Let’s not forget that almost all memory is made by a cartel of 3 companies known for price fixing. They’re all being as slow as possible about increasing production capacity.

Yes because it takes time and money to build them so you don’t do it on a monthly basis, you do it when the performance gain is worth it

They acquire and somewhat understand it but it doesn’t get saved. It only lives in the context window.

The things the big ones can do now are amazing when you have tasks that can be iterated upon until completion. And then there’s Deepseek V4 flash, a much smaller model with a still huge context window that costs an order of magnitude (nearly 2) less than the American frontier models and managed to do some spectacular things for me with essentially no input from me beyond the original prompt. It took hours, but for me to learn the tools and everything would’ve taken days or weeks.

Still not intelligent the way humans are. Next session starts, it’s all wiped clean until it reads the last session’s notes. But in terms of understanding information and reasoning about it, it’s going to be better than a human that isn’t a domain expert.

I would argue it used to be faster since it was indexed.

Not going to install win 11 to find out if it’s any good again. But in 7 it was.

You get all that through android auto. The manufacturer app is for when you’re not in the car.

They still sell consumer memory IIRC, just not their own brand. They’ll make DDR5 for others at inflated prices though.

You could argue that that’s what Palestinians are saying to the Israelis.

But while we have this ouija board thing going on, there’s things I’d like to ask you about the house, grandpa.

No, it’s a valid question, Jews also used to live in that land millennia ago.

Takes months to train a model, there were already models that got it right when the question was popular, as long as thinking was enabled.

Also if they were optimising for this question, why not update their lower end model (Haiku) as well?

The interesting question would be what percent of humans get it wrong. Smaller than LLMs for sure, but I somehow doubt it’s 0.

Plenty have given up after graduating and being on the job search for 2 or 3 years.

At least as a plumber or electrician you can actually still get a job.

I do. I reverse engineered some proprietary software using an agent. A pro could’ve maybe done it faster, but I did it AFK with little knowledge about reverse engineering.

An agent could similarly try tons of attacks against online targets. Fairly sure some are doing it.

If they don’t do that for GPT 6 when it lands, we’ll know it’s just retaliation for the whole thing where Anthropic didn’t want to give the US military unrestricted access or whatever (as they claim anyway)

Heads up, Kimi is the only one that requires a fucking email to opt out of having your data used for training. Everyone else has a toggle switch.

GLM 5.2 is pretty nice according to benchmarks though.

64 years. 64 bit computing and all that

The carwash thing applies to low end models and older models. Here’s Claude from lowest to highest model, ignoring the banned Fable

Point of a CEO is that there’s one public person for everyone to hate that can be given an even more hated golden parachute if shit goes down. You can’t do this with AI. Replacing AI with AI doesn’t have the same effect on people even if it has the same effect in reality

The US could easily afford single payer. It might even save the government money because the current situation isn’t very cheap either with multiple different government options that all require eligibility checks, etc. The issue is how many people profit from the current situation. Insurance, hospitals… Hell, do you think doctors want a pay cut?

Well there’s still like 2 or 3 computers on it most likely if you go by the definition of computer automotive journalists like to use.

PCM for sure, but could also have TCM if auto, ABS if equipped might have a computer, potentially some kind of BCM…

When they say a modern car has 100 computers, most of those are actually fairly simple controllers and the reason there are so many is that you can just route canbus and power to them and then run the necessary wires to the sensors and actuators from the modules instead of running a bunch of wires from one single controller to everything. Keeps the harness simpler and lighter.

My own 20 year old car has 26 “computers”. 4 of them are door controllers that just actuate windows, locks and mirrors.

Charged by whom? Israeli courts? Yea right.

No, LLMs can definitely be useful for cyber too. It’s the whole reason the US government banned Claude Fable for export.

An LLM can not just try existing exploits like a script kiddy, but with iteration it can try variations and if you know what runs on the server, inspect the source for potential exploits.

They can also look at your setup and say what issues they see (reverse proxy config, etc).

Doesn’t replace an expert, but can be useful for a first pass before you get the highly paid people involved.