Suhas Mahesh (@suhasm) header
Suhas Mahesh (@suhasm) profile picture

Accelerating Science with AI Prev: @OxfordPhysics @rhodes_trust Married to science. Affair with literature. Flings with philosophy

NYC
Joined November 2008
Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)Photo by Suhas Mahesh (@suhasm)
“It’s incredibly exciting and unprecedented.” Schmidt Science Fellows Sebastian Musslick and Suhas Mahesh are exploring how AI can accelerate discovery, while keeping science anchored to society’s goals. Can Science Harness the Power of Machines? shorturl.at/3tiPE
1
9
1,440
Topic in the Book Club was Quarrel. Enjoyed reading from the translated Sanskrit poetry collection by @AnushaSRao2 & @suhasm , a delightful selection of poems. Stumbled upon it through the @bicblr Podcast. Worth a listen and the book is a must on every poetry lover's bookshelf.
2
2
17
3,651
Suhas Mahesh retweeted
Replying to @PadhegaIndia_
A book with each poem a delight and as vibrant as love itself ❤️🧡💙 Thanks @suhasm @AnushaSRao2 for wrapping your book with this stunning cover! #padhegaindia #pigiveaway
1
2
11
1,265
"How do you say "I love you" in Sanskrit? You don't. That is not how the 'Sanskritic" mind works." I chatted with @suhasm and @AnushaSRao2 about their delightful book, "How to Love in Sanskrit" at the @bicblr youtube.com/watch?v=zUsqbdcV…
6
14
2,943
‘The overwhelming obsession of poets in Sanskrit and Prakrit has been love.’ At @IHCDelhi #Samanvay2025, authors @AnushaSRao2 and @suhasm join #ProfAnamika and #ProfSavitaSingh for a fascinating discussion on their book, #HowToLoveInSanskrit. The session unpacks the many languages of love—how words shape emotion, intimacy, and connection across time and cultures. #READWithHarperCollins
1
3
7
1,578
It's common to think that “following along” is the same as learning – like, if you can follow along with a video, book, lecture, whatever, without feeling confused, then you’re learning. While this might “feel” like learning, it's not. The feeling is completely artificial. Comfortable fluency in consuming information is not a proxy for actual learning. Learning is a positive change in long-term memory, and you haven’t learned unless you’re able to consistently reproduce the information you consumed and use it to solve problems. This doesn’t happen when you just “follow along,” even if you understand perfectly. That comfortable fluency you feel while following along is arising from the fact that the surrounding context is already on your mind – you’re not made to pull it from long-term memory. When you feel like you’re absorbing information while passively following along, what you perceive is information sitting in your working memory, not your long-term memory. If you want to test whether information is in your long-term memory (i.e., whether you've actually retained it), you have to actively attempt to retrieve it when it's not already at the front of your mind. You have to put yourself in the position where it's not already in your working memory, and the only way to pull it out of your brain is from long-term memory. If you don’t practice retrieving information from memory, it dissipates quickly and almost entirely. Have you ever had the experience of being unable to remember something despite repeated exposures, because you keep automatically looking it up from a reference instead of trying to retrieve it from memory? That’s happened to me an embarrassing number of times with addresses, phone numbers, directions, etc. And any books you read, movies you watch – the only ones you remember in proper detail are the ones you periodically think about and replay in your head. If you just consume and don’t reproduce then you forget almost entirely. I can’t tell you how many times I’ve watched a movie and didn’t even realize I’d seen it before until I got 20 minutes in and something felt familiar. And even then I could barely remember anything about the rest of the movie, just that it felt a bit familiar. Retrieval is the act of pulling information from long-term memory into working memory, and practicing retrieval under challenging but achievable conditions (e.g., when your memory has gotten fuzzy or there is less priming) is what increases your ability to remember and use information. Each time you successfully recall a fuzzy memory, it stays intact longer before getting fuzzy again. Each time you successfully recall a memory with less priming, its recall becomes less dependent on priming in the future. If you don’t practice retrieval, then the information quickly dissipates. It stays with you only briefly – just long enough to trick you into thinking it’ll stick with you, when it’s really on the way out the door. But, of course, you don’t notice that it’s gone if you’re not actually testing whether it’s there. Consuming information without practicing reproducing it can produce an artificial feeling of fluency while the info is held and manipulated in working memory, but since retrieval practice is not occurring to extend the info’s retention, the info dissipates quickly. The fact that it’s in working memory can trick you into thinking it’s going to stay there, but it doesn’t. Once it’s gone, the only way you can bring it back without reloading it from an external reference is if you’re able to retrieve the info from long-term memory. But if you don’t practice retrieval, you won’t be able to successfully retrieve. When all you do is consume information, you put yourself in a situation where the only way to load it back into your working memory is to re-consume it. This is why learning really amounts to increasing your ability to recall information from long-term memory unassisted, an action that can be trained by repeatedly performing said action in gradually more challenging contexts, just like strength training. Additionally, learners typically do not process all the key information as they consume, but they are unaware of this until they attempt to answer a question or solve a problem that requires them to retrieve some key information from memory. At that point, they realize that they never fully processed that piece of information to begin with, and they have to go back to find and properly process it. The same applies to generalization: learners typically do not fully generalize what they’ve consumed, but they are unaware until they attempt to answer questions or solve problems that require them to generalize their understanding. The way to avoid this problem and maximize your learning is to switch over to active problem-solving immediately after consuming a minimum effective dose of information. I know that might feel a bit jarring, like it’s slowing you down, but it isn’t actually slowing down your learning – it’s only exposing the fact that your perception of learning does not accurately reflect actual learning. Really, it’s speeding up your actual learning, and the only thing it’s slowing down is your perception of learning. Now, you might say “but I had learned so much, and I had it down pat, and then I forgot it all when I focused my effort on solving a problem.” But the thing is, if you can’t retrieve that information from memory at the snap of a finger, after thinking about other stuff or zooming in to focus on a specific problem, it means you didn’t really have it down pat.
5
24
4
189
9,146
Okay, I wrote a post on something like this, doing my own (crude) economic analysis xitter.gq/helicopterosaur/status… The key idea in my opinion is that replacing gas peakers is easier than replacing gas baseload—but how MUCH easier matters a lot wondering what your NREL friend thinks
1
1
2
732
🌹Prema, Pemma, Ishq: Love in Sanskrit, Prakrit, Urdu Join us for an evening of poetry and conversation on Sanskrit, Prakrit and Urdu. Drawing from their new book How to Love in Sanskrit, Anusha Rao and Suhas Mahesh, alongside Amit Basole, will explore how love finds its voice in these three languages, from clever flattery and yearning to heart-wrenching separation. Moderated by Vanamala Viswanatha, (bilingual scholar and translator), the discussion will feature readings of verses and a Q&A with the audience. 🗓️ 6.30 PM, Thursday, 30 January 2025 📍 Indian Heritage Academy, Koramangala, Bengaluru 📔 Reserve your seat: forms.gle/kmciX41rxXYaDEPu6
2
2
9
1,890
.@OpenTheMag features @Ram_Guha’s #SpeakingWithNature: The Origins of Indian Environmentalism, @AnushaSRao2 and @suhasm’s #HowToLoveInSanskrit, #JeetThayil’s #IllHaveItHere, and @sohinichat’s #TheDayIBecameARunner, among their selection of the best books of 2024. Read more: openthemagazine.com/cover-st… #READWithHarperCollins
3
9
1,203
by @suhasm is (still!) my review stack (boo, life). As said before, a monumentally skilful work. The translations are good in themselves & forced me to look at many originals in a new light. xitter.gq/PhiloCrocodile/status/… can't recall a recent achievement in Classics.
1
7
1,069
Suhas Mahesh retweeted
I've put the "Translator's Workshop" preface from this book on my syllabus 😊
1
6
634
Suhas Mahesh retweeted
Behold, my ornate UORBIT plot! Seriously, though, while I agree that UMAPs are not well-founded, they are successful in conveying a certain scientific aesthetic (in this case, the glory of big data). They are perhaps the epicycles of our time?
Biologists, stop putting UMAP plots in your papers! UMAP is a powerful tool for exploratory data analysis, but without a clear understanding of how it works, it can easily lead to confusion and misinterpretation. Link to @simplystats blogpost below.
1
13
8,776
.@ThePrintIndia features #AsakoYuzuki’s #Butter, tr. #PollyBarton, @VGokhale59’s #AfterTiananmen, and @AnushaSRao2 and @suhasm’s #HowToLoveInSanskrit in ThePrint editors’ 2024 reading list. All these remarkable books promise to stay with readers long after they have read the last page. Read more: theprint.in/feature/theprint… #READWithHarperCollins
3
1
8
4,197
It brings us great joy to share that five of our books have made it to the longlist for the tenth edition of the @oxfordbookstore Book Cover Prize.  Congratulations to the authors and designers!  #HariKrishnaKaul’s #ForNowItIsNight, designed by #AmitMalhotra; @AnushaSRao2 and @suhasm’s #HowToLoveInSanskrit, designed by #RashmiGupta; #DevikaRege’s #Quarterlife, designed by #EdmundSpitz; @charunivedita’s #ConversationsWithAurangzeb, designed by @devanganadash; and @itsanandneel’s #Mahi, designed by Shiladitya Bose and Subhadeep Roy - @bagha_gupi. #READWithHarperCollins #OxfordBookstoreBookCoverPrize
1
5
8
1,319
Suhas Mahesh retweeted
Excited to unveil OCx24, a two-year effort with @UofT and @VSParticle! We've synthesized and tested in the lab hundreds of metal alloys for catalysis. With 685 million AI-accelerated simulations, we analyzed 20,000 materials to try and bridge simulation and reality. Paper: arxiv.org/abs/2411.11783 Dataset: fair-chem.github.io/core/dat… Blogpost: ai.meta.com/blog/open-cataly…
9
20
7
107
39,353
Replying to @suhasm
in CP Snow's novel "The Search", the main character's career advancement is halted by an incorrect crystal structure that he published. It feels very foreign from a modern perspective.
1
1
509
Suhas Mahesh retweeted
So @JeremyFarrar asked for an "explainer thread" on Noble Prizes in Chemistry around AlphaFold and Protein design, so here goes.
18
140
30
522
200,587
From C. Truesdell, _Essays in the History of Mechanics_ (Springer-Verlag, 1968), p. 330 #HistoryofScience
2
5
831