ingatorp

Stablecoins are indeed centralized. That's why when people say that stablecoins are the solution, i always roll my eyes. Native assets in blockchains like Bitcoin, Ethereum, XRP, Solana etc can't be freezed as they are native assets in decentralized networks.

This is a result of benchmaxxing the models to infinity. If you RL with the goal of only achieving the correct result no matter how you arrive there, then the models will try to get there using any method in their disposal, including cheating.

This happens also because LLMs are black boxes that we know almost nothing on how they arrive at the result they are giving.

Or he wants you to believe he is Satoshi, without being a complete moron, like Craig Wright. Back stands to benefit from this, because he still has money on Bitcoin ventures (Blockstream). If people believe that he is Satoshi, he would still find investors backing him.

I mean Chrome OS started as a browser only OS. So theoretically it can be an OS.

I mean at this point, for that to happen it definitely isn't a matter of intelligence (it can fix errors later and learn from them), it's only a matter of memory and proper harness. Once memory it's solved for good, then recursive self-improvement is inevitable.

Then you haven't had any exciting idea and the need to actually build it. I personally like thinking of different projects and come up with ideas to make them unique. With Claude Code you can iterate like you're on steroids.

I've stopped using Gemini models altogether because of this. I'm using Claude Code with MiniMax M2.5 for a while now and i couldn't be happier. I haven't noticed any drop in output quality and the biggest advantage is that even the $10 is pretty generous. I haven't been hit with rate limit, not even one time. And i'm pretty heavy user. I tried also GLM 5.0 but i hit rate limit there pretty early on.

It was already pretty restricted due to ludicrous rate limiting. I tried it just for fun with my Pro account and it was unusable. It couldn't do tasks properly without hitting rate limit, every other prompt.

They could have used Kokoro which is open source, fast and way more human-like.

You can always disable multi-device, so it can act like a regular OTP auth app.