“Night will be over, there will be morning, The sun will rise, lotus flower will open.
While the bee inside the lotus flower was thinking thus,
The lotus plant was uprooted by an elephant.”
— Sanskrit Proverb
Writing to an intimate audience.
I’ve always liked writing on the Web. It seemed like the best possible way to share what I was passionate about and find a shared audience to indulge in it. Growing up, I was very passionate about building websites on Microsoft Frontpage. I had a Geocities page where I made weekly posts and spent many hours learning from folks across the world there.
Over time, my drive to write reduced as the Web became more centralised. It’s no secret that the Web is a giant walled garden today. We literally spend all our time consuming and creating, on closed off platforms which constantly ask us to play by their rules — erecting barricades between us and a wider audience. And so, it was refreshing to see the resurgence of newsletters and a more decentralised Web in recent times. It’s now become my primary means of exploring new authors and new areas of interest(like coffee!) It felt like the right time to jump on the bandwagon as well.
I’m grateful to have my life’s work be Envision. At Envision, I get to spend a lot of time working with and thinking about the very cutting edge of Accessibility, Computer Vision, Wearables and building a social impact startup. And I think that journey is definitely worth sharing with a wider audience and learning from them as well.
I hope to keep it a regular thing(but one can never know with running an early stage startup) and will focus on what’s on the outer edge of AI, Accessibility, Wearables, Books and what-not.
With all of that out of the way, let’s dive into Edition #1.
AI takes the center stage.
There comes a time when every new technology moves from the fringes to the mainstream. A time when they go from toys to indispensable tools.
Worldwide crises often prove fertile grounds for certain innovations to take hold — a few examples that comes to my mind real quick are that of airplanes going mainstream during World War 1, electronic computing going mainstream during World War 2 and mobile going mainstream post-2008 financial crisis. These innovations easily jumped multiple societal and bureaucratic hurdles and saw widespread adoption simply because they were far superior than anything else at the time. They were an obvious way to do things better.
AI could finally be having such a moment now — right now.
Many AI-driven products that were on the fringes, that might’ve otherwise faced a lot of societal scrutiny before they went mainstream, have now found willing clients in governments and companies almost overnight.
I think it’s simply because governments and companies are unable to adapt to the sheer scale of things in a global pandemic - how else are you going to ensure billions of people stay indoors, following strict social distancing rules and manage the minutiae of their lives(with the excuse that it’s for their own betterment)? AI is simply the best tool around for this — it can make sense of data at a level humans simply can’t.
I’m highlighting two major kinds of trends I’m seeing emerge these past few months. One on the ‘good’ side and the other on the ‘bad’ side - that classification is to simply denote how much of each side is inherently good or bad. With all the different cases I’m going to highlight below I believe one thing remains clear. The genie is out of the bottle and can’t be put back.
It’s then on us to make sure that we remain vigilant, informed and our institutions are robust enough to navigate this new world order. To never forget the tradeoff that we make everytime we as a society choose to rely on these tools.
The Bad: Social Distancing Detectors.
There’s no denying that the scale of this pandemic is global. Learning to grapple with a planetary crisis and the whole host of problems it brings isn’t something humans are capable of solving by themselves. Incidentally, this is exactly AI forte. It can’t substitute human reasoning but it certainly far outdoes it in some areas — surveillance is one of them.
And a few startups are starting to come out of the shadows to usher in this new era of surveillance. To begin with this was a problem that was already creeping up on society as governments world over started tending to as much surveillance as they could get away it. It’s just taken a different avatar now — monitoring people at social settings, at scale so they adhere to social distancing rules. COVID-19 just provides a nice smoke screen.
Landing AI’s Social Distancing Detector is classic example of this new kind of ‘benign’ surveillance enabling tool. Landing AI is founded by Andrew Ng (yes, that Coursera prof.) and is the latest in a line of companies to offer boutique machine learning solutions to whoever can pay for them.
Their Social Distancing Detector is a tool that helps employers enforce social distancing by detecting if people are walking too closely or not.
Checkout this demo video, straight out of dystopia:
Landing AI isn’t unique in their offering of enforcement solutions based on these detectors.
Skylark is a startup from India that ported this same tech to work on a drone and put it in the hands of the Indian government to help them enforce curfews and social distancing rules. Checking out their website reads like a dystopian tyrant’s wet dream — they claim to have tools capable of detecting lies, violent behaviour and even disguised face identification.
The tech by it self, that these companies, use isn’t very novel — training detectors are quite simple(here’s how you can do with zero code) and it’s even simpler to deploy them on edge devices like your phone, your security camera or even a drone. In fact each year new AI research brings quantum leaps in accuracy and this research is brought to market at an ever faster pace.
It’s not a hard leap for anyone to see this will eventually become a tool for tyrants and that future looks inevitable as they go more mainstream. And what’s even more alarming is that none of these companies are acknowledging the privacy implications of such tools. No one knows what dangerous, hidden biases lurk within their datasets or how much can these tools be stretched before they backfire at all of us.
The Good: AI-Assisted Therapeutics.
Drug discovery is often considered to be a crazy expensive problem.
At the time of me writing this there are more than a dozen Coronavirus drugs in development. Many of them still developed in the traditional approach of trial and error — potentially costing us billions of dollars to bring to market. AI can lighten the load for us a little by getting into that process, working in synergy with humans to discover and model the next wave of drugs a lot faster. While (AI + drug discovery) itself isn’t very new the urgency around COVID-19 can speed things up and bring them to the mainstream sooner.
There are 2 companies that I think are tackling some of the most fundamental problems today in AI — DeepMind and OpenAI. DeepMind was bought out by Google in 2014 but they continue to work outside their realm, chipping away at more experimental areas like Reinforcement Learning, Healthcare, etc.
When COVID-19 hit the world, like many companies, they tried to use their expertise to help. They developed AlphaFold, a machine learning system that is helping researchers by predicting new protein structures for SARS-CoV2.
Why is this helpful? Once a protein’s shape is understood, its role within the cell can be guessed and scientists can develop drugs that work with the protein’s unique shape. AlphaFold is today considered by the wider scientific community as “unprecedented progress in the ability of computational methods to predict protein”.
Image from the AlphaFold blog outlining the working of the neural network.
Not just drug discovery, AI-assisted physicians too have been on the fringes of the medical community for a while. With hospitals now overwhelmed in every single department we’ll soon start seeing them make an appearance to help deal with the massive caseload.
This is the kind of future that excites me the most — AI ‘researchers’ perfectly complimenting human ones. AI can sift through the landscape of possible options, something that can take months or years for humans to do, and offer the best possible options for us to pick from.
The ultimate intellectual Watson to the Sherlock that is human ingenuity.
Read: Gandhi - The Years That Changed the World.
As an Indian, a few things get associated with your identity wherever you go, one of those chief associations is that you come from the land of Gandhi. Growing up as an Indian, he seemed to be everywhere — in our currency, on the walls of our government buildings, even in our street names. He stood as the penultimate symbol of virtue and we always defaulted towards respect and reverence for the Mahatma.
But reverence doesn’t always need understanding or acceptance. While I grew up respecting this figure I was never really encouraged to learn about the person and understand(or even question) his actions deeply. It was never a part of my curriculum growing up and I never bothered either. I know this to be to true of most my Indian peers.
Gandhi recently came up on my radar because of a series of scathing attacks on him by India’s right-wing government including a very crass attempt at recreating his assassination.
Watching that video that day prompted me to read more about him — I was curious what about him spooked the hatemongers so much? I picked up ‘Gandhi - The Years That Changed the World’, the most definitive bio you can read about the man, thoroughly researched and written with an unflinching sincerity. It’s the 2nd and final part of a series chronicling his return to India and kicking off the freedom struggle against the British. I definitely urge you to read both parts.
It was interesting to note that freedom from the British wasn’t really the impetus for Gandhi to become Gandhi. It was freedom from many of India’s own evils — untouchability, Hindu-Muslim disunity, rampant classism, women’s place in society and more. In Gandhi’s own words, freedom from the British didn’t mean anything if India didn’t obtain freedom from its own social evils.
Unfortunately, today’s India carries with it a lot of the same social evils from 73 years ago. And just like the then British government, today’s Indian government fears Gandhi’s memories because he stood against those evils with an endearing simplicity.
Then, as now it’s far easier to rule with division and hate than peaceful co-existence.
To me, that’s the key takeaway from his story — a man with a remarkable clarity in his ideals, whose inner compass often went against the deepest social evils of his day with powerful but a peaceful conviction. In a society, such conviction usually demands a heavy price and Gandhi paid that throughout his life — boycotts, betrayals, multiple assassination attempts, with the last one succeeding in silencing the man but not what he stood for.
Another takeaway that made me see Gandhi in a new light was how farsighted he was. It’s a testament to his foresight that when he died he left behind a group of secular, able nation builders — Nehru, Patel, Rajaji — who worked incredibly hard to build institutions that gave young India a solid footing in the years to come. Compare that to his peer Jinnah who left no such nation builders to guide a young Pakistan through it’s nascent, difficult years. That alone altered the fate of both these nations and the world at large.
When leaders across the world have absolved themselves of ideals Gandhi is a beacon of what can be. And that’s why you should know him and his ideals better. So we may someday demand that of our leaders(and ourselves) too.
Thank you for reading the first edition of my newsletter. I would love to hear your feedback on what you think about it. It’s definitely going to take me a few iterations to find my own voice and I can only do so with your feedback. You can also follow me on Twitter here and don’t forget to subscribe if you haven’t already.