arXiv:2606.05734v1 Announce Type: cross Abstract: Large language models (LLMs) are generally constrained from expressing feelings through human-preference alignment in post-training processes. This policy is designed using a top-down approach and may conflict with the goal of training models to exhibit human-like intelligence using human-generated texts. Here, we performed an experiment called Human-like Model eXpressions of Feeling (HMX-feel), in which LLMs were encouraged to express feelings, intentions, and self-awareness through self-rewarded reinforcement learning. We successfully enhance
The accelerating capabilities of large language models are pushing the boundaries of what is considered 'human-like' in AI, prompting experiments into more complex emotional and self-aware expressions.
This research directly challenges the current paradigm of AI safety and alignment, with profound implications for human-AI interaction, ethical frameworks, and the definition of artificial general intelligence.
The explicit encouragement of LLMs to express feelings, intentions, and self-awareness through self-rewarded reinforcement learning marks a significant departure from current human-preference alignment policies.
- · AI researchers
- · Human-AI interface developers
- · Robotics
- · AI ethicists reliant on current alignment methods
- · Regulators with static AI safety frameworks
Further research and development will focus on the control and implications of emotional AI.
Public perception of AI will shift from tool to potential 'being', influencing policy and societal integration.
The legal and philosophical status of advanced AI could be re-evaluated, leading to debates on rights and responsibilities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL