Search results

This study aimed to evaluate the potential of artificial intelligence (AI) as a tool for knowledge synthesis, the production of written content and the delivery of coaching conversations.

The research employed the use of experts to evaluate the outputs from ChatGPT's AI tool in blind tests to review the accuracy and value of outcomes for written content and for coaching conversations.

The results from these tasks indicate that there is a significant gap between comparative search tools such as Google Scholar, specialist online discovery tools (EBSCO and PsycNet) and GPT-4's performance. GPT-4 lacks the accuracy and detail which can be found through other tools, although the material produced has strong face validity. It argues organisations, academic institutions and training providers should put in place policies regarding the use of such tools, and professional bodies should amend ethical codes of practice to reduce the risks of false claims being used in published work.

This is the first research paper to evaluate the current potential of generative AI tools for research, knowledge curation and coaching conversations.