AI đđ¤ LLM Deception Universal motif when LLMs 'knowingly lie' đ¤đ¨ LLM Jailbreaks Seeking commonalities between successful jailbreak attacks đ§đšī¸ SAE steering Steering LLMs with SAE features đ¤đ¨âđŠâđĻâđĻī¸ LLM Agent Multi-agent interaction and tool-use Neuroscience đ§ đˇī¸ Memory Consolidation Decoding the 'dream' (replay) of rodents while they sleep. đ§ đˇī¸ Event Representation How discrete episodic events are represented in the brain
