I will be starting a CS PhD in formal methods (mathematucal formalizations and proofs for programs and languages) in October. By when will AI be executing most of my research, with me reviewing and setting goals?
Writing code and formal proofs (e.g. lean, which I'm probably going to be doing the formalization in) is not sufficient to resolve YES. A YES resoution requires e.g. AI taking over most low and medium level design decisions, autonomously proposing, reseaching, comparing, and evaluating on different design approaches competently enough that I am satisfied with the reasoning presented to me and do not feel like I have to step in to fix bad decisions and missed opportunities.
In my current reeearch so far, I have not really seen this with claude at all. I think it's mostly because good high level decisions require keeping a wide context of self deveolped solid concepts in view, and somehow LLMs seem to tend to keep everything they come up with very liquid and subject to reinterpretation.
Resolves when I look back on the last month, and decide that I think AI has been executing most of my research in the last month.
I'll update this market with how things ate going if I feel like there has been notable progress in this direction.
I will not bet in this market.
Update 2026-06-09 (PST) (AI summary of creator comment): The creator clarified that if they are directing AI agents in a way analogous to how a professor oversees grad students (with AI doing most low and medium level work while the creator sets direction and reviews output), that would be sufficient to resolve YES.
Update 2026-06-11 (PST) (AI summary of creator comment): The market will not resolve YES if AI becomes capable of executing most research but the creator cannot afford it or refuses to use it. Reasonable pricing and accessibility are part of what the market is predicting. If the creator refuses to use AI due to wanting to preserve the research experience, this would prevent YES resolution.
1,000People are also trading
@vluzko Would not resopve YES. I'd generally have no way to tell when this happened, and somewhat reasonable pricing/wife accessibility is part of what I want the market to predict. Notably when I start my position, I'll gat a big increase in disposable income, so I imagine something like the current claude max price tag wouldn't stop me.
Refusing is more complicated. If the whole field starts getting steamrolled by AI assisted people, than I don't see a world where I (could) refuse. E.g. I think this will probably be the case for regular programming by the end of the year.
In a scenario closer to the margin, I'll have to consider the tradeoff between productivity/more better publications, and losing out on the research experience.
@ChurlishGambit I have no idea, at a gut level I generally do not like researchers being replaced with AI, but AI solving cancer and all that sounds nice I guess.
Maybe in the future all researchers are just AI wranglers, so learning to make AI do stuff will actually be learning to be a researcher. Maybe a year from now it will be impossible to get your human produced paper published, or to defend your thesis, because all the AI using people will be putting out way more impressive work. The intention of the market is to be a proxy for all of this.
@ChurlishGambit also I suppose professors take a large amount of credit for the output of their research groups all the time, even though most low and medium level work is done by grad students, and that's considered legitemate. I think if I was doing the equivalent of that over AI agents, that would be enough to resolve.
@consnop Professors already HAVE their doctorates, though—earned by producing original research. & they credit their assistants & research groups.
>Maybe in the future all researchers are just AI wranglers,
Not a future anyone except the most ardent slop addicts wants.
>but AI solving cancer and all that sounds nice I guess.
It can't & won't.
>all the AI using people will be putting out way more impressive work
They really won't.
@ChurlishGambit if you're confident in those beliefs, you have profit to be made on this market. I am certainly not going to make myself use it if it will have a negative or neutral effect, including the lost value of myself losing research experience.