Juhi Parekh

I was part of a group that said the same thing about Scale AI when we met Alex back in 2019 at a fellowship. It’s one of the top 3 things needed to make the magic happen, and its importance is only increasing with time. I get the argument, as the models get better, the business does get harder and the revenue isn’t recurring. These businesses do have flaws, yes. But I’ve never seen such a high concentration of exceptional people in the same place. It’s true for almost all the names here. The future for them is unpredictable, yes. But so is life right now. Excited to see how they do, but I’m definitely always cheerleading for them from the sidelines. What they do is EXTREMELY hard, and I don’t think we’d have this technology, which we all love, without them. They’re doing the most thankless job, and it’s unfortunate that we don’t give them much credit. They deserve a lot more. Thank you guys for working 18 hours a day non-stop to make the magic happen, which you never get the credit for, I hope that changes in coming months & people realize the true value of the work you do.

so just to recap this week (so far) - musk industries is real (spacex, tesla, xai merger) - clawdbot explosion leading to a bankrun on mac minis but then anthropic released their own version - tesla dropped the bomb they’re halting production on model s and x to scale 1M optimus humanoid robots this year instead - china dropped the mother of all open source models kimi k2.5 that turn video into production-ready apps but then google dropped a gemini update ON THE SAME DAY that does the same thing gg - google said fuck it and also launched the worlds greatest world model genie and switched on gemini for 3.8B chrome browser users AND released alpha genome model that one-shots 1M dna base pairs for 3000 researchers across 160 countries AND teased new veo model - microsoft crushed earnings, launched a new ai chip but stock still tanked 10% because they *only* grew rev 39% - anthropic round 2X oversubbed raised to 20B 🏌️ - openai raising another $100B, 750B val 🏌️ - intel leaked they’re gonna help produce nvidias next gen feynman gpus - hello americas tsmc - a robot (built by figure) washed the dishes with zero human interaction - apple acquired stealth startup for $2B that can lip read - integrating their tech for new ai consumer airpods with cameras and mics - demis confirms google glass 2.0 coming this summer fckin hell

How do you benchmark continual learning? What’s the eval?

Turing built 3,800+ expert-authored physics simulation tasks spanning Python and JavaScript across 2D and 3D environments. Each task included: -A unique simulation prompt -Critiques of AI-generated code for logic, visuals, and performance -Executable rewrites ensuring physical realism, smooth rendering & correctness Results: -90% QA acceptance rate -Thousands of labeled failure modes -Structured metadata across frameworks like PyGame, P5.js, and Three.js This dataset helps evaluate model reasoning, realism, and adherence to physical laws, paving the way for more grounded & embodied AI.

“Just look at the degree on that chick”