Kamal Ghorui
Running a smooth global event isn’t rocket science. Reputational costs of embarrassment can be high
Wednesday should have been Sarvam AI’s day to bask in LLM glory, but the buzz was stolen by a private university and its Chinese robot dog. What the Galgotias team did isn’t surprising. Right from kindergarten, schools commend and reward children for projects polished by parents and paid artists. Five-year-olds are scored on the finesse of thermocol solar systems. Genuine rough-edged effort never finishes on the podium. Team Galgotias merely followed old instincts in choosing store-bought celebrity. That it did so at the world’s largest AI summit is really the organisers’ fault, for they should have carefully vetted every entry.
Let’s get this straight, this AI summit isn’t any old mela. It’s drawn tech royalty, global media attention follows, and there will be a guiding declaration at the end. That’s the template set by Britain in 2023, and followed by S Korea and France in 2024 and 2025. India agreed to host this year’s summit last Feb, so it had a full year to get everything right. Instead, we’ve faced some serious embarrassment. Day 1 was chaotic. People struggled to enter the venue. There were complaints about the registration system and digital payments. Both were missed opportunities. Here was our chance to show the world we’re first-rate, tech-enabled organisers. It was also a chance to sell UPI to the West, but weak connectivity made it a cash-only outing. Some participants complained about stolen exhibits, and then the Chinese robot scandal spoilt the effect on Day 2.
IT minister Ashwini Vaishnaw has apologised for the first day’s mess, but in hindsight, did we really need to make this summit a carnival? By now, India has decades of event management experience. Major shows, like the Auto Expo, have separate business and public days, and that should have been the format of this summit too. That way, all the good stuff that happened over the past three days wouldn’t have been overshadowed by bad press. Sarvam’s a prime example. Here’s an AI that beats the likes of ChatGPT hands down in reading and translating Indian languages. You’d want it on your phone to read menus and street signs away from your home state. You’d also need it for researching Indian texts. And states could use it to speedily digitise old land records, for instance. Sarvam shows that regardless of the heft of Western AI models there is room for niche LLMs. AI isn’t one race but many.
END OF ARTICLE
