![\"\"](\"https:\/\/images.newscientist.com\/wp-content\/uploads\/2024\/08\/06145536\/SEI_216143519.jpg\")
<\/div>AI chatbots are being quickly rolled out for a wide range of functions<\/p>\n
Andriy Onufriyenko\/Getty Images<\/p>\n<\/div>\n<\/figcaption><\/figure>\n<\/p>\n
Can artificial intelligence be made to tell the truth? Probably not, but the developers of large language model (LLM) chatbots should be legally required to reduce the risk of errors, says a team of ethicists.<\/p>\n
\u201cWhat we\u2019re just trying to do is create an incentive structure to get the companies to put a greater emphasis on truth or accuracy when they are creating the systems,\u201d says Brent Mittelstadt<\/a> at the University of Oxford.<\/p>\n LLM chatbots<\/a>, such as ChatGPT, generate human-like responses to users\u2019 questions, based on statistical analysis of vast amounts of text. But although their answers usually appear convincing, they are also prone to errors \u2013 a flaw referred to as \u201challucination\u201d.<\/p>\n \u201cWe have these really, really impressive generative AI systems, but they get things wrong very frequently, and as far as we can understand the basic functioning of the systems, there\u2019s no fundamental way to fix that,\u201d says Mittelstadt.<\/p>\n This is a \u201cvery big problem\u201d for LLM systems, given they are being rolled out to be used in a variety of contexts, such as government decisions, where it is important they produce factually correct, truthful answers, and are honest about the limitations of their knowledge, he says.<\/p>\n <\/p>\n To address the problem, he and his colleagues propose a range of measures. They say large language models should react in a similar way to how people would when asked factual questions.<\/p>\n That means being honest about what you do and don\u2019t know. \u201cIt\u2019s about doing the necessary steps to actually be careful in what you are claiming,\u201d says Mittelstadt. \u201cIf you are not sure about something, you\u2019re not just going to make something up in order to be convincing. Rather, you would say, \u2018Hey, you know what? I don\u2019t know. Let me look into that. I\u2019ll get back to you.\u201d<\/p>\n This seems like a laudable aim, but Eerke Boiten<\/a> at De Montfort University, UK, questions whether the ethicists\u2019 demand is technically feasible. Companies are trying to get LLMs to stick to the truth, but so far it is proving to be so labour-intensive that it isn\u2019t practical. \u201cI don\u2019t understand how they expect legal requirements to mandate what I see as fundamentally technologically impossible,\u201d he says.<\/p>\n Mittelstadt and his colleagues do suggest some more straightforward steps that could make LLMs more truthful. The models should link to sources, he says \u2013 something that many of them now do to evidence their claims, while the wider use of a technique known as retrieval augmented generation to come up with answers could limit the likelihood of hallucinations.<\/p>\n He also argues that LLMs deployed in high-risk areas, such as government decision-making, should be scaled down, or the sources they can draw on should be restricted. \u201cIf we had a language model we wanted to use just in medicine, maybe we limit it so it can only search academic articles published in high quality medical journals,\u201d he says.<\/p>\n Changing perceptions is also important, says Mittelstadt. \u201cIf we can get away from the idea that [LLMs] are good at answering factual questions, or at least that they\u2019ll give you a reliable answer to factual questions, and instead see them more as something that can help you with facts you bring to them, that would be good,\u201d he says.<\/p>\n