Yesterday was a very fulfilling Sunday after a long long time! I spent good 3-4 hours in the morning at the chess club. There a friend also shared a photo from early 90’s which reminded us of our chess times. You can barely see me because I was too engrossed!
In the afternoon I watched the epic Australian Open final between Nadal and Medvedev which lasted for over 6 hours!
It reminded me of the 2008 Wimbledon final which I had watched in the UK with my MBA classmates. Everybody was rooting for Federer and I was the only one backing the underdog Nadal. I always back the underdog! To my surprise and everybody else’s envy, Nadal (and I!) won in 2008 in a fiercely contested match.
Yesterday was a similar occasion and I missed company of some of my friends (though wife was with me for last 1 hour). Like always I was supporting the less favourite – Medvedev. And he was off to a great start! It looked almost as if the match would be over in 3 sets. But Nadal once again showed how gritty he is. He endured, persisted and finally won! Amazing way to set the world record of 21 Grand Slams!
While watching the match and listening to commentary one thing caught my attention at this point in the match 👇
At this point Medvedev was winning hands down; he was leading 2 sets to nil and had a break in the third set. Anybody would have said that Medvedev was winning. And so did the tournament AI (Artificial Intelligence) which consumed all the data and statistics at its disposal and gave us “insights”. Nadal had come back to win a match after being 2-0 sets down only twice. The last one was in 2005 or 2006 and not against a great opponent and certainly not in a Grand Slam final. He was also playing the Aus Open final after 13 long years! So no wonder that AI predicted an easy win and the amazing Nadal proved everybody wrong.
That made me think about the topic I discussed fervently few weeks back – Is AI/ML overhyped?
I quoted one of my favourite authors Nassim Nicholas Taleb. Taleb is against the mindless application of technology and technology taking over human intelligence. He gave one example in one of the talks. He said something to this effect. Imagine that you are crossing a road and a car is coming towards you. All you do is see how wide the road is, how fast is the car moving, and whether you would be able to cross the road or not. You don’t need to know eye colour of the driver, whether he has diabetes, what’s is his credit rating etc. That’s what the Big Data and all such technologies do. Just because there is abundance of data and because we have technology and processing power to use that data, it doesn’t mean we have more insights. Principle of Garbage In Garbage Out (GIGO) applies for most of such Big Data applications. However, the technology companies want you to believe that you are better off in the decision making with al these tools/platforms and humongous data.
Worse yet is the argument that now machines will become so intelligent that they can take decisions for you. While it’s a wishful thinking, we are far far away from that stage. And maybe we’ll never reach that stage. In fact, we should NOT aspire to reach that stage. There is tremendous value in using tools/machines as a force multiplier of human capability and we should harness that. However, the thought of building Artificial Intelligence that can beat Human Intelligence is absurd.
What happens usually is, we become too optimistic and attach too many superlatives to a new technology. AI/ML have made some great inroads into some of the fields and some kind of problems. For example, the AlphaGo (AI based program that defeated the human world champion in game Go) or AlphaZero (AI based chess engine that defeated other chess engines and can potentially beat the world chess champion). The same company DeepMind that built AlphaGo and AlphaZero are now trying to solve some complex problems in the field of medicine (DeepMind was bought by Google few years ago).
All such efforts are great and must continue! But when some IT services company (no matter how big it is) talks about using AI/ML in “reducing customer churn” or “boosting customer acquisition”, all that is pure nonsense. They are not using AI/ML; they are just using AI/ML label.
I liked this funny crack at all these pseudo companies (most of them are still in the old Body Shopping business) who try to sell basic Data analysis/validation and reporting as AI👇
To conclude, I am not saying AI/ML is no good or it is a hype. I am saying that the companies that are making noise about AI/ML are (mostly) mis-selling it and tarnishing the real AI/ML initiatives.