Which AI Models Do Experts Use and Trust?
AI assistants have moved from novelty to everyday infrastructure. Which models does the expert community actually use, which do they find reliable in their own fields, and which companies do they trust to build them?
In the space of three years, AI language models have become embedded in scientific and professional work. A large international survey of medical researchers found that roughly 40% report using AI chatbots in the scientific process (international cross-sectional survey), and Nature's reporting indicates broad adoption across fields despite well-documented limitations (Nature, 2025). What remains poorly measured is which models experts actually rely on, how reliable they find the output in their own areas of expertise, and how much faith they place in the companies building these systems.
This SciPoll asks the SciPinion expert community to report their own usage and assessments directly: the models they use in professional and personal contexts, the reliability of model output in their fields, and their confidence in the companies developing AI. All responses are anonymous. Results may be published in aggregate form.