More than a thousand AI experts, researchers and funders have joined the call for an immediate delay in the creation of “giant” AIs for at least six months, so that capabilities and risks can be properly investigated and mitigated, such as GPT-4 systems.
The petition contains an open letter signed by major AI players, including Elon Musk, co-founder of OpenAI, the research lab behind ChatGPT and GPT-4; Emad Mostaque, founder of the London company Stability AI; and Steve Wozniak, Apple co-founder.
Its signatories also include engineers from the companies DeepMind, Microsoft, Meta, Google and Amazon, as well as academics, including the notifying scientist M. Gary.
“Recent months have seen AI labs practice in an unbridled fashion so that digital minds can be increasingly developed and developed in ways that no one—not even their creators—can truly understand, predict, or control.” The letter indicates. “Powerful artificial intelligence systems should only be developed when we know for sure that their positive effects will be positive and that their risks can be managed.”
The authors, coordinated by the Future Life Institute’s “long-term” study center, cite OpenAI’s co-inventor, Sam Altman, to justify their claims. Later in February, Altman wrote: “At some point it may be important to review independently before planning future systems and developing projects to agree more on creating new models to limit the use of computing.” .
The letter continued: “I agree. That’s the moment now.”
If researchers do not stop their work on AI models superior to GPT-4, the literature refers to the “giant” models, then “the government will step in,” the authors say.
“This does not mean a pause in the development of artificial intelligence in general, but simply a step back in the dangerous category of ever-larger, vague black-box models with emerging capabilities,” they add.
Since the release of GPT-4, OpenAI has added capabilities to the AI system through plugins, giving it the ability to search the web for data, plan vacations, and even make purchases. However, the company must deal with “overcapacity” – it realizes that the problem of its own system is more important than that of the customer.
As researchers test the GPT-4 over the next few weeks and months, they are likely to find new ways to “boost” the system by improving its ability to solve difficult problems. A recent finding was that artificial intelligence is significantly more accurate in solving problems when it first requires “expert style”.
The call for strict rules is in stark contrast to the British government’s official document on regulating artificial intelligence issued on Wednesday, which includes no new powers. But the government says, the focus is to organize the existing regulators, such as the authority and the authority and the Health and Safety Executive Competition and the market, five “principles” by which they should think about artificial intelligence.
“Our new approach is based on sound principles so that companies can trust that tomorrow’s technology will work,” said Secretary for Science, Innovation and Technology Michelle Donelan.
The Ada Lovelace Institute was one of the ones to criticize. “The UK has significant access gaps, which have been released without any harm, and is hampered by the stimulus and scale of the challenge,” said Michael Birtwistle, who gives policy and law, and artificial intelligence at the study center.
“The government’s implementation deadline of one year or more will leave some risks unaddressed as artificial intelligence systems become increasingly integrated into our daily lives, from search engines to search software for office files.
The Labor Party joined the criticism, with Shadow Culture Secretary Lucy Powell accusing the government of having “broken their side of the deal”.
Powell noted: “This regulation will take months, if not years, to take effect.” Meanwhile, ChatGPT, Google Bard and many others are making AI a regular part of our daily lives.”
“Government confirms the risk of being stuck in the current regulatory system, and the system makes it overwhelming for businesses and citizens to manage, while also undermining those fundamentals coming through the law.” .