Earlier in the week, Elon Musk openly posted about himself to X. I'm having trouble setting up a new laptop running Windows. He ended the issue by suing OpenAI for recklessly developing human-level AI and handing it over to Microsoft.
Musk's lawsuit is being filed against OpenAI and two of its executives, CEO Sam Altman and president Greg Brockman. The pair founded the company in 2015 in collaboration with rocket and automotive entrepreneurs. The two claim that the original “establishment agreement” was violated. The company said it was committed to working with Musk to develop AGI openly and “for the benefit of humanity.”
Musk's lawsuit alleges that the company's for-profit arm, which he founded in 2019 after he parted ways with OpenAI, created AGI without proper transparency and invested billions of dollars in the company. It claims to have licensed it to Microsoft. He is calling for OpenAI to be forced to publicly release its technology and to prohibit Microsoft, Altman, and Brockman from using it for financial gain.
A large part of the lawsuit revolves around the bold and dubious technical claim that OpenAI has developed so-called artificial general intelligence (AGI). AGI is a term commonly used to refer to machines that can match or surpass humans in overall performance.
“On information and belief, GPT-4 is an AGI algorithm,” the complaint states, referring to the extensive language model behind OpenAI's ChatGPT. The report cites research that found the system can help people achieve passing scores on the Uniform Bar Examination and other standardized tests as evidence that the system exceeds some basic human abilities. ing. “GPT-4 is not only able to reason; it is better at reasoning than the average human,” the complaint alleges.
Although GPT-4 was hailed as a major advance when it was announced in March 2023, most AI experts do not believe it is proof that AGI has been achieved. “GPT-4 is common, but it's clearly not his AGI as people usually use the term,” said Oren Etzioni, a professor emeritus at the University of Washington and an expert on AI. says.
Christopher Manning, a professor at Stanford University who specializes in AI and language, said AGI's claims in Musk's lawsuit “would be considered wild claims.” Manning says that within the AI community he has different opinions on what constitutes AGI. While some experts may set the bar low, arguing that GPT-4 can perform a wide range of functions and thus warrants being called AGI, others argue that it can perform most or all of the functions. We would like to reserve this term for algorithms that can outperform humans at anything. “Based on this definition, we clearly don't have AGI, and in fact I think we're still a long way from AGI,” he says.
Breaking the limit
GPT-4 gained traction and new customers for OpenAI because older AI programs were typically dedicated to specific tasks, such as playing chess or tagging images, whereas GPT-4 was designed to answer a wide range of questions. This is because you can answer. Musk's lawsuit states in a March 2023 paper, “Given the breadth and depth of GPT-4's capabilities, it is difficult to view GPT-4 as an early (and still incomplete) version of the Artificial General.'' “We believe that this is reasonable,” Microsoft researchers said. Intelligence (AGI) system. ” Despite its great capabilities, GPT-4 still makes mistakes and has significant limitations in its ability to correctly parse complex questions.
“I feel that most of us researchers in the field think that large-scale language models are important. [like GPT-4] “It's a very important tool that allows humans to do more, but it's limited in that it's far from standalone intelligence,” said John C., a professor at the University of California, Berkeley, who specializes in machine learning. Michael Jordan, an influential figure in the field, adds: