Advanced artificial intelligence systems such as OpenAI’s GPT could destabilize democracies unless governments take swift action to “protect their citizens,” AI pioneers warned.
Yoshua Bengio, who won the Turing Award in 2018 alongside Jeffrey Hinton and Yann LeCun, said the recent rush to launch AI products by big tech companies has become “unhealthy”, adding that the “political system , democracy, and society itself,” he added. nature of truth.”
Bengio is the latest in a growing faction of AI experts that are sounding the alarm about the rapid deployment of powerful large-scale language models. His colleague and friend Hinton resigned from Google earlier this month to speak more freely about the risks AI poses to humanity.
In an interview with the Financial Times, Bengio pointed to society’s increasingly indiscriminate access to large-scale language models as a serious concern, and the lack of scrutiny currently being applied to this technology. pointed out.
“Companies can rent access to ChatGPT . . . for example,” he said. “It is important to closely monitor this access, know who is using these systems, and be able to track potentially illegal or dangerous uses.”
The warning comes as technology groups have spent the past six months racing to launch groundbreaking generative AI products — computer software that can write text or create images. Google last week unveiled an improved search engine powered by its new AI system PaLM 2.
OpenAI’s state-of-the-art software GPT-4, backed by Microsoft, went public in March and is rapidly being adopted by banks, consulting firms, startups, educational institutions and governments to improve productivity.
“The dynamic of the race .
Regulators and governments around the world are stepping up scrutiny of the technology amid concerns about its potential abuse. OpenAI CEO Sam Altman on Tuesday expressed concern over OpenAI’s potential to provide “two-way disinformation” ahead of next year’s U.S. election, prompting U.S. lawmakers to regulate the area. called to do so.
Bengio is best known for his work on deep learning, a field of AI that has driven some of the most recent breakthroughs in the field. Bengio is one of the few senior AI researchers left in academia after a massive brain drain to tech companies in 2018. last ten years.
Academic colleague LeCun joined Facebook owner Meta in 2013, and Bengio’s older brother Sammy spent almost 15 years as a senior researcher in Google’s AI division before leaving to join Apple in 2021. .
Bengio was one of more than 1,000 AI experts and tech executives last month who signed an open letter calling for a six-month moratorium on developing advanced AI systems.
He said he didn’t believe the letter would work, but supported the idea of a moratorium. “From what I’ve heard, there seems to be a lot of mixed opinions within OpenAI, Microsoft, and even Google on this,” Bengio said. “The whole community is divided on this issue.”
He pointed out that the disagreement among AI experts is an important signal to the general public that science does not yet have the answers. “If we disagree, it means we don’t know . . . if it can be dangerous. It means we have to act to protect,” Bengio said.
“If humanity and society want to overcome these challenges, there should be no competition and very weak international cooperation between people, companies and nations,” he added.
Short-term dangers of this technology include misinformation, insidious prejudice, and discrimination against marginalized groups and minorities. These societal harms are focused on non-profit research organizations such as the Distributed AI Lab, founded by former Google AI researcher Timnit Gebul.
But the long-term threat Bengio, Hinton and others are concerned about isn’t about today’s chatbots providing human-like answers to questions. Rather, he fears that research into making AI systems more autonomous and correcting their weaknesses, such as their tendency to fabricate facts, could accelerate within the next decade.
Bengio said the most urgent need for regulators to do is to increase the transparency of AI systems, including auditing the data used to train them and their output.
He has also proposed, among other peers, a United Nations to fund AI research in areas of public interest such as climate and medicine.
“Like European investments in CERN and the space program, today’s AI public investment should be at this scale, not just to make big bucks, but to truly bring the benefits of AI to everyone.” said he.
Bengio added that he is currently discussing ideas with Hinton for future collaborations to address these challenges.
“There are a lot of emotions and cries being raised within the broader AI community right now. rice field. “That’s the scientific way.”
