ChatGPT, Galactica, and the Progress Trap
For WIRED, the authors hone in on the need for accountability in LLM deployment.
Overview
"The release of LLM's such as ChatGPT and Galactica has reignited conversations about what these models can do. Their capabilities have been presented as extraordinary with evangelists claiming that these models even resemble consciousness. However, such hype is not much more than a distraction from the actual harm perpetuated by these systems. Every day, individuals are harmed by the practical shortfallings of model deployment, requiring that failures of those who built the tech, are held accountable.
People get hurt from the very practical ways such models fall short in deployment, and these failures are the result of their builders’ choices—decisions that they must be held accountable for."
