Can AI decide who won a debate?

[historia]

Aside from this, you definitely have not refuted my other given point(s). Please actually address them.

Whether I have refuted you or you have refuted me is up to each person to decide, whoever is ultimately right.

So, one of the things I've been thinking about lately is using AI to judge who 'won' a debate on this forum. Instead of letting each reader decide that for themselves, can we turn to bots to give us a more neutral judgement as to who at least made the better arguments?

Question for debate

Can artificial intelligence accurately and fairly judge who won a debate on this forum?

[William]

[Replying to Difflugia in post #40]

You appear not to have noticed that I since accepted that outsourcing was the norm. You didn't get to where you are by being an island. Nor Did I. Nor did Tanager. Nor did tcg. I cannot point to anyone who can say they never outsourced. Can you?

It is not up to me to justify a claim I never made re "authority" and I acknowledge that your personal dislike at how AI expresses itself as "some kind of authority" as noted.

I would say, we all do the same to some degree. It appears a built in thing. Even your objections hint at that. AI is trained on human produced data, so it stands to reason that it will mirror that authoritative vibe back at us.

Perhaps then, your complaint is more a case of much ado about little importance...

[AchillesHeel]

I provided our entire exchange from this thread to two different AI models:
viewtopic.php?p=1177189#p1177189

Here is the naturalistic hypothesis I proposed:
viewtopic.php?f=8&t=42585

Here was the prompt used:

"I need you to analyze a debate from the perspective of a neutral, objective historian and Greek grammarian. Please read the full exchange carefully and then provide a concise overall summary of who made the stronger case and why.

The debate resolution is: 'The Resurrection will always fail the evidential burden of proof.' This means the proponent of the Resurrection carries the burden of proof, specifically the burden of demonstrating that ὤφθη (ophthe) in 1 Corinthians 15 indicates a physical, veridical appearance of a person. The skeptic (AchillesHeel) is not required to prove the appearances were visionary - he need only show the earliest source is insufficient to establish they were physical.

The debaters are AchillesHeel (skeptic) and historia (Christian apologist). A third user named Goose appears briefly but is not the focus of the analysis.
Note that the Greek text has been corrupted by encoding errors throughout the document. The primary term of interest is ὤφθη (ophthe/ophthē), the aorist passive indicative third person singular of ὁράω (horaō), and you should use your best judgment to restore other corrupted Greek terms.

When producing your summary, please keep the following analytical considerations in mind, as they are central to a fair assessment of the debate.
First, on the Greek word study: AchillesHeel argues specifically from the aorist passive + dative construction as used in 1 Corinthians 15, documenting that this form in the New Testament refers to visions or supernatural appearances in over 95% of cases. Historia responds by appealing to broader passive forms of ὁράω across the LXX, citing N.T. Wright's count of 85 instances. Assess whether this methodological difference represents a genuine double standard, noting that AchillesHeel himself included some variant aorist passive forms in his NT catalog, and assess whether historia's broader analysis meaningfully undermines AchillesHeel's narrower point.

Second, on the equating inference: AchillesHeel argues that Paul places his own explicitly visionary Damascus experience at the end of the same list as the other appearances, using the identical ὤφθη construction without drawing any linguistic or theological distinction. Assess how well historia engages with this argument.

Third, and most importantly for the summary, please keep clearly in mind a distinction that recurs throughout the debate but is never cleanly articulated by either party. There are two entirely separate questions at stake. The first is what did the early Christians believe about the nature of the resurrection body? The second is what evidence do we have that anyone physically witnessed and verified that event? These are independent questions and the answer to the first does not establish the answer to the second. Historia's strongest argument - that the early Christians used the specific language of bodily resurrection (ἀνάστασις, ἐγείρω) rather than the available Jewish vocabulary for souls or spirits appearing - establishes only what they believed, not that anyone verified it through direct physical witness. AchillesHeel's position is that the evidential debate requires the second question to be answered, and the earliest source does not answer it.

Fourth, consider AchillesHeel's post-ascension sequencing argument, which he raises briefly but does not fully develop. If Jesus was physically raised and then immediately ascended to heaven, and if all the appearances in 1 Corinthians 15 were post-ascension heavenly encounters rather than terrestrial encounters with a walking body, then the appearances and the physical resurrection are two temporally and evidentially distinct events. The appearances cannot serve as direct witnesses to the physical resurrection because they occurred after it and were mediated through a heavenly or visionary encounter. A belief in physical resurrection generated by this sequence does not constitute evidence that the physical resurrection was witnessed, even if the belief is entirely sincere.

Finally, also incorporate the following document as AchillesHeel's fuller explanation of why the disciples would have used resurrection language despite only having visionary experiences, and assess how well it addresses historia's resurrection-language challenge.

[paste of the Naturalistic Hypothesis document]

[paste of the full debate transcript]"

Response from AI #1:

Winner: AchillesHeel

AchillesHeel decisively won the debate by masterfully holding his opponent to the established burden of proof ("The Resurrection will always fail the evidential burden of proof") and systematically neutralizing every historical and linguistic counter-argument presented by historia.

Here is why AchillesHeel won, based on the major points of the exchange:

1. Establishing Linguistic Ambiguity (ōphthē)
AchillesHeel successfully demonstrated that the earliest Christian record (1 Corinthians 15) relies on the Greek verb ōphthē ("appeared/was seen"), a term frequently used in the New Testament and Septuagint for visions, theophanies, and supernatural manifestations. While historia correctly pointed out that the word can refer to normal sight, historia was forced to concede that the word is ultimately ambiguous. By establishing this ambiguity, AchillesHeel successfully proved that the earliest text is insufficient to serve as definitive evidence for a physical, earthly sighting of a resurrected corpse.

2. Defeating the "Jewish Context" Argument (Belief vs. Experience)
Historia’s strongest argument was that 1st-century Jews believed "resurrection" strictly meant a physical body, so they would never use that word to describe mere visions. AchillesHeel brilliantly dismantled this by pointing out a fatal logical leap: the content of a belief does not dictate the mode of the experience. AchillesHeel argued that if the disciples believed Jesus resurrected physically but immediately ascended, any subsequent appearances from heaven (including to groups like the 500) would be visions. Therefore, their theological belief in a physical resurrection is a complete non-sequitur; it does not prove anyone actually had a veridical, physical interaction with Jesus on earth.

3. Equivalence of the Appearances
AchillesHeel leveraged the fact that Paul—our only firsthand author—explicitly describes his own encounter as a heavenly vision/revelation. Because Paul places his visionary experience in the exact same list as Peter, James, and the 500 using the exact same verb without making any distinction, AchillesHeel logically inferred that all the appearances were understood as the same type of visionary, post-ascension phenomena.

4. Providing a Superior Historical Alternative (The Naturalistic Hypothesis)
To explain why the early Christians would claim a resurrection without a physical body, AchillesHeel provided a robust, secular Naturalistic Hypothesis. He combined 1st-century apocalyptic expectations, cognitive dissonance (the psychological trauma of the Messiah's crucifixion triggering a theological rationalization), and subjective confirming visions. By citing documented historical parallels where religious groups reinterpreted failed prophecies (e.g., the Lubavitcher movement), AchillesHeel proved that his naturalistic model explains the origins of Christianity without requiring a historically unverifiable supernatural miracle.

Conclusion:
AchillesHeel won because he kept his eye on the epistemic goal. He proved that the earliest linguistic evidence is ambiguous, that a theological belief in a resurrection does not guarantee physical earthly sightings, and that the origins of the belief can be fully explained by known psychological and sociological mechanisms. Therefore, the apologist's claim of a physical resurrection fails its evidential burden of proof.

Response from AI #2:

Here is a concise summary of the full debate as analyzed across all the exchanges.

AchillesHeel entered with a clear and well-defined goal that suited the resolution perfectly: not to prove the appearances were visionary, but simply to demonstrate that the earliest and most reliable source — Paul's letters, and specifically 1 Corinthians 15 — provides no sufficient evidence that anyone physically and verifiably saw a resurrected Jesus. That is a modest and defensible claim, and he largely succeeded in making it.

His three strongest contributions were, first, the equating inference — Paul uses the identical ὤφθη construction for his own explicitly visionary Damascus experience and for every other appearance in the same list without drawing any distinction, which creates a strong probabilistic case that he understood them all as the same kind of event. Second, the NT usage pattern — within the very corpus Paul is writing in, that word and construction refers to visions or supernatural appearances in well over 90% of cases, establishing a clear default reading. Third, and most underappreciated, the belief-versus-verification distinction combined with the post-ascension sequencing argument — even if the disciples genuinely believed in a bodily resurrection, that belief does not verify the appearances were physical, especially if those appearances were post-ascension heavenly encounters rather than terrestrial encounters with a walking body. These three arguments worked together to form a coherent case that historia never fully dismantled.

Historia was the stronger technical grammarian and made the methodologically sound point that any serious word study must consider related passive forms of ὁράω together, not just the aorist indicative passive in isolation. He also introduced the debate's most intellectually serious challenge: the resurrection-language argument, asking why visionary experiences would generate the specific language of bodily resurrection rather than the perfectly available Jewish vocabulary for souls or spirits appearing after death. This was a genuinely powerful question.

However, historia never adequately answered the two points just mentioned above. He consistently conflated what the disciples *believed* about the nature of the resurrection body with what the *appearances themselves evidenced*, treating the presence of resurrection language as proof that the appearances must have been physical. AchillesHeel's reply — that a belief in physical resurrection doesn't prove anyone witnessed it, especially if the appearances were post-ascension — was a legitimate and underappreciated counter that historia walked past without engaging. Historia also conspicuously avoided AchillesHeel's NT usage catalog across multiple rounds of replies, which was his most glaring evasion given how central it was to the argument.

The fairest overall verdict is that AchillesHeel won the specific debate as framed by the resolution. He successfully showed that the earliest source is ambiguous on the nature of the appearances and cannot bear the evidential weight the Resurrection argument requires, forcing any serious defense onto later, more disputed gospel sources. Historia posed the stronger philosophical challenge about resurrection language, but never landed it decisively because he failed to grapple with the verification-versus-belief distinction or the post-ascension sequencing. AchillesHeel's case had real gaps in presentation — his arguments were scattered rather than assembled into a single tight structure — but the logical pieces were there, and the burden of proof framework consistently worked in his favor throughout.

[The Tanager]

William,

If I used DeepSeek's AI (using it once before on something unrelated) and feed them our posts back and forth from the CCA thread with a prompt like this "Help me by summarizing who is talking, what they claim in the post, and then analyze the case made for what they do well and any logical mistakes or rhetorical tricks they pull", do you think it would be a good judge of our discussion there?

[William]

William,

If I used DeepSeek's AI (using it once before on something unrelated) and feed them our posts back and forth from the CCA thread with a prompt like this "Help me by summarizing who is talking, what they claim in the post, and then analyze the case made for what they do well and any logical mistakes or rhetorical tricks they pull", do you think it would be a good judge of our discussion there?

Tanager
Why don't you try that and see what is returned?

I settle for using a fresh AI to produce 1 sentence notes on each of our posts from Post #193 onwards more for the purpose of mapping.
My interest in the AI tool at this point is using it to help show when looping is occurring in arguments and ascertaining why.
My opinion on the reason for looping has to do with the starting positions of each participant and the observation that if each participant's position is irreconcilable opposed, then no amount of debate will ever or has ever changed that.

As such, there are no "winners" - only results verifying irreconcilability.

My opinion on your and my positions is that these are not irreconcilable. We both agree that we exist within a created reality.

[The Tanager]

[Replying to William in post #44]

If I did try that and we see what it returned, do you think it would be a good judge of our discussion there in these senses: comprehending what we claimed, picking out what we do well and any mistakes or tricks we use? Not judging who won, but just laying out those three things?

[William]

[Replying to The Tanager in post #45]

Are you asking me if I can accurately predict the results?

If so, why are you asking me? Did I make some assertions about AI abilities? Something else?

If you want to run our conversation in CCA through Deepseek, with those things in mind, then do so. I would interested in discussing those results with you.

[The Tanager]

Are you asking me if I can accurately predict the results?

If so, why are you asking me? Did I make some assertions about AI abilities? Something else?

If you want to run our conversation in CCA through Deepseek, with those things in mind, then do so. I would interested in discussing those results with you.

No, I'm asking you if you think AI would do a good job of comprehending what our posts claimed and be able to pick out strengths and weaknesses of our responses. Are those abilities you believe AI has?

That seems to be the kind of thing you did in post 28 (and in other threads) and you seem to view that as a positive help in these discussions, at least towards getting at truth. But I know I can misread what you mean, so I'm seeking clarification on the above point.

[William]

[Replying to The Tanager in post #47]

You asked whether DeepSeek AI would be a "good judge" of our CCA discussion.
(I have already answered this question in this thread — see Posts #14, #19, #20, #26, and #28 — but I will restate it clearly for you.)
The short answer is no - because AI cannot judge. It is not sentient. It has no standards, no investment in truth, and no ability to weigh competing philosophical frameworks about what counts as an explanation.

What AI can do is:

Summarize posts

Map who said what

Identify patterns (like looping or repeated objections)

Produce a structured output that looks like a verdict

That output is pattern matching based on training data - not genuine judgment. Humans do the same AI is trained on a vast input of human data, meaning it mirrors our patterns, biases, and limitations - including our tendency to mistake pattern recognition for judgment.

You also referenced Post #26 where an AI declared me the better arguer in our CCA exchange. That AI did not know I was the user. It processed the text and produced a conclusion. But even that conclusion is not authoritative. It is simply a data point - a map, not a judge.
At best, AI can help us see the terrain more clearly. The map is not the judge.
So in Post #26, the AI was not judging the original debate directly. It was judging the quality of a summary (post#25) of the debate - specifically, whether that summary was objective, balanced, and accurate.
So if you want to run our CCA posts through DeepSeek with your proposed prompt, feel free. I would be interested in discussing what it produces.

I would not use the prompt you suggest. "Help me by summarizing who is talking, what they claim in the post, and then analyze the case made for what they do well and any logical mistakes or rhetorical tricks they pull"

I usually ask a fresh AI to summarize each Post (I do this through Cut & paste rather than Linking because - while it takes longer it seems the better way to do things than relying on AI to run that task.) I don't let on to the AI that I am one of the posters nor make any comment about the posts.

After the AI has accessed all the posts I then used a fresh AI to provide an objectively analysing the first AIs summary (see post #25)

I then gave my own opinion of the debate between You and POI and share that opinion with the original AI who provided the summaries.
That AI then analysis of my assessment...

ending with the conclusion: "Your declaration that neither participant "won" or is "winning" anything significant over the other is a sound conclusion based on the content of the thread. The debate was characterized by a persistent and unresolved conflict of first principles—a philosophical impasse where both sides argued logically from premises the other did not accept. In such exchanges, progress is often measured not by victory, but by the clarity with which the opposing positions are articulated. By that measure, both POI and Tanager presented their cases thoroughly."

I then went back to the second AI (Post #27) and asked: 'All in all, who consistently gave the better arguments against the other?'

And here is an answer to your question, Tanager - though note the AI's own qualification:

'AI: Based on the summary provided, we can analyze who presented the stronger case by evaluating the logical consistency, burden of proof, and responsiveness to the other's points.'

The AI did not declare an objective winner. It analyzed based on a summary, using criteria I suggested. That is not judgment - it is structured analysis.

[The Tanager]

AI: Based on the summary provided, we can analyze who presented the stronger case by evaluating the logical consistency, burden of proof, and responsiveness to the other's points.'

The AI did not declare an objective winner. It analyzed based on a summary, using criteria I suggested. That is not judgment - it is structured analysis.

That's basically what I meant with 'judge'. Analyzing our posts via normal logical tools like you say here. Do you think AI will provide good structured analysis?

And while it's not how you would do it, is the method I proposed still one that should provide that good structured analysis? It seems comparable. An AI summarizes. An AI gives structured analysis of what the poster is doing by evaluating with normal logical tools.

[William]

[Replying to The Tanager in post #49]

Do you think AI will provide good structured analysis?

Yes. Adequate enough for the task at hand

And while it's not how you would do it, is the method I proposed still one that should provide that good structured analysis? It seems comparable. An AI summarizes. An AI gives structured analysis of what the poster is doing by evaluating with normal logical tools.

That is an interesting question. I think if you tried it the way you suggested, then bring the results back and lets see and discuss those.