Large Language Models (LMMs)

2 Posts

Graphs compare human and LLM performance strategies in rock-paper-scissors, highlighted by stars.

Strategic Thinking in LLMs vs. Humans: Researchers at UT-Austin and Google model human decision-making in Rock-Paper-Scissors

While large language models can behave in human-like ways, the similarities are superficial. A simple strategy game revealed clear differences in their strategic approaches.

Diagram comparing original transformer model with a replacement model using token-level attention and neuron-level outputs.

Large Language Models (LMMs)

Ordinary LLMs Implicitly Take Reasoning Steps: Anthropic experiment finds Claude shows signs of unprompted reasoning

Even without explicit training in reasoning, large language models “think” in ways that may be more deliberate than previously understood.

Large Language Models (LMMs)

Strategic Thinking in LLMs vs. Humans: Researchers at UT-Austin and Google model human decision-making in Rock-Paper-Scissors

Ordinary LLMs Implicitly Take Reasoning Steps: Anthropic experiment finds Claude shows signs of unprompted reasoning

Subscribe to The Batch