Real risk of AI

Goes beyond adapting to evolving technologies

American science-fiction anthologist Groff Conklin in 1954 called Roald Dahl’s short story “The Great Automatic Grammatizator” “an unforgettable bit of biting nonsense”. One wonders how Conklin would react when he realises that Dahl’s short story is playing out — almost word for word — in the real world today. In Dahl’s story, the protagonist creates a machine that computes rules of grammar with mathematical precision and can write a novel in roughly 15 minutes. The story ends on a fearful note as an increasing number of the world’s writers are forced into licensing their creative freedom to the machine. The overarching concern around the future of artificial intelligence (AI) in creative practices so far is primarily pivoted on the fear of losing jobs. The other dominant concern has been the question of fairness in academic environments. Both questions, however, miss the larger picture.

Take, for instance, the most recent few events that have sent out troubling signals around the evolving role of generative AIs. It was one of several issues on the agenda for Hollywood’s film and TV writers when they declared a strike on Monday. John August, a member of the negotiating committee for the Writers Guild of America, said the writers had two concerns regarding AI. They don’t want their material feeding AIs. And they also don’t want to fix sloppy first drafts. Then, Geoffrey Hinton, nicknamed “the Godfather of AI”, recently ended his decade-long stint at Google, citing concern over the future of generative AIs. In an interview recently, he said he was worried that generative AI services would lead to the dissemination of fake information, photos, and videos across the internet.

The comments of Messrs Hinton and August, therefore, go beyond the immediate worry of AIs affecting jobs. Generative AIs work with human language — verbal, visual, and auditory. And language is the very foundation of human culture and history. To truly gauge the magnitude of the threat that AIs pose to occupations across sectors, one must understand how most jobs are rooted in the use of language. In the case of TV and film writers, the job is not merely to write catchy dialogues or weave in the right twist in the right moments of the plot. The writer is also translating an entire time-space on to the screen. For instance, Jubilee, a web series recently released on Amazon Prime, captures the formative years of Mumbai’s cinema industry in the years following the nation’s independence. Here the writer’s job is not merely to recreate the speaking patterns, language usage, and mannerisms of personalities from that era. They are also trying to make this era accessible to the young 21st-century consumer of the web.

From news writers to scriptwriters, professionals have developed conventions, structures, and specific usages of language which help convey not just the text but also layers of subtext and context to their audience. Generative AIs have shown a remarkable capability to replicate not just the form but these textual layers as well. As the code for these AIs becomes more accurate and proficient in hacking into the very core of human culture, they would absorb biases, phobias, and layers of cultural history, and first categorise and then deploy these with mathematical precision. The question, therefore, is not only about adapting to evolving technologies but also to a non-human writer of human histories.