@wtgowers @AlecStapp I'm not a lawyer but possibly yes. I wonder if " busy committing another murder" can be used as reasonable doubt in a legal defense.
English
Manoj Gopalkrishnan
547 posts
Manoj Gopalkrishnan
@manoj_333
CEO at Algorithmic Biologics, former faculty member at IIT Bombay and TIFR Mumbai, researcher in Molecular Computing, Molecular Programmer
Mumbai, India Katılım Kasım 2008
799 Takip Edilen423 Takipçiler
@manoj_333 @AlecStapp Or maybe the point is that in such a situation the two murders would have to be tried together.
English
Just remembered the story about a computer scientist who had his bike stolen and tried to explain binary search to a cop
English
@wtgowers @AlecStapp Here is a Boolean example. If two crimes happened, and it can be proved that you have for sure committed one of them, but it can not be established which one, and these crimes are being tried separately, then you can not be convicted for either crime.
English
@AlecStapp I was once told that the mean value theorem is not admissible in a court of law in the UK. More precisely, if you are driving from A to B and your average speed breaks the speed limit, then you can't be convicted unless a specific time you were speeding can be established.
English
There is a great battle going on in the world today. In one corner is the challenger, Artificial Intelligence, and in the other corner is the reigning world champion Natural Stupidity. Each human alive must pick a side. So which side are you on, Team Stupid or Team Artificial?
English
@satnam6502 Very valuable advice. People who will be honest with us to kill our bad ideas can save us years of effort. "Grapple them to thy soul with hoops of steel" material.
I do think there is merit to a low energy machine learning accelerator chip with hardware native random variables.
English
In 2012 Hinton told me my proposal for a machine learning accelerator chip that exploits analogue characteristics was complete nonsense. I suspect in that case I really was not onto something. Happy birthday Geoff Hinton!
Association for Computing Machinery@TheOfficialACM
Happy Birthday, Geoffrey Hinton! Hinton received the 2018 #ACMTuringAward for conceptual and engineering breakthroughs that have made deep neural networks a critical component of computing. In 2024, he won the Nobel Prize in Physics: #_ftn3" target="_blank" rel="nofollow noopener">amturing.acm.org/award_winners/…
English
For example, sequence alignment algorithms in genomics are incredibly sophisticated. Everyone uses them, very few understand what's happening under the hood in detail.
English
What is the hardest thing in your field that everybody uses but few understand in detail?
English
Manoj Gopalkrishnan retweetledi
The next in the series by #AnanyaGanapathy, featuring conversations with entrepreneurs and technologists about emerging technologies, @manoj_333, founder and CEO of @AlgoBiologics, sheds light on making molecular testing more efficient.
#FuturePerfect
tinyurl.com/4w6zn7tx
English
It is a recently expired passport with old address. New address is in another state. I have applied for passport renewal under the "recently expired passport" category.
English
@passportsevamea I am trying to renew a passport using an existing profile. Despite correctly entering new address in the submitted application, I am only seeing appointment options in the old RPO state. What do I do?
English
"We need policies like congestion pricing which discourages single-occupancy vehicles and incentivises buses, metro, and shared mobility"
Yes please! And cycling lanes. Make Bengaluru roads cyclist friendly.
thenewsminute.com/karnataka/beng…
English
@purwarrahul2 Rahul would love to chat and brainstorm, will message separately.
English
Access of CAR-T is essential for patients in global south- majority of Asia (probably except China and Japan), Latin America and Africa still don't have access. We need to find new ways to make it more affordable to ensure access.
Gaurav Narula MD@DrGauravNarula
Thanks for the shoutout @Satyayadav__ Really happy to’ve presented Indian Cell & Gene Rx journey @ASTCT #Tandem25 Great discussion on the stuff that matters to all! Innovation, Cost/Price, Market reach, Access & Equity @TataMemorial @ImpacctF @CTCTC_tmc @purwarrahul2 @AlbeenaDr
English
Very interesting work @KaixuanHuang1 . Have you compared how humans do? Is it fair to say humans would also be vulnerable to similar "gotchas"? What if we prime the llm to expect such gotchas, does the performance improve?
Kaixuan Huang@KaixuanHuang1
Do LLMs have true generalizable mathematical reasoning capability or are they merely memorizing problem-solving skills? 🤨 We present MATH-Perturb, modified level-5 problems from MATH dataset to benchmark LLMs' generalizability to slightly perturbed problems. 🔗 arxiv.org/abs/2502.06453 🧵 [1/n]
English
"In the 90s, my hometown dismissed education. Now they know better. When innovators succeed visibly, minds change. "
indi.ca@indica
This is a 7/23 interview with DeepSeek founder Liang Wenfeng, autotranslated by DeepSeek itself. How the Price War Began Dark Tide Waves (DTW): DeepSeek V2 ignited a price war. Are you the disruptor? Liang Wenfeng: Unintentionally. We priced based on costs, aiming for modest profit—no subsidies. DTW: ByteDance matched your price, forcing others to follow. Liang: Big Tech’s costs are higher—we never expected loss-leading. It’s becoming an internet-style subsidy war. DTW: Why redesign architectures when copying Llama is easier? Liang: AGI demands structural innovation. Llama lags cutting-edge by two generations. To close gaps in training efficiency and data usage, we need 4x more compute. DTW: Why focus purely on research? Liang: China must join global innovation, not just monetize others’ work. For decades, we’ve ridden Moore’s Law and now Scaling Law as free riders. It’s time to contribute. The Originality Gap DTW: Why did Silicon Valley notice DeepSeek V2? Liang: They’re shocked a Chinese firm innovates, not imitates. But this is normal abroad. DTW: Can China afford such idealism? Liang: Capital isn’t the bottleneck—confidence is. With giants like ByteDance earning globally, we lack neither funds nor talent, just conviction. DTW: Won’t rivals copy MLA? Where’s your moat? Liang: Secrecy can’t stop disruption. Our moat is the team’s accumulated know-how. Open-source earns respect—it’s cultural, not just commercial. DTW: What’s your rebuttal to skeptics like Zhu Xiaohu (VC who prioritizes fast ROI)? Liang: His model suits quick wins. But America’s most profitable firms are tech giants built over decades. DTW: What’s at stake beyond tech? Liang: China’s AI gap isn’t 1-2 years—it’s originality vs. imitation. Without change, we’ll forever follow. NVIDIA’s lead reflects entire ecosystems. China needs its own. Building the Innovation Machine DTW: How’s DeepSeek structured? Liang: Fully bottom-up. No hierarchy. Anyone can requisition GPUs or form teams. We hire for curiosity, not resumes—many are fresh grads or interns. DTW: Jack Clark called your team "cryptic geniuses." Liang: (Laughs) Just passionate youths. MLA began as a junior researcher’s side project. DTW: How do you compete for talent without VC hype? Liang: Tackling AGI attracts the best. China’s top minds are underutilized—we offer hard problems. DTW: OpenAI’s slower pace—is Scaling Law fading? Liang: Progress continues. OpenAI isn’t invincible. We bet on three paths: code/math, multimodal, and pure language. DTW: AGI timeline? Liang: 2-10 years—within our lifetimes. DTW: Will you stay open-source? Liang: Yes. Strong ecosystems need shared foundations. The Hardcore Future DTW: With China’s economy slowing, can innovation thrive? Liang: Downturns breed real innovation. When easy money fades, people build. DTW: Final thoughts? Liang: In the 90s, my hometown dismissed education. Now they know better. When innovators succeed visibly, minds change. We need patience and proof. original: mp.weixin.qq.com/s/r9zZaEgqAa_l… via: pekingnology.com/p/ceo-of-deeps…
English
Manoj Gopalkrishnan retweetledi
@GenomicsCow Boltzmann's law, normal distributions, and Pascal's triangle are the same thing
English
Today India learnt the true worth of @cheteshwar1 Cheteshwar Pujara.
#INDvNZ
English
@RBI @rbinnovationhub this needs urgent attention. Immediate relief through a one year relaxation of kyc requirements, and use the time to find a solution.
Road Scholarz@roadscholarz
1/6 The “KYC crisis” – voices from Chhattisgarh. Shanti (Pahadi Korwa, PVTG) says she "lost hope" of seeing the money she earned from NREGA work and selling tendu leaves, after multiple costly trips to the bank. @DasShaktikanta @RBI @suchetadalal @FinMinIndia
English
@daniel_271828 Had the same general idea a few days back :-) Even better, I got an LLM to write the arguments. Easy to tweak the prompt and make it more academic.
linkedin.com/feed/update/ac…
English
Has anyone written a paper on “Can humans actually reason or are they just stochastic parrots?” showing that, using published results in the literature for LLMs, humans often fail to reason? I feel like someone should write that paper.
English
Manoj Gopalkrishnan retweetledi
First they are funny toys, then they’re our competitors, then they are superior forevermore. The competition window is tiny in perspective, but it dominates our vision. But "losing" isn’t the end, it’s the beginning of a new world of powerful tools that we very much need.
English