Shaan M.Khan retweeted
"When liability exposure approaches 0—when no one pays the price for unverified failures—verification budgets collapse...Deployers flood the Runaway Risk Zone with unmonitored agents, capturing the gains of automation while socializing catastrophic risks."
Peace! 🇺🇸 🇵🇰 🇮🇷
With the greatest humility, I am pleased to announce that the Islamic Republic of Iran and the United States of America, along with their allies, have agreed to an immediate ceasefire everywhere including Lebanon and elsewhere, EFFECTIVE IMMEDIATELY.
I warmly welcome the sagacious gesture and extend deepest gratitude to the leadership of both the countries and invite their delegations to Islamabad on Friday, 10th April 2026, to further negotiate for a conclusive agreement to settle all disputes.
Both parties have displayed remarkable wisdom and understanding and have remained constructively engaged in furthering the cause of peace and stability. We earnestly hope, that the ‘Islamabad Talks’ succeed in achieving sustainable peace and wish to share more good news in coming days!
@realDonaldTrump
@JDVance
@SecRubio
@SteveWitkoff
@SEPeaceMissions
@drpezeshkian
@mb_ghalibaf
@araghchi
Shaan M.Khan retweeted
New on the Anthropic Engineering blog: writing effective tools for LLM agents.
AI agents are only as powerful as the tools we give them. So how do we make those tools more effective?
We share our best tips for developers: anthropic.com/engineering/wr…
This is huge development because the best in industry is open sourced. #GPTOSS
The internet situation in Gilgit Baltistan is hopeless. After a week here, I’m desperate for stable internet to conduct business. It’s critical that @Starlink and other satellite internet providers be permitted to operate in this disaster- and climate-affected region asap. Unable to tag state officials as X considers GB part of India, where those accounts are banned. #Internet4GB
Rewriting prompts is more frustrating than debugging the code. AI Engineering Expectations vs Reality. #PromptEngineering #AI #LLM
Shaan M.Khan retweeted
New from Ilya Sutskever
He talks about AI recursively building better AI.
He argues such systems could compress decades of biomedical RAG into months, erasing many diseases and extending life, yet the same loop could outstrip any safety protocol.
Prediction fails because super-capable agents remain both unpredictable and unimaginable.
From 'The Open University of Israel' YT Channel (Full video link in comment)
It's never enough when you focus on your current situation, but reflecting on where you started brings a different sense of accomplishment. #selfnotes #reflection
Shaan M.Khan retweeted
Stress and anxiety thrive when there’s a gap between your ambitions and your actions. Between what you say you want and what your actions show you’re willing to do to achieve it. Identify that gap, then adjust your actions to close it. The answer is found in the action.
The most impactful skills in today's (LLM) world involve articulating your thoughts and ideas with clarity and context. #LLMS #ContextMatters #AI #GENAI