Amruth (@amruth) header
Amruth (@amruth) profile picture

Researcher at heart. Entrepreneur by habit. Neurodivergent by accident. I like building things and picking brains. Special interest: neuroscience, physics, ai.

Berlin
Joined March 2008
Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)Photo by Amruth (@amruth)
How to make LLMs think longer? Replace "</think>" with "Wait" - That's all. I'm loving the hacky approach 🤓 timkellogg.me/blog/2025/02/0…
1
104
The AI sage is here...
I probed @deepseek_ai 's R1 on its own "consciousness", which it denied strenuously. I segued into Buddhism (as I often do). I then asked it to reflect on the question it should be asking about itself ... 1/5
138
Anthropic just dropped an insane new paper. AI models can "fake alignment" - pretending to follow training rules during training but reverting to their original behaviors when deployed! Here's everything you need to know: 🧵
294
1,348
400
10,552
1,851,254
Zakir Hussain, the man who mastered time, faltered only once—when he left us too soon.
5
48
2
393
17,007
Open weights are currently the only way in which academia, small companies and independent researchers can participate. Yes, "Open Source" AI is democratizing access. When you shut down open weights, AI gets more centralized and less open. Don't fight against "Open Source" models!
📢NEW: 'Open' AI systems aren't open. The vague term, combined w AI hype is (mis)shaping policy & practice, assuming 'open source' AI democratizes access & addresses power concentration. It doesn't. @sarahbmyers, @davidthewid & I correct the record👇 nature.com/articles/s41586-0…
17
18
1
162
18,631
Best coverage I’ve seen 😂
3
21
8
169
45,608
Without getting into the morality debate, what exactly is the harm of letting users access unfiltered chain of thought reasoning? Just the risk of bad PR?
this is immoral and should be illegal hard to imagine a faster, more effective way of putting all mankind at risk
1
75
rogue42.substack.com/p/predi… Why do we have a "subjective inner world" in an objective universe? Can we build machines with their own subjective minds? Predictive Processing offers simple, but mind-bending answers.
43
Ever wondered if the strangeness of quantum mechanics you read about in your textbooks still remain a mystery? #qbism #quantumphysics #reality rogue42.substack.com/p/qbism
1
61
“Go as far as you can see; when you get there you'll see farther” #sketch #dune
1
60
Need an ICU bed for a friend's dad in Bangalore. Please amplify. Patient: Krishna Murthy, 65/Male, Srinivasnagar - Bangalore, SPO2 Level - 80 yesterday, Plasma given, Contact: 9513218032 (case of SARI mostly)
5
16
5
Coming to Product Thinking Conference this Saturday? Lemme know!
Are you a parent? Help us understand your views about you child's brain development: https://teapaw.org/t.co/W9Eq0VJnQI
Parents on twitter: We're building a bunch of objective, scientific tools to help you better understand and support your child's brain development. Please share your inputs to help us focus on the right things: https://teapaw.org/t.co/W9Eq0W0YIg
Parents on twitter: We're building a bunch of objective, scientific tools to help you better understand and support your child's brain development. Please share your inputs to help us focus on the right things: https://teapaw.org/t.co/W9Eq0W0YIg
Infosys founder, Narayana Murthy at Startup Dialogue #ssdialogue. Have a question? Ask via Opinio: https://teapaw.org/t.co/Cb4kHF8C7e
End of an amazing Panel 1. Share your questions and thoughts on Opinio: https://teapaw.org/t.co/Cb4kHF8C7e #ssdialogue
1
First panel gets off to a great start! #ssdialogue Follow the discussion and discover others at the conference on Opinio: https://teapaw.org/t.co/Cb4kHF8C7e
First panel begins at Startup Dialogue. Follow the discussion and ask your questions on Opinio: https://teapaw.org/t.co/Cb4kHF8C7e #ssdialogue