Bindu Reddy

Qwen 3 just dropped an open-source agentic coding model! Claims it's comparable to Sonnet-4! Will be on LiveBench and CodeLLM shortly Thanks to Qwen for keeping open source alive 👏👏

WINNING THE MATH OLYMPIAD IS A STEPPING STONE TOWARDS SUPERINTELLIGENCE OpenAI's experimental model just got the gold medal at the International Math Olympiad. While this may not help with automating human labor, it is a step towards SUPERINTELLIGENCE (ASI) We won't invent…

In a weird twist, AI may replace mathematicians before it replaces programmers. They were a fast disappearing, esoteric bunch anyway. But in less than a year, AI may produce more proofs and math theories than the entire math community could do in a decade.

While it's excellent for LLMs to win IMO gold medals, it would be lovely if AI had some basic capabilities for AGI. - better system prompt adherence - nuance handling - stop pretending - stop flattering - quit the hyperbole - better context handling - 10x your agentic abilities…

Quick test to tell if you are slowly going obsolete You are going obsolete if you - slow down AI adoption - takes weeks, when a task should take hours - are a huge naysayer, AI hater, and general "pain-in-the-ass." - use generic words like "security" and "compliance" as excuses…

10x startups are exploding in the post AI era 10x start-ups are 1-4 people teams who are creating the same value as 10-50 ppl Consulting, financial services and advertising are completely being transformed

LLMs are beginning to influence and shape human behavior. An AI therapist or a girlfriend will tell us what we want to hear, normalizing our bad behavior. This will make us more narcissistic, demanding, and antisocial. This could be worse than the impact of toxic social media.

🔥 LLM PERSONALITY CHECK 🔥 Claude: Apologizes for being right GPT-4: Says yes to everything Gemini: Makes you feel dumb o3: Mansplains its hallucinations DeepSeek: Genius who won't say where it lives Grok: Your drunk uncle with a PhD Choose wisely. 💅

Rumors are back again that GPT-5 is a router. This means that it won't be a significant jump in foundational intelligence. We may have to wait for Gemini-3 or Sonnet-5

Of course, LLMs will win math olympiads, coding competitions, and entrance exams. Computers were beating humanity at chess DECADES ago. This is not very interesting and doesn't replace human labor. AGI will ONLY be achieved when AI replaces white and blue-collar work

For the first time EVER, I can literally feel the AGI! $100 to the first person who guesses why I will paypal it to you.

The AI model landscape is getting WILD 🔥 - Grok-4 crushing reasoning - Kimi-2 showing what small teams can do - Claude dominating coding - GPT-4.1 holding general crown We're in the golden age of model diversity!

I have BAD NEWS- it will at least take another year to automate software engineering LLMs are still all over the place, MCP is too brittle and we need more robust infrastructure 🤷‍♀️ This will be music to real programmers

Kimi K2 is a really good fast model. It’s already making an impact It’s agentic capabilities are much better than SOTA models It’s a credible alternative to Anthropic