-
サマリー
activate_samplebutton_t1
あらすじ・解説
Large Language Models (LLMs) like ChatGPT seem to be awash in ethical problems like bias, hidden labour, lack of transparency, energy use…but since when did they become too Woke? When Constance and Darian enlisted ChatGPT to expand their horizons into conservative and far-right ideas they ended up in the same liberal bubble as always, what gives? Thankfully, data scientist Jerry Spanakis was on hand to try to explain what’s going on and how both training data and tweaking by socially liberal engineers can lead to a liberal status quo being reproduced by some models. Is this such a bad thing? Listen and find out.