Alpan Raval

India's upcoming #AIImpactSummit2026 can shift the global dialogue on #ArtificialIntelligence, with New Delhi potentially positioning itself as a voice for the global south and advocate for AI that is inclusive, affordable, and aligned with regional needs On #GlobalLens, Anjali Kaur, Senior Associate, @CSIS and Former Deputy Assistant Administrator, USAID and @alpanrav, Chief Scientist, @WadhwaniAI, break down how India can lead for the Global South in conversation with @Parikshitl @CSISIndiaChair youtube.com/watch?v=k35Wuh…

@sabeer Actually it ranks 102/123, a minor detail (globalhungerindex.org/india.html). But India's GHI has been significantly decreasing year on year and, while the GHI is serious, it is no longer alarming or extremely alarming. This improvement is what we need, isn't it?

India ranks below North Korea and Sudan on the Global Hunger Index (105 / 127). Let that sink in. Economic growth means nothing if stomachs are still empty.

@achyutchetan This criticism by an IITian of the work of Sen obviously appears to be flippant, but I also think that genuine criticisms of individuals and sometimes even fields should be welcome. A Nobel cannot be a shield from criticism, and should not automatically stand for "the best".

When Amartya Sen won the Nobel for Economics, many of his ideas, and less technical essays, started coming down from the high academic world to common readers. After casually flipping through them, a friend of mine, an IITian, and hence considered brilliant by everyone (including himself), told me that there is nothing great about his work, and anyone could do all that development thinking: a bit of probability, elementary mathematics, a lot of rhetoric, choice, equality and other everyday words. In other words, even a low-ranking engineering graduate from a mediocre Engineering College (In 1998, there were no world rankings of institutions, but in 2011, IIT Kanpur was ranked somewhere below 250 in the world), could presume that he was intellectually superior to a Nobel Prize winner in a social science discipline. Through some strange twists, the Francesca Orsini affair has sparked a war of disciplines––more accurately, an assault on the humanities and social sciences––on social media and other platforms. It's a caricature of what CP Snow had once called the battle between "The Two Cultures". For it to be a genuine war, we need the best practitioners of both 'cultures' speaking to each other, understanding one another, and THEN critiquing each other. In an age of superspecialization, that seems hardly possible, and that is quite understandable. I have no way of proving it, but I know it by intuition, that the best in any field will not berate the best in another field, even if their own ideas lead to very different conclusions. While we celebrate Dante's Ptolemaic geocentric world in his great epic, we do not say Copernicus and Gallileo, and the whole of modern science is a mere superstition. Tagore and J C Bose were friends, but Bose never wanted to write a lyric, and Tagore never wanted to do science. There were many challenges to Freud, including some from his own disciples, but Einstein never questioned Freud's intellect. So when a science or commerce guy criticises top humanities and social science scholars, you know how mediocre they are in their own fields; and the vice-versa.

@AnimaAnandkumar I am not sure if solving nonlinear PDEs should be considered AI for Science. Perhaps AI for numerical solvers is a better term. Imo AI for Science would be discovering the PDEs themselves from data.

How do we build AI for science? Augment with AI or replace with AI? arxiv.org/pdf/2408.05177 Popular prescription is to augment AI into existing workflows rather than replace them, e.g., keep the approximate numerical solver for simulations, and use AI only to correct its errors in every time step. The other extreme is to completely discard the existing workflow and replace it fully with AI. We have seen this approach win in areas like weather forecasting. Such end-to-end AI is significantly better for speed: 1000-million x faster. In our latest paper, we show end-to-end learning also wins in data efficiency, which is counterintuitive. Where do these savings come from? The former approach that augments AI relies only on fully accurate training data that is expensive. But end-to-end learning can use both approximate and accurate training data, if the model can learn how to mix them correctly. In many physical systems, coarse-grid numerical solvers yield approximate data while fine-grid solvers fully resolve the scales and yield exact answers. It turns out that Neural Operators offer a perfect solution when such multi-fidelity and multi-resolution data is available, and can learn with high data efficiency requiring only a small amount of fully resolved data, since it can also utilize approximate training data. In contrast, the standard approach of augmenting AI to a coarse-grid numerical solver (closure model) can only train on fully-resolved simulations, making it very expensive and hard to train. Our results are applicable in multi-scale chaotic systems that have traditionally required running long simulations at high resolution such as climate change or plasma in nuclear fusion and astrophysics. Now you can replace expensive simulation fully with AI (Neural Operators), and also train it without requiring such simulations in large numbers for training in many scenarios.

Instead of reacting to student dropouts, we're preventing them. In Gujarat, our Early Warning System with @EduMinOfGujarat & @UNICEFIndia proactively identifies and supports students at risk. Discover the story behind the system ⬇️ linkedin.com/feed/update/ur…

@dr_alphalyrae If this is not obvious to those working in any technical field, then maybe they should question whether they are in the right field in the first place.

Everyone working in a STEM field should read this - ‘writing is thinking’ nature.com/articles/s4422…

@NTFabiano Have you taken into account the fact that Nobel prize winners are overwhelmingly decided by committees from countries that eat a lot of chocolate? 🙂

Chocolate consumption enhances cognitive function. There is powerful correlation between chocolate intake & the number of Nobel Prize recipients.

@octonion How do you increase diversity of estimates without increasing average individual error? After all the latter is bounded below by the former, and the two aren't independent.

We could interpret this as [collective error] = [average individual error] - [diversity of estimates], which implies we can reduce our collective error by increasing the diversity of estimates, even if the average individual error doesn't decrease. 2/

It's interesting to dive into different mathematical formulations for why a wisdom of the crowd theory might work. One example is to consider the crowd's average guess G vs the true value T. Then we have (G-T)^2 = 1/n [sum (G_i - T)^2] - 1/n [sum (G_i - G)^2]. 1/

Gujarat and Rajasthan never engage in linguistic politics. They only speak the language of business, investments, and job creation.

@im_mansigupta Depends on the city. Mumbai, for example, is very pedestrian friendly.

There's often such a disdain towards pedestrians by drivers, as if sitting in a car makes them better. And often pedestrians have to walk for kms to find a safe junction to cross, hence, they are forced to take their life into their hands everyday

@sonukg4india f is not a function.

#morning with #math

@_amitbehere I returned 16 years ago from a wonderful job in the US to happily live in India. No regrets at all! So there is at least one counterexample to your claim.

Country of birth is a coincidence. Country of residence is a choice. If you are born in a slum, you don't show loyalty to that slum and live there till you die. When you have money, you move to a bungalow. Same logic for country. 99.99% Indians would love the gutter it has been turned into by Modi, but they can't. I feel bad for them. Yeh sasti deshbhakti apni yah apne baap Modi ki gaand mein daal dey.

@octonion This should be quadratic in the number of pieces to leading order. About 25 hrs.

Assume you can solve a 1000-piece jigsaw puzzle in 1 hour. About how long would you expect it to take you to solve a similar, but 5000-piece jigsaw puzzle?

Friends, need your help. @antarikshB, a senior from IIT B has launched an incredible project of organizing all Sanskrit literature in one place, in a user-friendly manner. The service is free, not-for-profit, created purely out of passion. Media coverage will go a long way in ensuring the service reaches the right people. Could you help by RT-ing and perhaps tag the right people? (link below)

@readswithravi Me

Is there anyone else who still favors reading physical books over e-books?

@neonpajamas The Old Man and the Sea, Hemingway

What's your favorite novel under 200 pages? No poetry, no short story collections. Novels & novellas only.

Awesome to see our team's work on newborn anthropometry featured on state TV and radio news networks youtube.com/watch?v=T1ucvY…! 👶 Also, funnily enough our work on oral reading fluency, also deployed across Gujarat, has taught me enough Gujarati to understand this 🤓 @WadhwaniAI

@Saichand_Kasi @sagarcasm You mean Fourier transform. And technically when you have less time you have more “bandwidth” in frequency space, and vice versa. So we don’t use the term bandwidth correctly. When we have less time we should really say we have a lot of bandwidth. :)

@sagarcasm So MBA is like Laplace transform, converts time domain to frequency domain (bandwidth)…

Time ko bandwidth bolte hain, difference ko delta aur majdoori ko hustle

Happy 131st Birthday to Satyendra Nath Bose (b.OTD in 1894), Indian physicist who made significant contributions to quantum mechanics, leading to the development of Bose-Einstein statistics and the prediction of a new state of matter called the Bose-Einstein condensate. Despite his groundbreaking work, Bose never received a Nobel Prize.

@markmeyourze Veena Stores in Bangalore has to be on the top of this list.

Idli- Tier List in India Tier 1: - Cafe Mysore, Mumbai - Saravana Bhavan, Delhi - Gatti Chutney, Pune - IDC, Bangalore - The Filter Coffee, Bangalore - Chutneys, Hyderabad - Mahamudra, Hyderabad Tier 2: - Everyone else