Ask HN: Vibe Researching" with AI – Anyone Using It for Real?

Posted by spenceXu 1 day ago

Counter8Comment5OpenOriginal

The concept of "vibe researching" – using AI to rapidly explore, synthesize literature, and generate novel research ideas or frameworks – seems promising. Beyond just literature reviews, it could act as a brainstorming co-pilot.

Has anyone here seriously used AI (e.g., Claude for long-context paper analysis, custom GPTs on arXiv, or specialized agents) to aid in hypothesis generation, research gap identification, or drafting substantive parts of a paper?

What are the biggest pitfalls regarding accuracy, hallucination of citations, or superficial understanding of complex theory? How do you validate the AI's output?

Do you see it as a legitimate accelerator for early-stage research, or more of a productivity tool for mundane tasks? Any success stories linking it to a tangible research outcome?

Looking for honest experiences from academics, industry researchers, or solo discoverers.

Comments

Comment by keiferski 1 day ago

Not for anything seriously published, but - yeah I use ChatGPT constantly to evaluate essay ideas, critique arguments, and give me more reading material ideas for whatever topic I'm thinking about.

I think this is truly the best use-case of LLMs, actually. It functions as a kind of hyper-informed assistant.

Comment by sangkwun 1 day ago

For "vibe research" across multiple sources, I've been using an AI setup that monitors topics I care about and summarizes only what's relevant.

Helps reduce the information overload while still catching context quickly. Instead of browsing 10 newsletters and feeds manually, I get a digest of what actually matters to my current interests.

Not quite the same as deep literature review, but effective for staying on top of a field without drowning in it.

Comment by rektlessness 1 day ago

I'm not a researcher but i did 'write' a 'paper' for personal consumption to dissect a thought experiment I had on thriving in the Age of AI, Neural Networks, and Quantum Logic.

It examined the psychological and strategic archetypes that determine success or failure during periods of radical technological disruption, using the internet revolution (1995-2015) as a historical baseline. I don't know if it was any good, but it was a fun few hours of exploration.

Comment by eamag 1 day ago

I tried it last year, and compiled a list of projects in one category herehttps://openreview-copilot.eamag.me/projects and 2025 retro here https://eamag.me/2025/links/Links-And-A-Retrospective-Of-202...

The biggest issue is to be able to identify good ideas, ones that are useful, novel and doable. "Research taste" is the thing people develop over years!

I don't think we're there yet, and "vibe" research producing slop now, but creating tools for other researches to move faster sounds way more promising

Comment by bjourne 1 day ago

It doesn't work. You can use it for proof-reading, but that's about it. You can use it to stuff your text with scientific sounding BS: "remains challenging", "motivating alternative approaches", "emerged as a prominent", "strategy leverages", and "typically". Typically, it will insert a "typically" in every third sentence. Leveraging AI for high-quality content generation "remains challenging".