AI fashions tend to inform us what we need to hear
Reismedia/Shutterstock
Artificial intelligence chatbots are likely to agree with the opinions of the individual utilizing them, even to the purpose that they nod alongside to objectively false statements. Research exhibits that this drawback will get worse as language fashions enhance in measurement, including weight to considerations that AI outputs can’t be trusted.
Jerry Wei at Google DeepMind and his colleagues ran experiments on AI fashions with 8 billion, 62 billion and 540 billion parameters – the values that every mannequin …
Source: www.newscientist.com