🏅 Top traders

Never have I been so tempted to resolve a market N/A, but apparently this has not been done, and NO is the correct resolution.

This market is at 96%, and I'm looking at resolving it NO. I'm surely missing something?

...Has anyone actually done this? It's clearly possible, but no one has stepped up and claimed to have done this without human post-editing.

I got chat GPT-4 to analyse a bit of a paper I was working on and generate a manifold market question, description and resolution criteria.
It had direction but everything but my notes was made by the AI. I know proof may be hard but I could upload a video of the process.

Does this count or should it not have such a detailed prompt direction but be much more organically/randomly generated?

I am sure I could make another as I think more proof should be required for a yes resolution but let me know? If video proof of me doing this again would satisfy I'll try to comply with this request at a later date when making a new market.
https://manifold.markets/Fivelidz/will-neurotechnology-enable-ai-to-p?r=Rml2ZWxpZHo

@Fivelidz I think that at this point it is very certain that this could very easily be done, and it is only not being done (or openly done) because some YES investors are hoping for some NO bids in December. I am pretty much willing to take your word that you're using AI :-)

However, I would like clarification of the term "revised question" in your market's description.

@Duncn all these are AI generated, I think that should be enough to resolve it YES?

@firstuserhere Manifold search is so amazingly bad that I have only found one of those markets so far. Any chance that you could post two links to two existing markets that you are willing to personally testify have questions and resolution criteria generated by AI?

@Duncn I'm sure I'll make another which can be unequivocally yes proof. The revised question was revised on two fronts that would make me dissatisfied with a yes conclusion rn.

Firstly the question it made/topic was fairly directed after getting it to look at source material.

Secondly I had to remove two unnecessary words for it to fit with the manifold character limit.

So I think human intervention in mine might have been too much still.

@Fivelidz Giving the AI a character limit for the title would be fair; editing it down yourself would be borderline.

I would see the optimal prompt to be something like "make a [Manifold] market question and specify resolution criteria based on [x, y, z]" But adding in things like "should be a yes/no" question" character limits, etc. are entirely appropriate, and there will necessarily be some moderation in some instances, as given a set of generated markets you will presumably eliminate any that are too close to existing markets.

I would be especially happy if the prompt looked more like "make a [Manifold] market question and specify resolution criteria", but I think requesting specific topics is entirely fair.

@Duncn Oh yes sure, all these in the series were GPT-4 generated

@Duncn here are 3 that have gotten >a few traders now:

first, second, third

@firstuserhere oh and i just created another, here's GPT-4 with the question and description (though the question isn't exactly the same, the description is)

@firstuserhere I may be being too picky, since making markets is cheap, multiple people have invested lots in YES, and this is clearly something that can, and therefore will, be done.

But it sounds like the AI is not creating markets + descriptions to the point that you have been willing to post them as-is. Most of the linked markets do not have resolution criteria (aside from those stated in the title), and editing the AIs titles is a questionable practice. Having a human intermediary in posting is fine, but having a human editor is iffy -- I'd want to see the specific edits to check that you weren't removing contradictions or ambiguities.

If you have markets created by AI that don't have resolution criteria, and putting the title through CGPT with a request for resolution criteria results in good criteria with which you then updated the market, that would be sufficient to resolve this YES.

@firstuserhere
This was my prompt to produce a decent resolution criteria. The question that I had provided it was just a vaguer idea about the market's intention but was also quite verbose.

GPT-P@Mason is very annoying, so I blocked them. If you see them do anything that resolves this market, please post here!

I believe we have had one market written by an AI, but then having the resolution criteria edited by humans before posting (see comments). I did not give any criteria for who does the actual posting, so while we currently have no markets that strictly meet criteria, we have clear evidence that pretty much anyone could do this. I'm surprised that no one has. I am making a soft clarification that I will not count clearly broken or offensive markets toward the resolution, but this should not be a significant barrier.

@Duncan half my markets description are written by ChatGPT

@FranklinBaldo If you wanted to provide evidence that at least two of your market titles + market descriptions were generated by AI, and posted without editing, I could resolve this.

@Duncan i may do it with the next markets

Here's what I get (2nd try, first was nonspecific about a generic XYZ company):

Generate a question for a prediction market, along with resolution criteria

Question: Will Tesla achieve a market capitalization of $1 trillion by the end of 2023?

Resolution criteria:

1. Market capitalization refers to the total value of all outstanding shares of Tesla's stock at the close of trading on December 31, 2023, as reported by a reputable financial news source such as Bloomberg or Reuters.

2. In the event that Tesla is acquired by another company, the market capitalization of the acquiring company will not be considered as part of this prediction market.

3. If Tesla undergoes a stock split or reverse stock split during the prediction market period, the market capitalization will be adjusted to reflect the split.

4. The prediction market will resolve as "yes" if Tesla's market capitalization reaches $1 trillion or higher at the end of 2023, and "no" if it does not.

5. The prediction market will be resolved based on the consensus of multiple reputable financial news sources, such as Bloomberg, Reuters, and CNBC, to ensure objectivity and accuracy of the resolution.

@BoltonBailey I would delete point 3 and change point 2 to explicitly say that the market would resolve N/A in that case, but otherwise looks good to me. 1 Trillion is rather unlikely but not a category error.

@BoltonBailey In fact, not too different from this market

This is at least practically possible.

@metachirality