Avatar
arun
66749e7e3301133e813e6d19c8213d5d64a0eabca3ebe3770a232ded96151a2c

This is the only logical take on the MIT report on AI that I could find. Almost all of the other "influencers" have not even seen the actual report and are going by fortune's take on a fundamentally flawed "research"

https://youtu.be/5QzqyrnL010

If you have noted system prompts of many popular chat apps, they have explicitly stated who the current US president is and when he took office. I guess it is to avoid this kind of confusion. #ai #perplexity https://cdn.nostrcheck.me/66749e7e3301133e813e6d19c8213d5d64a0eabca3ebe3770a232ded96151a2c/1cff9499b48294a3e13126699f0461a747c86695f1ab8af9fa9261c9e518fd0e.webp

Unlike Google DeepMind, OpenAI didn’t officially collaborate with the IMO for result verification. OpenAI announced their gold medal performance before the IMO closing ceremony, despite organizers requesting #AI companies wait a week to avoid overshadowing student achievements. https://cdn.nostrcheck.me/66749e7e3301133e813e6d19c8213d5d64a0eabca3ebe3770a232ded96151a2c/1db71cace1b9eae2ef5e459d24fcc2cef07ab3d4ea123c8d32e509c9854cee7a.webp

Amazon acquired Bee, a San Francisco AI startup that makes a $49 wrist-worn device that continuously listens to and transcribes users’ daily conversations

https://www.geekwire.com/2025/amazon-is-acquiring-bee-maker-of-a-wearable-ai-assistant-that-listens-to-conversations/

After using Perplexity Comet I am now clear on how we will get to AGI.

Computer use agents and agentic browsers are not an excuse to not fix broken UX.

Context Engineering - Unlocking Agentic AI’s True Potential

Context Engineering is the art and science of giving the right information to LLMs. As AI systems become more complex and sophisticated, prompt engg does not convey the complexities in deisgn. #AI

https://deepgains.substack.com/p/context-engineering-unlocking-agentic

If you are worried about AI use causing cognitive decline or skill erosion, you are using AI incorrectly. Stop being lazy and use AI to amplify your own abilities.

#AI

The part I don't get about Meta's strategy is how a company exec who specializes in manual data labeling is going to help them get to AGI.

#Meta #AI #AGI #scaleai

Apple Research: AI is not intelligent

Anthropic Research: AI is intelligent

Having used both Apple Intelligence and Claude, I think they are both correct in the context of their own models

Uber for X

is now

Cursor for X

We could be in the “good old days” of open AI (not OpenAI). With MCP and Agents at its peak everyone is opening up their doors to LLMs and agents via MCP. I am afraid that this may not last long and walls will come up faster than we think.

Listen to latest of BG2 podcast.

Codex

Jules

GitHub Copilot Agent

Claude Code with Claude 4

We have now passed the vibe coding phase and moved on to the phase of agentic peer programming.

Google announced AlphaEvolve, an evolutionary coding agent powered by LLMs for general-purpose algorithm discovery and optimization. #AI

https://deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/

Software development is evolving with #AI. Earlier, it was like writing a recipe book where you only had to pay for the writer’s time.

Now, it is more like cooking where you not just paying for the time in the kitchen, but also for each ingredient they use (tokens).