MIT professor cautions about the downward trajectory of AI tech companies

Physicist Max Tegmark contends that tech executives are unable to impede AI progress amid intense competition

The scientist who spearheaded a notable letter calling for a temporary halt in the development of powerful artificial intelligence systems has mentioned that technology executives did not suspend their efforts due to being deeply entrenched in a “race to the bottom.”

Max Tegmark, one of the co-founders of the Future of Life Institute, orchestrated an open letter in March, advocating for a six-month pause in the advancement of extensive AI systems. Despite receiving support from over 30,000 signatories, including Elon Musk and Apple co-founder Steve Wozniak, the initiative did not succeed in achieving a pause in the development of the most ambitious AI systems.

In a discussion with The Guardian six months later, Tegmark disclosed that he hadn’t expected the letter to dissuade tech companies from pursuing AI models even more powerful than GPT-4, the extensive language model behind ChatGPT, primarily due to the escalating competition.

“In my conversations with corporate leaders, I sensed that many of them privately wished for a pause, but they found themselves entangled in a fierce competition against each other. As a result, no single company could afford to pause independently,” he explained.

The letter raised concerns about an unbridled competition to develop intelligences that would be beyond comprehension, prediction, or effective control. If leading AI firms such as Google, OpenAI (the owner of ChatGPT), and Microsoft couldn’t reach an agreement on a moratorium for systems more advanced than GPT-4, the letter urged governments to intervene.

It posed fundamental questions: “Should we continue advancing non-human intelligences that could eventually surpass, outsmart, make obsolete, and replace us? Are we prepared to take the risk of losing control over our civilization?

Max Tegmark, a physics professor at the Massachusetts Institute of Technology (MIT), considered the letter to be a triumph.

“The influence of the letter has been more substantial than I originally foresaw,” he commented, emphasizing a noteworthy political awareness regarding AI. This awareness has manifested in US Senate hearings involving tech executives and the UK government’s arrangement of a global summit on AI safety in November.

Max Tegmark observed that the expression of concern about AI had shifted from a once-taboo subject to a mainstream perspective since the release of the letter. He highlighted that his thinktank’s letter was followed in May by a statement from the Center for AI Safety, endorsed by numerous tech executives and academics. This statement asserted that AI should be considered a societal risk on par with pandemics and nuclear warfare.

“I felt there was a substantial amount of suppressed concern about the rapid advancement of AI—concerns that people worldwide hesitated to voice for fear of being perceived as alarmist critics. The letter gave legitimacy to these discussions, making them socially acceptable,” explained Tegmark.

Tegmark warned against framing the advent of digital “god-like general intelligence” as a distant future threat, emphasizing that some AI experts believe it could materialize in just a few years.

The Swedish-American scientist expressed excitement about the forthcoming UK AI safety summit in November, set to occur at Bletchley Park, describing it as a “remarkable initiative.” His thinktank has outlined three key objectives for the summit: promoting a collective understanding of the seriousness of AI-related risks, recognizing the need for a coordinated global response, and advocating for the urgency of government intervention.

Moreover, he underscored the ongoing requirement for a temporary halt in AI development until universally accepted safety standards are defined. He asserted, “Progressing models beyond our current capabilities must be temporarily suspended until they can align with universally agreed-upon safety criteria.” He also pointed out, “Achieving consensus on these safety standards will inherently result in the pause.

Tegmark also urged governments to tackle the issue of open-source AI models accessible for public use and modification. Notably, Meta, led by Mark Zuckerberg, recently introduced an open-source large language model named Llama 2. A UK expert cautioned that such a decision was similar to “providing individuals with a blueprint for constructing a nuclear bomb.