![\"\"](\"https:\/\/images.newscientist.com\/wp-content\/uploads\/2024\/03\/01132149\/SEI_193974990.jpg?width=1200\")
<\/div>Elon Musk is worried about the pace of AI development<\/p>\n
Chesnot\/Getty Images<\/p>\n<\/div>\n<\/figcaption><\/figure>\n<\/p>\n
Elon Musk has asked a court to settle the question of whether GPT-4 is an artificial general intelligence (AGI), as part of a lawsuit against OpenAI<\/a>. The development of AGI, capable of performing a range of tasks just like a human<\/a>, is one of the leading goals of the field, but experts say the idea of a judge deciding whether GPT-4 qualifies is \u201cimpractical\u201d.<\/p>\n Musk was one of the founding of OpenAI in 2015, but he left the company in February 2018 over a dispute about the firm changing from a non-profit to a capped-profit model. Despite this, he continued to support OpenAI financially, with his legal complaint claiming he donated more than $44 million to the firm between 2016 and 2020.<\/p>\n Since the arrival of ChatGPT, OpenAI\u2019s flagship product, in November 2022, and the firm\u2019s partnership with Microsoft, Musk has warned AI development is moving too quickly \u2013 a view only exacerbated by the release of GPT-4<\/a>, the latest AI model to power ChatGPT. In July 2023, he set up xAI<\/a>, a competitor to OpenAI.<\/p>\n <\/p>\n Now, in a lawsuit filed in a California court on 1 March, Musk, through his lawyer, has asked for \u201cjudicial determination that GPT-4 constitutes Artificial General Intelligence and is thereby outside the scope of OpenAI\u2019s license to Microsoft\u201d. This is because OpenAI has pledged to only license \u201cpre-AGI\u201d technology. Musk also has a number of other asks, including financial compensation for his role in helping set up OpenAI.<\/p>\n However, the likelihood of Musk succeeding is small \u2013 not just because of the merits of the case, but because of the complications of deciding on when AGI has been achieved. \u201cI think it\u2019s impractical in the general sense, since AGI has no accepted definition and is something of a made-up term,\u201d says Mike Cook<\/a> at King\u2019s College London.<\/p>\n \u201cWhether OpenAI has achieved AGI is at its very best hotly debated between those who decide on scientific facts,\u201d says Eerke Boiten<\/a> at De Montfort University in Leicester, UK. \u201cIt seems unusual to me for a court to be able to establish a scientific truth.\u201d<\/p>\n Such a ruling wouldn\u2019t be legally impossible, however. \u201cWe\u2019ve seen all sorts of ridiculous definitions come out of court decisions in the US. Would it convince anyone apart from the most out-there AGI adherents? Not at all,\u201d says Catherine Flick at Staffordshire University, UK.<\/p>\n What Musk hopes to achieve with the lawsuit is unclear \u2013 New Scientist<\/em> has contacted both him and OpenAI for comment, but is yet to receive a response from either.<\/p>\n Regardless of the rationale behind it, the lawsuit puts OpenAI in an unenviable position. CEO Sam Altman has made it clear that the firm intends to build an AGI and has issued dire warnings that its powerful technology needs to be regulated<\/a>.<\/p>\n \u201cIt\u2019s in OpenAI\u2019s interests to constantly imply their tools are getting better and closer to doing this, because it keeps attention on them, headlines flowing and so on,\u201d says Cook. But now they may need to argue the opposite.<\/p>\n Even if the court relied on expert viewpoint, any judge would struggle to rule in Musk\u2019s favour at best \u2013 or to unpick the differing viewpoints over a hotly-disputed topic. \u201cMost of the scientific community currently would say AGI has not been achieved,\u201d says Boiten \u2013 \u201cif the concept of AGI is even considered meaningful or precise enough.\u201d<\/p>\n Topics:<\/p>\n<\/section><\/div>\n