Just read through a paper from Arxiv. This topic is very interesting. “High order theory of mind”: the human ability to reason about multiple mental and emotional states in a recursive manner (e.g. I think that you believe that she knows). This step by step recursive reasoning should be what AI are good at since they should be good at recursively applying the same logic. The issue is to save within the AI framework, any mid-results that is found.

In addition to the concept that is being studied, the experiment is interesting since it propose a new set of questions for AI to answer. This can be an interesting new benchmark for AI to study in the future.

As for the outcome, we observe that GPT4 seem to be performing well for level 5 and 6 on the benchmark dataset. What is interesting is that humans seems to be performing well for odd levels and not so well for even level. GPT4 performs the best from the first impression but Flan-PaLM performs on par with GPT 4. For level 6, GPT 4 outperforms humans, leading to this interesting paragraph at the end: “However, LLMs possessing higher-order ToM at human levels, or potentially higher, also incurs risks including the potential for advanced persuasion, manipulation, and exploitation behaviours [El-Sayed et al., 2024]. Indeed,‘ringleader’ bullies have been shown to have higher-orders of ToM in comparison to their victims [Sutton et al., 1999a,b] and reinforcement learning agents with higher-orders outcompete their opponents or have a competitive advantage in negotiations [De Weerd et al., 2022, 2017]. LLM-based agents with ToM capacities that exceed those of the average human (as GPT-4 has in our study) could provide a powerful advantage to their users, and a disadvantage to other humans or AI agents with lesser ToM capacities [Street, 2024, Gabriel et al., 2024]. Further research is required to understand how LLM higher-order ToM manifests in real-world interactions between LLMs and users, and to devise technical guardrails and design principles that mitigate the potential risks of LLM ToM without quashing its potential benefits” This leads to uses of LLM for counter bullying.

Reference

Street, Winnie, John Oliver Siy, Geoff Keeling, Adrien Baranes, Benjamin Barnett, Michael McKibben, Tatenda Kanyere, Alison Lentz, Blaise Agüera y Arcas and Robin I. M. Dunbar. “LLMs achieve adult human performance on higher-order theory of mind tasks.” (2024).