In this blog I discuss The Great A.I. Awakening by Gideon Lewis-Kraus (New York Times Magazine, 2016).

The Great A.I. Awakening

Back in 2010 I once employed Google’s AI-powered Translate service to flirt in Flemish, but my love interest proved unimpressed with the results. My ego burned, Translate went the way of the bin.

As a high school French student in 2006, I’d signed up for the WordReference online dictionary and crowd-sourced translation forum (user name: fat bahstard), and I leaned on the site all through the 2010s as I continued to dabble in foreign languages.

Google Translate I relied on for a quick and dirty on-the-go translation, but WordReference I turned to for a more delicate tournure de phrase or newer corporate and internet jargon.

Interestingly, The Great AI Awakening points out that this was one of Google’s challenges to improving Translate’s ability to produce realistic-sounding phrases and paragraphs. The app’s main use case had not been to translate the entire works the Bronte sisters in one go - people tended to use Google Translate to translate “weird little shards of language.”

In 2016 this article caught my eye because I’d actually noticed that all of the sudden, Translate had been, well, good. Accurate even. And admittedly it’s only gotten better - I recently wrote about my love for the Camera feature on my post Remembering why I love the internet: Searching for crushed tomatoes in Germany.

While the story looks at AI innovation through the lens of translation, the leaps and bounds it covers are responsible for the current AI-powered everything. It reads with renewed relevance in this 2023 era of DALL-E, ChatGPT, and the sudden proliferation of both incredible and unsettling AI.

As the article describes, the rapid acceleration and improvements in neural networks over the last couple of years is a story of multitudes. Refreshingly, it’s not the cliché Silicon Valley story of overly optimistic Zuckerberg-types who lead with technology as the ultimate panacea, or some geek tinkering in his garage til he hits it big time. And it’s not about TechCrunch-style D I S R U P T I O N either.

Rather, the article weaves together three stories - one of Google Translate and company engineers, another of Google Brain and Silicon Valley AI competitors, and the story of deep learning that has unfolded over seventy years across research labs from Canada to Scotland, Switzerland to Japan.