Suggestions
Tathagat Varma
PhD Scholar #GenAI #Business #Strategy | LinkedIn Top Voice North America | GCC Leader (4x) | TEDx Speaker (2x) | Entrepreneur (1x) | Published Author (1x)
Dr. Tathagat Varma is the Global TechOps Leader at Walmart Global Tech, responsible for leading strategic business initiatives, enterprise agile transformation, technical learning and enablement, strategic technical initiatives, startup ecosystem engagement, and internal events across Walmart Global Tech.13 He also provides support to horizontal technical and internal innovation programs in the company.
With over 27 years of experience, Dr. Varma has held significant technical and leadership roles in various organizations. He started his career as a Computer Scientist with DRDO and has since played key roles in establishing and growing companies such as NerdWallet, ChinaSoft International, McAfee, Huawei, Network General, NetScout System, 7 Innovations Labs, and Yahoo!.13 He has also worked in engineering roles at Siemens and Philips.
Dr. Varma holds a PhD in Generative AI from the Indian School of Business.2 He is a prolific author, having published a book titled "Agile Product Development" in 2015.2 His research interests include strategy, complexity, culture, dynamics, and innovation.2
Throughout his career, Dr. Varma has been involved in various aspects of technology and leadership. He has experience in building billion-dollar software products, scaling operations to over 10,000+ sizes, and leading cross-cultural teams in high-tech domains including retail, e-commerce, fintech, consumer internet, network management, datacom, consumer electronics, healthcare, and telecom.
Dr. Varma is also known for his passion for learning and sharing knowledge. He reads up to a hundred books each year, is a prolific blogger, and has given TEDx talks.4 His unique experiences include staying sixteen months at Maitri, the permanent Indian station in Antarctica, as a member of the 13th Indian Scientific Expedition during 1993-95.4
Highlights
With #GenAI's illusion of simplicity, it is almost too tempting to assume that the well-crafted responses are too correct! One might almost be forgiven for naively believing in that new urban myth, but such a belief could lead to significant issues in how one uses these tools. It might even be dangerous, perhaps even fatal, if such blind faith in system responses were taken for face value, especially in mission-critical applications for making life-or-death decisions. So, what's the solution? Well, introducing some friction might be an interesting idea. I will give a personal example. Over the last few months, as part of my #doctoral #research, I have been conducting interviews and using software that uses #AI for automatic transcription. I found that even the best ones only offer some 95%-97% accuracy. But here's the catch - I don't really know where all the errors in automatic transcription might reside. So I am literally forced to go over each line and each word of the transcript lest a mistake might have crept in the area that I skipped over. With the human+AI combo, I am able to get to almost 100% accuracy, even if it takes me a good amount of time to go over each part of the text (~1.3-1.5x of the recording time!). So while there is some notion of #productivity (perhaps without such software, I might be required to spend 2-3x of the recording time for manually transcribing the interview recordings, who knows?), it gets offset by the necessities of #accuracy. However, the very presence of errors makes me alert and forces me to suspend a belief that I can readily trust the tool output, and of course, furthermore requires a much higher level of knowledge and skills to inspect, audit, and correct the mistakes than what a normal transcription service might offer. So, perhaps having the #humanintheloop isn't so bad after all - for it triggers a balancing feedback loop that restores and maintains system stability at all times.
So what would happen if there system produced more errors? Of course, I would have a tipping point vis-a-vis my context - clearly, a new tool should help me in some ways, or else why will I consider it? And what if it is good...maybe too good...perhaps perfect...with 100% accuracy. Well, aside from the possibility of probabilistic software as opposed to deterministic software, #AI solutions can never be expected to give exactly the same result as the system state is dynamic, especially when the data is used that is constantly changing. But suppose it could? Well, maybe we might consider what happened when safety belts were introduced in cars, and what was supposed to improve safety actually started leading to more accidents. When we have the illusion of higher safety, we tend to take it easy :). So, maybe we don't need a perfect #GenAI after all? You don't want the driver to sleep at the wheels because, you know, just in case...!
Goodbyes are not really goodbyes, right? Except that we narrowly escaped one today at Bangalore airport.
We had just gotten down our cab and my wife Shikha was holding the trolley in the front. As she hadn't been to the T2 terminal yet, I was taking her pics from behind. I called her to look back so I could get a good shot. Suddenly I heard some noise from my 5 o-clock, but by the time I could react, a car shot past me like a bullet and hit my wife's trolley before hitting the steel bollard. The impact of collision threw her cart, the luggage went flying away....she probably missed getting hit by perhaps just a few inches. She was badly shaken and in extreme shock...but luckily unhurt!
Soon the police and staff of Bangalore International Airport Ltd came and took charge of the situation. They offered prompt help to my wife. There was a doctor and there were a few nurses...they offered medical checkup. Some stranger appeared with a water bottle. Everyone around was well-meaning and offered help whatever was possible! I want to call out the BIAL team for all the prompt help and support right through boarding!
I think Shikha survived because of her faith and love, and I survived due to some very very slim statistical probability of being hit by that flying car. It still gives us goosebumps to think that instead of the steel bollard, it could have been the two of us....and to think that both of us would have been over that very instant. Somehow...today it wasn't to be!
#Grateful #lucky
