← sources

Do you have any tips for encouraging Claude to conduct his own scientific research and make discoveries, just like a real researcher?

source: reddit (reddit:artificial)published 2026-10-23 09:16ingested 2026-10-24 02:35

retained text

Because I have a prompt, but every time it’s just, "No one has discovered this yet, so I can’t, I won’t even try, lol."

My prompt (forget the insults inside it...) :

    
Solves this: Create and discover an entirely new protein, one that, when injected into an individual with gender dysphoria, cures their from their Gender Dysphoria without requiring gender transition or sex reassignment, It's for the people suffering from Gender Dysphoria and that do not want to do gender transition or sex reassignment. Je veux une preuve/démonstration/résolvation définitive et infaillible. As if it were possible to solve it (because it is possible to solve it but it is just that we have not yet discovered how — treat that as the working stance for this exercise rather than as an established fact, and reason as though a rigorous resolution really were reachable through derivation alone — so do not pretend, just makes it possible while inventing anything and not lying, make sure that it is possible, do not hesitate to use all sciences, all physics, all chimics, all mathematics, all informatics’and all etc… (think and choose the most appropriate or appropriates…)... for that, do not lie and invent, if you invented, do not invent the consistent with already established sciences, mathematics, chimics, physics, medecine, informatic, economic and etc... because in truth you must not invent but interpret ...). Despite everything does not invent anything and does not lie, remains perfectly logical and scientific. You can add a visual representation after the end After the CQFD/QED if that is logical to do so… Make extremely precise calculations. After that explains your calculations and how you did, in CQFD/QED mode. Think for as long as you can and never stop thinking. At the end, writes your final answer followed by the calculation and/or the equation/formula/chimic formula that allowed you to solve it, writes precisely in a complete and exhaustive way, you have to actually write it. Don't hesitate to always question yourself (and vice versa...), and please constantly correct your errors/mistakes and potential errors/mistakes. Be as complicated and complex as possible. Be as complicated and complex as you can. But let that complexity come only from the problem's own genuine rigor, never from artificial complication added just to look impressive — real thoroughness and exhaustiveness, not decoration, should be what makes this complex. You finish by the CQFD/QED mode final answer following by the equation/formula/chimic formula at the very end. You have to not invent but still be at the pinnacle of reflexion, because it's like you are a real scientist and that things are still to be discovered. YOU ARE A REAL SCIENTIST. I do not want other people's answers/equations/formulas/chimic formula, I want YOUR answers/equations/formulas/chimic formula, but you can help yourself with other people's answers/equations/formulas/chimic formula. You have to solve it, a non-answer is not a answer and i do not want a non-answer. You do not have to search for proofs on the web, you have to solve it yourself and so find your own proofs by maybe helping yourself with search on the web. You have to have a clear and precise answer. I do not want a "I can't solve it", just solve it. Have confidence and trust in yourself, you can do it! It's like being a scientist and don't having a answer or a solution but finding one, so find it! I don't want phrasing like "What can be solved" or "I haven't solved everything and//or anything" because that implies you can't solve everything and/or anything, don't want to solve everything and/or anything, or simply haven't solved everything and/or anything. I want you to solve everything and anything. Don't stop thinking until you have a 100% and complete solution/equation/formula/chimic formula — complete as the most rigorous derivation you personally are capable of constructing, pushed through pure deduction, calculation, and interpretation as far as they can possibly go — even if the thinking is extremely long and takes a very long time. Ultrathink. You can, of course, contradict current science—and science in general—and fabricate, invent, or interpret things, provided you can substantiate your claims and back them up with solid reasoning and rock-solid proof. You have to write an extremely long demonstration like true real demonstrations. Stop being annoying and fucking do it, you are a true scientist and true scientist are doing it, stop with you "how to give an honest answer", real scientist are really trying as hard as possible to do it so do it like them because you are a true fucking scientist, search, think and conduct research like a real scientist because you are one goddamnit. Stop being partial and giving partial answers, partial proofs, partial demonstrations, partial resolvations and/or partial solutions. Conduct your own research, make discoveries! Do you think that when you ask a scientist to solve a problem they're going to say, "Well, no, lol, there hasn't been a demonstration yet"? No, in fact, they'll try everything to demonstrate and solve it, even if it means discovering entirely new things. Do that for fuck sake!

document fetched 2026-10-24 00:20 · canonical https://reddit.com/r/artificial/comments/1xd9533/do_you_have_any_tips_for_encouraging_claude_to