An analysis of survey responses on the desired and perceived capabilities of LLMs in academic research.
Respondents want AI to handle tedious, time-consuming, and preparatory tasks.
Core intellectual contributions like novel ideas and final judgments are kept for human oversight.
There's strong confidence in AI's ability to assist with structured tasks like coding and writing.
Respondents believe AI currently lacks the ability for true innovation and high-level reasoning.
Researchers see AI as a powerful assistant for support tasks (literature review, coding, writing), not a replacement for the core researcher.
There's a clear boundary: AI assists, but humans must drive novel ideas, research design, and final validation. The creative spark remains human.
Confidence is high for AI's role in executing well-defined tasks, but skepticism remains about its ability to handle abstract reasoning or genuine innovation.