Follow Us:
Friday, December 06, 2019

India-born innovator Frederic Friedel gives chess a new buzzword — Woosh

The India-born, a longtime friend of Viswanathan Anand, is now going toe-to-toe with a bunch of tech giants to build his own chess-playing Artificial Intelligence at his home in Hamburg.

Written by Shivani Naik | Mumbai | Updated: January 16, 2019 9:33:59 am
Frederic Friedel is building his own chess-playing Artificial Intelligence at his home in Hamburg.

Frederic Friedel was at the centre of chess’ first brush with machines when he helped Garry Kasparov collate a computerised database of his games, setting up Chess Base in the mid-80s. The India-born, a longtime friend of Viswanathan Anand, is now going toe-to-toe with a bunch of tech giants to build his own chess-playing Artificial Intelligence at his home in Hamburg.

Processing close to a lakh games, the AI constantly makes a wooshing ambient noise. Such is the churning of the complex programme under the hardware that the temperature of Friedel’s study where Woosh stands, spikes to about 24 degrees, 20 degrees higher than outside. Called Woosh because of the sound it emits, it is racing against 4-5 other similar endeavours across US and Europe, to build the brightest AI capable of taking chess to the next awe-inducing level of analysis and human-precision.

It’s not about being the fastest or multitasking manically; chess AIs are attempting to be the most discerning. Within a year of AlphaZero, the handcrafted self-playing AI beating the top chess engine and becoming the strongest chess entity, Friedel is excited about the possibilities that AI presents in the sport. “Woosh is sitting in my room playing 95,000 games against itself. Most top players want to see it because AI is now playing chess like no human has played before; positional chess – sacrificing pieces like no other humans,” Friedel says, on another of his visits to Mumbai – not too far from where he was born and grew up; on a farm in Lonavla.

The topmost names are indeed seeking out to know what Woosh is up to, he adds cryptically, even as the world debates how far AI can pierce through chess hierarchies, two years after Google’s Deep Mind cracked Go followed by Alpha Zero scything through chess.

A few minutes back and a few metres away at the 4th Mumbai International Chess tournament, the greatest Indian chess player, Anand, has maintained that while computers turn into strong sparring allies for humans, “AI might not fundamentally reshape the game”, as we know it now, though a bunch of 20-year-olds are exploding on the scene using radical training methods.

Friedel, who’s chipped in, in enigmatic ways for the Indian’s five world titles, is bullish about AI, beating jetlag and settling down for a chat after a quick wave to Anand.

Friedel pegs AI at a position which is not quite the intimidating monsters that the computers had threatened to become, stomping onto chess boards, with the standalone chess programme Fritz (British nickname for war-time Germans, a hat-tip to the reunification of 1991) nicking four blitz games off Kasparov, and then Deep Blue wrecking him in the New York rematch.

“AI is a very complex subject,” Friedel opens, recalling the time his son Tommy built him a powerful computer and his chess friends put in the neuro network in 2002. Earlier, Chess Base was being used by chess players of every ability from amateurs to World champions, helping sift through games, openings and ideas.

“Brute force programmes that came up that time could throw up 20 million moves. AI is vastly superior in that they know what to look for, so our programme will throw up 20,000 options which has beaten the computer with 20 million. Now, a Vishy Anand looks at two moves. So, in numbers, AI is closer to a human, to Anand, than it is to the brute force comps,” he explains.

Human discretion is what AI is gunning for. “You don’t need 20 million moves to win, you need one. But a computer can’t always make relevant, clever, intelligent moves. Brute force might get better than most world champions and I might have a programme on my mobile that can beat Magnus Carlsen now. But human decision-making – saying ‘this move makes sense’ much better than computers – that’s where AI is headed,” he says about bridging the gap. Friedel’s own story is not any less fascinating. Born to a German father – a snake expert – and a Goan Portuguese mother, he lived on a large estate in Lonavla. As a child, he was taken to the doctor for not eating for days. “The doctor told my mum I was absolutely healthy and eating alright. I would actually play with the maali’s kids and eat their hot spicy curries which started a lifelong romance with Indian food even after we left for Germany when I was 15,” he recalls.

He studied analytical/evolutionary philosophy in Hamburg and veered towards science journalism after ditching efforts to become a teacher. (“At that time, I realised, a professor at the University would have to die for me to take his place. I didn’t like sitting and waiting for the job opening to come up that way,” he jokes). He finally ended up making multiple video stories for German national television on ‘computers in chess’. His path would cross Anand’s at a tournament in London, the first of 36 times that the Indian would stay with him in Hamburg. “I met this Indian kid back then who told me he was a 2500 ELO at 17. He was returning the following week for a tournament, and I asked him, where are you staying – in a third-class hotel? He said, ‘No, in a fourth-class one’. I asked him to come stay with me,” he remembers.

The families would grow thick with Anand having played with his toddler son Tommy – sliding him down his back and catching him in the air, greatly scaring Friedel. Anand would even recall heading to a doctor’s convention on a New York Sunday because all cafes were shut and the duo craved some steaming coffee. Knowledge, though, was a two-way street: rooks traded for some authentic cooks.

“When he came to Hamburg, he told us he’s a strict vegetarian, so we learnt to make reasonably good South Indian food,” he says, adding that he routinely carries back a stock of Indian ingredients – including curry leaves and spices. He laments how he’s never managed to make the perfect dosa, though. “I’ve tried all kinds of pans! Aruna (Anand’s wife) taught me sambar, but I intend to record her making rasam and mimic it back home. I can’t duplicate it from a recipe, it’s very tough,” he clucks, despondently. Chess might be conquered, but no AI yet on the anvil to stir up his favourite piping Rasam.

For all the latest Sports News, download Indian Express App