The central lie behind these programs is that they are meant for artists. They’re not. We don’t need them and using them only hurts us. What our clients really need from us is what the A.I. button cannot and never will be able to give: a human expression in all its flawed, beautiful glory.
Evaluating Large Language Models (LLMs) as agents in interactive environments, highlighting the performance gap between API-based and open-source models, and introducing the AgentBench benchmark.