Difficulties in GenAi regulations
Nilesh Jasani
·
May 31, 2023

"We must control AI" has no opponents, like a chorus sung unanimously, without dissent. Politicians and activists of all ilk are already raising hackles, imputing whoever they dislike as AI devils. Yet, the ever-shifting quicksand of AI provides no firm terrain on which to erect robust regulatory structures.

The last post argued that there is no one definition of AI. Nobody knows how the current generative models work and produce their output. This ambiguity extends to our understanding of AI's trajectory, as outlined in this 2019 review (https://bit.ly/3MCIvmR).

The clarion calls on controls are right, but the realpolitik of competing nations and corporations has reverse-engineered the game theory established by nuclear weapons. Where nuclear deterrence discouraged initial aggression, AI instigates a race. The reluctant participant, who hesitates in deploying their arsenal, risks catastrophic competitive defeat.

While noble, activists' rallying cry for voluntary restrictions is hampered by diverging, mutable ideologies with broad strokes and scarce detail. Politicians are about to establish watchdogs, ironically destined to serve as everyone's favorite whipping boys for not doing the right thing they should somehow know. The world could not agree on how to act as one on the pandemic, despite a known enemy and a cadre of experts; the question of AI management seems a veritable chimera.

AI control could supersede sustainability as the zeitgeist of these years. Expect a diurnal flurry of statements of intent, remonstrations, political exhortations, committees, and private resolutions against specific applications to fill the news headlines for years to come.

If there were ever a domain demanding global unity, AI could well challenge the environment for the title. Expect the whole world to agree that we should act as one on AI control, but we won't. This writer would support every group, committee, and watchdog that will be set up, but would also remain skeptical about witnessing a genuine deceleration in AI advancement.

Related Articles on Innovation