Computer SciencePNAS
ChatGPT as Research Scientist: Probing GPT’s capabilities as a Research Librarian, Research Ethicist, Data Generator, and Data Predictor
S. A. Lehr, A. Caliskan, et al.
How good is ChatGPT as a research scientist? This paper probes GPT-3.5 and GPT-4 across librarian, ethicist, data generator, and novel data predictor roles in psychological science—revealing frequent hallucinations (especially in GPT-3.5), improving self-awareness in GPT‑4, strong ethical detection, reliable reproduction of known biases, but poor novel prediction. Research conducted by Steven A. Lehr, Aylin Caliskan, Suneragiri Liyanage, Mahzarin R. Banaji.
Related Publications
Explore these studies to deepen your understanding
Adjacent work that informs or extends this paper's methodology and findings.
Linguistics and Languages
The effectiveness of ChatGPT as a lexical tool for English, compared with a bilingual dictionary and a monolingual learner's dictionary
R. Lew, B. Ptasznik, et al.
Computer Science
Navigating the perils of artificial intelligence: a focused review on ChatGPT and responsible research and innovation
A. Polyportis and N. Pahos
Medicine and Health
Employing a systematic approach to biobanking and analyzing clinical and genetic data for advancing COVID-19 research
S. Daga, C. Fallerini, et al.
Medicine and Health
Psychiatric polygenic risk as a predictor of COVID-19 risk and severity: insight into the genetic overlap between schizophrenia and COVID-19
M. Alemany-navarro, S. D. Almeida, et al.
