Granted LLMs aren't very good at the ML family of languages. Isn't vibecoding a lot safer with a typed language? Going from Haskell to Python is going from one end of the typing spectrum to the other.
No. Steve's rage was justified, IMO. It was because Eric Schmidt was on Apple's board while simultaneously being Google's CEO and Google was surreptitiously building Android at the time. Mother of all conflict of interests.
There was a recent story that reminded me of it. Mike Krieger was on Figma's board and Anthropic's CPO, while Anthropic was surreptitiously building Claude Design.
If they were wary of dying, they wouldn't be in the business of terrorism in the first place. Also, they almost certainly believe in a better afterlife. Reminds of the old animated short: Saga of Bjorn[1].
The local LLM community is now teeming with erstwhile crypto and NFT hucksters who've brought the culture of hype from their former communities with them. There still are a few deeply technical people left, but their voices are being crowded out by the vapid marketers'.
When I was growing up in India, the family physician would prescribe single dose deworming syrup/tablets for kids, every year. I recently learned that Mebendazole is available over the counter, here in Ireland (also in the UK). But I have enough Albendazole from my trip to India last year, to last a couple of years.
This is a big fear of mine. I have a course of albendazole once every year just for this. It is de facto over the counter in India. I bought enough to last 4 years, the last time I was there.
You're absolutely allowed to release 20% time projects publicly. As in any large bureaucracy, there's a process for that which is taught during onboarding. What you're not allowed to do is skip the process. There's nothing Google specific about it and I've seen similar firings at other companies too. Skipping legal and corp comms review on any external public communication is grounds for termination.
For locally hostable image editing models, the edit variant of the recently released Boogu-Image[1] model is very good. Anecdotally, I'd say way better than Flux.2 Klein 9B and Qwen-Edit.
I’ve been using oh-my-pi with GLM-5.2 xhigh as the main model and GPT-5.5 medium as its advisor model. IMO, the combo works better than either of those models alone.
https://x.com/jeethu
Building: slopornot.app, privatellm.app and cleanlinks.app