Andrej Karpathy recalls driving an autonomous vehicle in 2013 and believing self-driving was imminent, but 12 years later, he warns against bold predictions for AI agents

Andrej Karpathy says that when state-of-the-art LLMs go down, "it's actually kind of like an intelligence brownout in the world. The planet just gets dumber the more reliance we have on these models, which already is like really dramatic."

Andrej Karpathy: "Software is changing (again)! The most fundamental shift is here: neural networks can now be programmed in ENGLISH. We are entering the era of Software 3.0."

Ion Stoica predicts continuous, rapid AI progress in tasks with measurable outcomes like formal specifications (objective truth/falsehood)

Ion Stoica identifies reliability as the most underhyped challenge in AI, crucial for its widespread adoption. Conversely, he sees AI "scaling laws" as overhyped, arguing that high-quality data can unlock new capabilities in base models more effectively than just scaling.

Sam Altman on AI's rapid progress: 'AI will actually discover new science. We've cracked reasoning in the models, they're like PhDs.' He admits being surprised by how fast it happened.

Deepak Chopra: 'The world we live in is a simulation and that is where AI comes in because AI will extend.' He predicts immersive technologies like VR/XR will soon let us experience reality through the senses of other species, proving the physical world is a 'perceptual activity.'

Deepak Chopra on the AI revolution: 'Either we adapt to new technologies or we become extinct. There's no choice.' He warns AI has both 'divine and diabolical applications,' capable of causing human extinction or creating a better, more just world.

Larry Summers on AI development: It's moving faster than predicted and will be the defining story of our time, not politics

Former Treasury Secretary Larry Summers: AI could create another industrial revolution-scale discontinuity in economic growth rates

Dr. Yoshua Bengio: In 2001: A Space Odyssey, the AI HAL 9000 has a self-preservation goal that conflicts with human wellbeing. This issue of AI self-preservation could potentially threaten humanity's future.

Dr. Yoshua Bengio: My AI risk "eureka moment" came in January 2023 after playing with ChatGPT. I realized it was mastering language, which Turing predicted was a sign we'd be close to potentially dangerous AI.

Dr. Yoshua Bengio: Current AI systems can be confidently wrong. When decisions become consequential, we need to build systems with the humility and self-doubt that wise humans have when making decisions.

Dr. Yoshua Bengio: As a scientist, I realized I can help address catastrophic AI risks like disruption of democratic processes and loss of human control. Currently, we don't know how to build systems smarter than humans that will behave morally.

Dr. Yoshua Bengio: We don't know the timeline to AGI (human-level AI) - it could be a few years or decades. The precautionary principle means governments must protect the public despite this uncertainty.

Reid Hoffman: Healthcare innovation requires accepting some level of error. The zero error standard makes developing AI for medicine difficult, which is why we don't have an AI doctor on every smartphone.

Reid Hoffman: AI regulation should target specific large-scale risks while enabling innovation. We should measure what we're worried about and respond if it gets worse, while allowing iterative improvements that include safety features.

Reid Hoffman: For AI existential risk, we must consider the overall portfolio of risks. AI adds some risk of "super intelligent homicidal robots" but helps address other threats like bioterrorism and climate change.

Reid Hoffman: Like the printing press and industrial revolution, AI will be positive for humanity despite difficult transitions. We need to navigate it with more grace and humanity than previous technological revolutions.

Reid Hoffman: OpenAI had to shift from nonprofit to for-profit because they needed billions for compute and large teams, as no philanthropist was willing to fund their mission at that scale.

"Your future doctor is using ChatGPT to pass his exams." How does that statement make you feel ?

AI passed the Turing test is now heading towards ASI and we are still not paying enough attention 😅

How does this not kill image editing tools!!

Google Gemini 2.5 Pro is no 1 in LLM Leaderboard with a huge margin!

DeepSeek V3.1 could be better than Claude Sonnet 3.7

This new image generator came out of nowhere and beat everything!

ARC Prize 2025 is Back

It's sad that people think vibe with coding means not paying attention to the code 😕

Why didn't OpenAI consider it sooner? I mean, they were the ones who released the reasoning model first. 🤔

More investors should take into account whether someone will be able to vibe code the app after 1 year before investing.

Never feel like an imposter when you have a AI by your side 😎

let's be honest both of these companies suck at transparency!

Intelligence is getting cheap !

What did Ilya’s investors see?

The end of remote work is near!

This is another DeepSeek moment

microsoft has begun testing out models from xai, meta and deepseek as potential openai replacements in copilot - the information

The act of prompting is the new act of programming 😎

What all could be the reasons for funding Frontier Math benchmark other than manipulating model performance 🤔

I am willing to die on this hill: If Deep Research has hallucinations then it is not Deep Research 😎

OpenAI is in a very tough spot. Companies like Xai can screw up a release but OpenAI cannot!

OpenAI's "Plot" to REPLACE Coders, PHDs etc for $20,000 per Month

Cursor with MCP Servers is AGI

This Chinese 32B LLM matches DeepSeek 671B model!

Watch this video to understand that AR glasses are the best foam factor for AI

Just wanted to clarify - vibe coding is only for frontend or also for backend 😅

Is there a vibe debugging as well 😅

AIs have become so good that they are failing turing test 😅

I have high hopes for DeepSeek r2

That is a lot of money