The past few weeks have been difficult times for those of us who live in India. We have lost loved ones, family, friends, acquaintances, teachers, bosses, colleagues and neighbours. I took a break from writing here to recover from being infected myself, fortunate to have a mild infection that left me without the need to hustle for oxygen or to fear for dear life. Amidst all this, you worry, you join others in pushing to find beds and oxygen - you are torn between your own illness and the fear of escalation, the desperate needs of others and the grief of loss.
Despite this, Indians are fighting admirably and yet going on with life. You work and you indulge in the same to join in this spirit of getting on with it. You are in awe of your people, their spirit to live, to fight, to help and to be there for each other. We will overcome this, you remind everyone and they remind you. You say, we shall not fear, even if it means tweeting Daredevil from Marvel Comics.
In March 2020, as the first signs of a prolonged pandemic arrived, I was coincidentally in the process of moving my practice into chamber/advisory mode and investing in something I had been looking to build for a while - eliminating the need for humans to spend hours manually reviewing documents. At the time, it was a dangerous cocktail of - the entrepreneur in me who saw an opportunity, the tired corporate lawyer who saw cost and time savings and the “non-”engineer inside who wished to know how much could be done and what was possible.
Over the next few months locked in at home was a great opportunity to learn, to reach out to friends who know better and to understand technologies relevant to my project. Natural language processing is complex, fascinating and moving very quickly, after a long lull except outside academia and niche engineering shops. What was even more striking is what is being built now is based on knowledge that has existed for a long time - but it is only now that the technology companies big and small have started investing at a scale that is transforming how speech, text, voice, communication and publishing work. Given the wide usage of structured language and documents across sectors and the possible range of applications, it became clearer that we are only standing at the beach. The wide expanse of ocean is before us, untouched and waiting. There is also a ship docked lying unused, while some farers have begun some explorations wading across on underused boats. The ship might require work, but its there. The directions are many, and we know what to do. It is for us to map out how to better apply what we know, select the problems that need solving. Isn’t it often that solutions travel with us while we seek the questions?
Much of recent developments in NLP/AI at big tech have been made possible by monopolistic capture of data by tech oligopolies that control e-commerce, advertising and internet/device ecosystems. The more the data, the more we can do by running math and metrics. NLP until sometime ago a relatively obscure corner in computing is now driving chatbots, self driving cars and Alexa, and directing new talent pools towards figuring how we can do more computing with language.
I developed a couple of headlines hypothesis:
Solving the problems of professions that work with voluminous and structured text is difficult but not impossible, and most of the technology required has been with us for over 20 years now.
Recent investments in AI and NLP have created a larger than ever data and talent pool capable of tackling these difficult if not impossible problems.
If these stand true, we are at the cusp of a language-linguistics-tech revolution that will transform how humans and machines interact with each other. Teaching computers how to handle language version 1.0 was about figuring out how we can organise inputs in a way that avoids the need for computers to engage with linguistics in generating output. Version 2.0 involves programming how computers engage with our strange accents, incoherent pronunciations and strange turns of phrase. In some ways, it might appear we are at the beginning of the road. Output generated today appears unambitious and sticklers for accuracy are terrified of whether our promises will match what is delivered.
Cutting back to now, the second wave of the pandemic has been even more cruel and destructive, but we are less fearful than when it first arrived. We are less fearful because we know more, we understand the beast. We have more brains at work, more tools at our disposal and more than anything, we benefit from the lessons of history. The silence of the lockdown is around us, but we know that this too shall pass. In the meanwhile, there is much to do and much to build. There will never be a better time than now to put our brains at work.
When the sun shines again, the work we do today will bring promise and happiness.
Hey, I liked today's piece as well. Looking forward to reading more, and hopefully learning about how you are solving the documentation problems.
Hey, I liked today's piece as well. Looking forward to reading more, and hopefully learning about how you are solving the documentation problems.