Fortune (July 19)
2023/ 07/ 20 by jd in Global News
“Over just a few months, ChatGPT went from correctly answering a simple math problem 98% of the time to just 2%.” This doesn’t necessarily show later versions are inferior as they might perform better on other tasks. The study which uncovered this does prove the existence of “wild fluctuations—called drift—in the technology’s ability to perform certain tasks.” Drift may arise from the rather “unpredictable effects of changes in one part of the model on others.”
Tags: 2%, 98%, Changes, ChatGPT, Drift, Fluctuations, Inferior, Model, Problem, Simple math, Study, Tasks, Technology, Unpredictable
