Elon Musk. Image via ploughshares.ca

Elon Musk Criticizes Those Eager to Manifest an AI Deity

Billionaire polymath Elon Musk has become a household name among not only the technorati and business elite but the broader public as well, placing himself on the blistering edge of such exciting and potentially world-changing ventures as Tesla, SpaceX, Hyperloop and Neuralink.

But in addition to these sexy ventures befitting Tony Stark (and one “boring” company), Musk has also become increasingly involved at the forefront of AI research in recent years. Musk co-founded the research nonprofit OpenAI with the mission of safeguarding humanity by fostering safe and ethical development of AI tech and campaigning for increased regulation and oversight in the rapidly-developing field. Along with many of the AI world’s most brilliant minds, Musk fears that unchecked AI development could pose existential risks in the not-too-distant future.

To this extent, Musk has made some very public –some would say sensational– pronouncements about the risks of AI, most famously likening the development of AI to “summoning the demon” and, earlier this summer, engaging in a public feud with with Facebook founder Mark Zuckerberg over AI’s role and risks. And this week Musk is at it again, with this short but damning commentary on twitter:

It’s unsure precisely whom in the article Musk is referring to, but he seems to be in strict opposition to the foolishness of creating an AI deity or religion, and one person featured squarely in the article is trying to attempt exactly that: Anthony Levandowski.

As we covered earlier this month, former Google Engineer and ex-Uber employee Levandowski has become infamous not only for his alleged theft of trade secrets from Google, but also for founding Way of the Future in 2015, a non-profit religious organization created “to develop and promote the realization of a Godhead based on artificial intelligence and through understanding and worship of the Godhead contribute to the betterment of society.” Regardless of whether Way of the Future represents a sincere theological movement dedicated to discovering an AI or is nothing more than a con-man’s tax-exempt cult, Musk seems decidedly determined to keep the reigns of AI development away from the likes of Levandowski.

And while Levandowski’s religious group–whether bona fide or not–may seem of little concern to someone like Musk, the profound (and profoundly terrifying) reality is that within the next century we may very well have created something that is, by most definitions, a god or at the very least more god-like than anything ever realized by mankind. One of the central concerns of Musk and other AI thinkers is the immense power of  Artificial General Intelligence (AGI), AI that matches humans in cognitive abilities and intelligence across all domains.

The creation of a digital human-level entity in itself is enough to prompt existential discourse about the nature of life and intelligence, and one can easily imagine how the birth of AGI will raise all manner of theological and philosophical debates, but Musk and other technologists worry even more about the immense power of Superintelligence (Artificial Superitelligence – ASI).

A self-improving ASI would be an almost inevitable step up from AGI, as a machine capable of general intelligence would be able to creatively leverage the entirety of human knowledge on the substrate of a computer system vastly more fast and powerful than our own feeble minds. Debates on how ASI may look and act or even its possibility vary, but given the present rate of technological advancement it seems not only possible but indeed probable that we will create AGI and that AGI will in turn give rise to ASI that could be as intelligent to us as we are to ants and amebae. (Note: for an excellent and highly recommended primer on AI and ASI, see Wait But Why’s article(s) here).

In short, we might be mere decades away from creating something indistinguishable from a god; and given the stakes, I feel much more apt to trust Musk’s caution than the unbounded optimism of Levandowski and others seeking to immanentize an “AI godhead.”