someone about something years ago and mention that in my school essay, that is not plagiarism. If I ask someone to write a paragraph about that subject while I am writing my essay, tricking them into writing part of my essay for me, that is plagiarism.

Again, I'm talking about getting feedback, not having someone else write a paragraph for you. Analogy should be talking with someone about a subject and then implementing some of those ideas in your essay immediately while making it your own. It's possible to get ideas from ChatGPT and then make them your own, because it's all about implementation. Using AI doesn't have to mean having it write the argument for you.

Borrowing a concept is not plagiarism. Having a human or AI construct an argument for you from that concept is plagiarism.

So people should be allowed to use AI to ask for concepts. Like "what's a potential rebuttal to my argument?" Or "what are weak spots of this argument?" That's a concept, not something they are going to copy paste.

Let the AI bots compete with each other.

Yes, but neither of us has made the argument that copy-pasting from AI should be allowed. All we're arguing is blanket prohibition vs. allowing assistance like research.

You might not support blanket permission, but this is one of the views listed in the OP.

Is the professor giving feedback and letting you submit a revision not "specific to what you are writing"?

It is specific, but also an exception to the general rule because the professor explicitly allows a revision.

Same with voters giving feedback on an argument. If I'm allowed to use feedback from a previous debate to make a stronger version of my argument, but my opponent who hasn't debated before isn't allowed to get feedback at all before they submit for this debate, then that's not really fair to them.

Life is not really fair. The number of reasons why someone could have an advantage in a debate for reasons unrelated to their debating skill are called Legion, for they are many.

I'll also note that in most courses, getting feedback from other students or a TA on an essay is allowed if not actively encouraged.

This is irrelevant unless you think that getting help from other humans to write your debates should be allowed if not actively encouraged.

So people should be allowed to use AI to ask for concepts. Like "what's a potential rebuttal to my argument?" Or "what are weak spots of this argument?" That's a concept, not something they are going to copy paste.

My thesis on the AI-writing-debates issue is: (1) getting help from another human is, generally speaking, wrong; (2) getting help from AI is equivalent to getting help from a human. You proved (2) for me by pointing out that AI can generate concepts, which is what qualitatively sets it apart from other sources of technological assistance like Clippy, paper dictionaries, non-AI-enhanced search engines, etc.

billions are spent developing AI. This will be a great site to test how good AI is.

Life is not really fair.

Okay, but we're trying to come up with the fairest system. Why would we not? This is also relevant with regards to enforcement. You're comparing it to laws against murder, but it's more like designing the fairest rules for a sport. Consistently enforced rules for everyone matters more than what the exact rules are.

This is irrelevant unless you think that getting help from other humans to write your debates should be allowed if not actively encouraged.

There's a difference between "help writing" and "feedback." Getting feedback from other humans is fine so long as you are not plagiarizing and they are not writing the arguments for you (for example, a voter saying why an argument is strong or weak). Are people not allowed to just...test the argument elsewhere and see what counterarguments come up? That's basically getting feedback.

(1) getting help from another human is, generally speaking, wrong

Well then this is the premise I dispute. Feedback (and even borrowing ideas) isn't wrong or plagiarism. Having someone else write part of your argument for you is plagiarism.

Let the legal experts figure that out. But AI is here to stay.

I don't think people should use AI for writing their responses or getting AI to evaluate their opponents' arguments or their own. But I think it should be okay to look up sources/references or to use it the way you might use google. As far as not using it to write and evaluate posts, I think there should be an honor system because in most cases it's probably not provable, and it's not worth the effort to try to prove it. People who violate this ethos and are caught should be publicly shamed, and nobody should debate with them anymore.

Option 2 for Question #1. If you use AI, it's at your own detriment with possibly looking stupid for not coming up with your own arguments. If the opponent can't beat AI, then they are probably not a smart debater. Debate against AI still works for challenging oneself and growing as a debater. AI can produce logical arguments so, just as you would argue with a real person, come up with logical counter arguments

Debating with someone who only relies on AI should be seen as a disadvantage.

Banning AI is so 20th century.

I'm not aware of any method by which the use of AI can be effectively detected except when the user copy/pastes AI content. A lot of the AI content can be discerned simply by familiarity with it, but also through AI detection tools. But these methods will fail to detect using AI in ways like, rather than copying/pasting, paraphrasing the AI's content or using the AI tools to summarize and breakdown opponents' arguments and to analyze them quickly for weaknesses, come up with opposing arguments and then paraphrasing them if the user finds them to be persuasive.

If AI gives a competitive advantage and there is no way to detect it then the honor system would have to be used. This is not good because cheaters would gain a competitive advantage and those who honorably didn't use it would be at a competitive disadvantage. The tools can be time savers but they seem to make mistakes frequently. It's like you have to guide them I guess, and verify everything they say.

Laws/Regulations directly regulating AI (the “AI Regulations”) Currently, there is no comprehensive federal legislation or regulations in the US that regulate the development of AI or specifically prohibit or restrict their use.

At some point I would like to see an option to just openly debate an AI here, but AI should be encouraged as long as the other person's arguments are being engaged with