GenAI: Wallflowers are now Wall-facers

It's a 4 body problem with GenAI. It's so tough to get that syzygy right- where data availability, data privacy, authenticity and enterprisereadiness align to the T and not cause chaos. But how tough, exactly? 

- Pratima H

“Overly literal translations, far from being faithful, actually distort meaning by obscuring sense.”

— Ken Liu, The Three-Body Problem

“Liars.” The super-intelligent and much-advanced-than-humans species from the other world is thrown off by just this one word. In ‘The Three-Body Problem’ series, the dark plans of aliens (for lack of a better word) to invade this planet go off the orbit just because they hear a story. They listen to a human as he casually narrates a fairy tale about the Wolf and Red Riding Hood. And they get stuck at the word, the idea, beyond rescue. They just don’t understand it. Humans can say one thing and mean something else. And the other person will actually ‘get’ it.

That’s exactly the power and potential of GenAI. It’s the most beautiful, complex and rarefied accomplishment of AI. Because it’s about human language. It’s about words, thoughts, nuances and context. That takes a whole new weight when we are thinking of a country like India- where vocal search, speech-based apps and mobile phones (in the hands of people unable to read and write) are as rife as rich literature, multiple languages and text waiting to be digitized. GenAI has to ‘get’ what humans say here. And if it can do that, there’s no stopping it from ruling this world.

This is where the classical challenge of data availability in GenAI assumes an entirely new complication. It has to dance well with other factors like data privacy, data authenticity, context and enterprise usability. Not that easy a dance though.