Emerging Applications of Generative Artificial Intelligence Tools in Parascience

Abstract

Following a rapid account of the history of Artificial Intelligence (AI), we introduce the current generation of AI tools based on ‘generative technology’, most notably ChatGPT, and outline why these tools are currently taking the world by storm. We then explore emerging uses of these tools in the parapsychology/parascience literature. We uncover two main themes: (i) analysis of narrative descriptions (e.g. for purposes of categorization), and (ii) elimination of bias in the comparison of ‘free-response’ data against target data (e.g. in remote viewing studies). Both themes explore the use of AI tools as a substitute for human researchers’ efforts in certain tasks. The potential benefits include reduced subjective bias, and the enabling and acceleration of larger-scale analyses. Outcomes so far show promising signs, with AI-tool judgements tending to match those of human experts, with similar levels of variance between judges (i.e. between different tools). However, different tools nevertheless exhibit specific nuances based on their training environment. This suggests further work is needed to understand these ‘tool-biases’ before unleashing AI-tool analyses at scale. A third theme, less explored but clearly signposted, is the development of specialized ChatGPT-style systems, enhanced with libraries of parapsychology material, to support literature study in the field (e.g. summaries, reviews).  This approach has clear merits in counteracting the biases in AI tools that emerge from typical LLM training material such as Wikipedia. Finally, after considering some intrigueing related topics, we end with a note of caution and some brief speculation regarding future developments.