Photo credit: www.technologyreview.com
In a thought-provoking essay, Arvind Narayanan, head of Princeton’s Center for Information Technology Policy, along with doctoral candidate Sayash Kapoor, argues for a more measured perspective on artificial intelligence (AI). They advocate for viewing AI as a routine technology rather than an extraordinary entity, countering the prevalent perception that AI constitutes a distinct and potentially superintelligent species.
The researchers contend that AI should be regarded as a general-purpose technology, likening its gradual integration into society to the historical adoption of electricity or the internet rather than nuclear weapons, although they acknowledge the limitations of this comparison.
According to Kapoor, a crucial distinction exists between the rapid advancement of AI methods—the remarkable capabilities demonstrated in controlled environments—and the slower pace of real-world applications. Historically, the practical integration of new technologies tends to lag significantly behind their initial development.
“Discussions surrounding AI’s societal implications often overlook this adoption process,” Kapoor remarked, suggesting that the transformative impacts of AI are unlikely to occur as rapidly as technological advancements. He believes that the transition to functional artificial intelligence will be gradual rather than overwhelming.
The authors present several compelling points in their essay: they argue that terms like “superintelligence” lack clarity and encourage speculation; that AI will not automate every job but will instead create roles focused on monitoring and supervising AI systems; and that it is vital to consider how AI could exacerbate existing societal issues rather than merely contemplating potential new challenges.
“AI supercharges capitalism,” Narayanan asserts, pointing out the technology’s dual potential to either alleviate or exacerbate issues such as inequality, labor market fluctuations, the independence of the press, and the deterioration of democratic processes, depending on its use.
One significant area of AI application that remains notably absent from their analysis is its military use. The rapid integration of AI technologies in defense raises serious concerns regarding the involvement of AI in life-and-death decisions. The authors did not delve into this topic due to the complexities arising from classified information but indicated that they plan to explore it in future research.
Embracing AI as a “normal” technology could profoundly challenge the stance adopted by both the Biden and Trump administrations, which prioritize the development of advanced AI as a matter of national security. This includes measures such as restricting the export of certain semiconductor technologies to China and intensifying focus on data center resources. In their paper, Narayanan and Kapoor describe the current rhetoric surrounding a US-China “AI arms race” as “histrionic.”
Source
www.technologyreview.com