me551ah

I just use their models via openrouter. I have an account with OpenAI, Ollama , even deepseek platform. But for all google ai stuff, I just use openrouter.

Not only is it easy to pay but I also don’t risk getting my account banned.

That’s the equivalent of a human doing brain surgery on another human.

The fact that they were able to 'freeze' some of the funds and recover it, does that not mean that crypto is also centralized like banks?

The general quality decline in software seems to be accelerating. I suspect it’s mainly due to AI slop. Engineers creating tons of code due to management pressure and lesser people available to test.

30fps in VR is nausea inducing

I average about 2-3 hours per week. I’ve gotten into simracing and an average Le Mans Ultimate race is around an hour which I play a couple of times per week.

The native experiences are too barebones and I don’t even use it untethered

Half-Life Alyx is great, but it still requires a tethered headset and a beefy PC to run with high fidelity graphics.

Beat Saber is still quite basic graphics, I wouldn’t call it high fidelity.

The biggest issue with VR adoption currently is not VR hardware. Quest 3, Pico 4 etc are all great headsets which launched with amazing price points. The issue is that these standalone headsets are pretty much useless. Sure you can watch some videos, or play some extremely basic games which remind me of N64 graphics. But that's about it. To get the most out of these headsets, you need a powerful PC. Driving in Assetto Corsa in VR feels amazing, you get actual realistic car physics and can drive in Tokyo with traffic around you which feels 'real'. VR in Microsoft Flight Simulator allows you to fly in any plane in any part of the world. VR in Elite Dangerous allows you to fly in outer space in a 1:1 scale Milky Way Galaxy. But require a beefy PC to run.

Apple, Meta and other VR companies just keep throwing more and more VR hardware with better lenses and form, but the issue is content. If someone is spending 1200 bucks on this headset and playing/interacting with N64 level content, the novelty wears off pretty fast.

I think that explains the race for IPO by the US AI labs, they know that the longer they wait, the less they will be worth.

> They need to start moving parts of the harness behind the API

This isn't without it's challenges however.

1. This will increase costs drastically, since they would need to a run a sandbox per use to ensure data isolation.

2. Increased latency, and this directly limits how much of the harness can be moved to the cloud before the users notice sluggishness

This makes me wonder if AI companies even have a MOAT in the first place.

All requests to an LLM are idempotent, for every API call you need to send it the entire conversation history so that it can process it. LLMs do not learn or remember anything, which makes it super easy for users to switch LLMs on the fly. Most popular AI frameworks, make this a one-liner change these days.

And that makes me wonder if the trillion dollar valuations for OpenAI and Claude are even justified. Cause if that is justified, then Kimi, Qwen, Deepseek etc are also valued at a trillion dollars. Or all of them are worth a lot less. One of those statements is true.

Also this makes me wonder if the next iteration of LLMs would be based on fine-tuning, where LLMs actually learn from your past behaviour so that it would grant some amount of stickiness to the product. OpenAI used to offer fine tuning runs for GPT-3.5, but they don't seem to do that anymore.

-Chinese labs only have 5-10% the valuation of OpenAI/Anthropic

Another reason could be that they want to close the gap between their valuations and OpenAI/Anthropic.

Either Chinese labs are worth more or OpenAI/Anthropic are worth less. One of those is true

I don’t know why India is mentioned here.

I live in India and have used 1Gbps Fiber since almost 10 years and pay only 40$ for it. Internet access in India is quite cheap and fiber is quite easily available

These are the same people who will give the LLM full write access on the disk and complain that it performed destructive actions.

If you don’t want an AI Agent to read private repos then you do not give the AI agent access to the private repos. This is not a permission bypass issue but a prompt injection issue which can’t be reliably solved at the Agent layer

If you think in terms of ownership, even then digital is not that bad. I’ve owned digital games since Xbox 360 and I can still play them to this day on my Xbox series X.

But not all of my physical games CD/DVDs are in mint condition and some have scratches.

Nobody downloads music these days and everybody just streams. Audio at 24 bit still takes a small fraction of the bandwidth that 1080p video takes, so I don’t understand the hate for it.

I use a DAC by focusrite which can do 24-bit, and if I want to listen to higher fidelity audio on my planer headphones then I should be able to. Why should I limit myself to 16-bit

I can do everything on my bank app from prepaying small amounts of a loan, spend analysis, opening fixed deposits and such.

Claude is also weird for being the only coding assistant that for some reason doesn't support AGENTS.md. Codex, Amp, Cursor all of them support it and read from it, but not claude which forces it's users to use CLAUDE.md instead.

The issue is the higest voted issue on their gitlab repo: https://github.com/anthropics/claude-code/issues/6235

It’s not just the hyperscalers, it’s also the boom in personal projects being launched. AI has massively decreased the bar to entry, and a lot more people need servers compared to before

So where does the api key reside? You can’t ship it on the iOS client since anyone can read and abuse it

AI is not supposed to write excellent code, it is supposed to write the “most likely” code which equates to your average engineer. People who are below the curve are going to see a much higher benefit than people who are higher than the curve.

None of these companies have any MOAT.

LLMs are getting better at a rapid pace and in a year or two, Chinese LLMs should easily catch up with the best of the models. Models have kind of reached a saturation point and that’s why OpenAI and Anthropic are doing an IPO now. Because a year down the line when their lead diminishes even more, they would be worth lesser money.

SpaceX has clients to put payloads into space, but those clients have always looked for good low cost alternatives. It is going to have competition in the future and will lose customers.

As a hypothetical example, what happens if you could train a custom LLM with your own chats. And it responds to people, and gives the same response as you would. Would that be a form of mental cloning?

This isn’t even a research paper.

Is there some code or results from experiments where we can see the speed up?

Even for making new music Suno is a godsend. My workflow has changed from making a whole track to just creating some nice loops with my favorite VSTs and asking Suno the rest. I get a song in the exact BPM and style that I wanted, while saving me a ton of time.

You didn’t break API Gateway or bypass it, you broke the company using incorrect api gateway config.

Your title is clickbait

I’m just wondering how much more RAM and VRAM chrome will use after these changes

With the number of wars that the US have waged over the years including in Vietnam, Iran and supporting Israel. I don’t think even the US has done a stellar job in defending human rights.

If you meant American citizen human rights, then you’re correct.

What am I supposed to do now? Copy code from stackoverflow like a caveman?

Makes me wonder if the Iran war was a result of the same.