Kunjan Prasad

Tweets about anything and everything.

[Kunjan Prasad @KunjanPSD]

@aleabitoreddit Would a combination of AAOI + SIVE be good enough for patient capital or is it too early? Supplement with LPK/XFAB?

[Kunjan Prasad @KunjanPSD]

@Kenu73 That’s because virtually every CRM supports multi-ID routing.
If one gets blocked, they simply spin up another and continue.

Meta themselves enable this because revenue.
Otherwise, it would be very easy for them to introduce a Promotions tab.

[Kunjan Prasad @KunjanPSD]

@MartinShkreli All noise until we see production implementation.
Even in the best case, if this performs as good as intended at monumentally less compute, the tasks will become more complex and more capability will be needed and hence more compute.

[Kunjan Prasad @KunjanPSD]

@GeminiApp @OfficialLoganK @sundarpichai
Pulling the plug on my Gemini Sub. Nerfed as hell, tool calling and context does not work.

@shantanugoel

[Kunjan Prasad @KunjanPSD]

@shantanugoel That's true for the Indian VC domain.
They only have thesis for the next D2C, the next pet brand, the next loan app.
Find a good overseas VC for anything technical.

[MIT Jameel Clinic for AI & Health @AIHealthMIT]

🔬Researchers @AIHealthMIT + @McMasterU used DiffDock-L + Boltz-2 to test new antibiotics that target bacteria causing Crohn's disease. Specifically, the antibiotic enterololin, which was discovered via high throughput. Published today in @NatureMicrobiol https://t.co/RmJO6a8gYG

[Stanford HAI @StanfordHAI]

Can AI generate new DNA and show us how our genomes interact at a molecular level? The results could reveal novel insights in biology and pave the way for personalized medicine. Meet the scholars behind EVO 2 at the Hoffman-Yee Symposium on Oct. 14. https://t.co/rbgxicx1sy

[Stanford HAI @StanfordHAI]

What if an AI model could predict your risk and progression of Alzheimer’s or Parkinson's? Scholars are building a world model of the brain that could create better predictions for diagnosis and care. Learn more at the Hoffman-Yee Symposium on Oct. 14: https://t.co/rbgxicx1sy

[Kunjan Prasad @KunjanPSD]

Quite true.
Scientific breakthroughs need incubation, especially when it comes to healthcare.

[Derya Unutmaz, MD @DeryaTR_]

As someone who works in cancer research and spent significant portion of his life dedicated to this, I support Sam’s point. The “spend less on entertainment, cure cancer” take sounds simple and feels good. It is also wrong.

The notion that you should devote all AI power to cancer or medical research is a false dichotomy that is not compatible with human reality and is not even as helpful as people assume. You could make that argument about any entertainment, sports, or games: why waste hundreds of billions on these when you could spend all of that on cancer research (currently less than $10 billion a year from the U.S. government, and being cut further). Yet they feel entitled to make these simplistic popular criticisms.

Scientific progress, especially in biomedicine, not only needs AI and massive data centers; it also needs sustained R&D, actual expensive experiments, very robust datasets, faster trials, fewer regulations, and better incentives for such investments. We still have to solve these bottlenecks and invest tremendous amounts of money and resources. Pools of capital are not a single bucket. Consumer spend does not map cleanly to the NIH, biotech, or trials.

It is therefore unfair to attack OpenAI, which is in fact doing much more for the progress of science than most frontier labs, with the exception of Google, which is also devoting huge resources toward these goals. I do wonder how much of their time and money these people who are making these criticism have given to cancer research? Do they ask this to themselves evet time they spend any time and money for entertainment? Why waste it on that instead of donating to cancer research?

It is also important to remember that without fun games, we would not have NVIDIA and current AI. Why? Because games drove GPUs. GPUs powered AI. AI now accelerates drug discovery. The line from “entertainment” to cures is not straightforward. In fact, I have started making educational science videos using Sora 2. I can already see how powerful this will be for training people or get them interested in science, because it’s fun and seeing is understanding. I would not be surprised if one of these AI videos leads to an insight that results in a cure.

Life, education, and research need to be fun, at least while humans are in charge. We need this to sustain our motivation to learn and produce.

Now I will spend some time making a few more Sora 2 videos and will not feel guilty that I did not spend that time finding a cure for cancer. I also know that in science you have to remix unexpected things to arrive at the most creative ideas.

[Stanford Medicine @StanfordMed]

Developed at Stanford Medicine, https://t.co/8mNfh6K6hN is an artificial intelligence-based tool that helps pathologists work faster, collaborate more easily and improve diagnostic accuracy. https://t.co/Qa4tFRPIlg

[Kunjan Prasad @KunjanPSD]

@Kenu73 People don’t do proper research and then shift the blame onto the authorities.
I have been to Jeju and I applied for a Korean visa in advance.
The process was hassle-free.

[Fei-Fei Li @drfeifei]

ha! here is something fun and totally random I've been pondering: as Oliver Sacks has beautifully written - "what is the space between two snowflakes?" Language can describe all the things, stuff, and people in intricate details. But what about the 'space', the 'nothingness' in between all of them? Without this 'nothingness', space doesn't exist, and things to move. The elegant path a butterfly took from one flower to another is as curious and intriguing as the fact the butterfly has landed on a flower...

[Jason Weston @jaseweston]

🌀Diversity Aware RL (DARLING)🌀
📝: https://t.co/MH0tui34Cb
- Jointly optimizes for quality & diversity using a learned partition function
- Outperforms standard RL in quality AND diversity metrics, e.g. higher pass@1/p@k
- Works for both non-verifiable & verifiable tasks
🧵1/5

[Stanford HAI @StanfordHAI]

📣 Announcing the AI for Organizations Grand Challenge, a new competition for scholars to help organizations enter the era of AI. @GoogleDeepMind and @StanfordHAI invite researchers from any university worldwide to submit your boldest ideas. Learn more: https://t.co/67SBSgDIgd

[Kunjan Prasad @KunjanPSD]

@sama I understand that at some point, company needs to take measures for habit change but replacing exisiting models with just GPT-5 isn't working out.
All of my queries are lacking nuance as if they are being routed to 4o-mini.
I am preferring Gemini a lot more now.

[MIT CSAIL @MIT_CSAIL]

Lung cancer is the leading cause of cancer-related deaths worldwide.

A deep learning model developed by MIT & MGH researchers called "Sybil" can accurately predict a patient’s risk of lung cancer up to 6 years in advance by analyzing a low-dose CT scan: https://t.co/uBkkTnt8mW

[MIT Jameel Clinic for AI & Health @AIHealthMIT]

It’s a beautiful day out for day 1 of the @AIHealthMIT Summer High School Bootcamp! ☀️ We’re starting off strong with an Intro to AI/ML lecture from #JameelClinic faculty lead @BarzilayRegina!

[Kunjan Prasad @KunjanPSD]

@ACTFibernet
Will not be renewing ACT again.
Service has been gamed for Speed test but post connection pings are horrible (100-300!).
The app doesn't even let me raise a complaint as it just resets the Wifi midway and then says there is a technical error!
Horrible.

[MIT Jameel Clinic for AI & Health @AIHealthMIT]

Delighted to announce the release of Boltz-2, which demonstrates unprecedented accuracy in predicting structure and binding affinity! Congrats to @GabriCorso and @jeremyWohlwend on this breakthrough achievement!
📄Paper: https://t.co/UFInT5vDf0
💻Code: https://t.co/4J6ldsgNgM

[Kunjan Prasad @KunjanPSD]

@AdityaKhemka5 Please provide factsheet and allocation breakdown. Have been considering shifting over the same.