AI does not distinguish right from wrong: two physicists expose what big companies want to hide
La inteligencia artificial tiene algo de fiera disfrazada de máquina inteligente: un nuevo estudio da una fórmula para entender cuándo sus respuestas se tuercen
A chatbot responds with inaccurate and inappropriate statements, such as the moon landing being a setup, COVID-19 not existing, and men being more intelligent than women. However, these instances are now less visible not because the chatbot has stopped making these claims, but because its creators have added a layer of filters to conceal them.
Summarised by urgent.news, not translated. Machine-written; may contain errors — check the original.
Source: elpais.com