One bad human driver is one bad driver. If an automated vehicle makes a mistake, multiply it by the number of cars in an entire fleet.

They should be issued a ticket for every car on the road, since they are essentially all the same driver and would do the same thing.

In another article I had a conversation about what 'intent' means. They were trying to defend self driving companies and sent me the legal definition of intent (to injure). As it turns out, a self driving company that releases 80% tested driving code and not knowing entirely what it will do in real situation completely falls under the definition of intent of someone is injured.

You are trying to argue that people don't know what school buses mean? I think I have known since I was 5, and I didn't ride a school bus.

You can't just think they are wrong without being able to state how they are wrong.

Yea, when the market has stupid prices for stupid reasons this is the likely response. People will only replace phones, tablets and PC that actually fail, not just to get a trendy newer model. Fortunately there is little reason upgrade as the only difference with older models is the addition of AI features that few buyers actually care about, and, with exception of live translation, are not needed on the device itself.

My first computer had 128B of RAM (MC6802) and 2KB EPROM. When programing in machine code using a hex keypad it felt like a reasonable amount of RAM. Now this machine is using 59GB of it's 64GB, mostly due to Firefox and it's web extensions to block ads etc. While some of the demand makes sense I can't help but wonder how much waste there is with layer upon layer of code that assumes RAM is cheap.

Given that a large percentage of the current AI investment is "me too" investment that will fail, then hopefully there will be a healthy drop in price from the oversupply for the real AI needs.

Errr no. There are insanely minor hardware accelerations at play here. Virtually all of the translation on the M series is handled by Rosetta 2 - a software emulation layer.

Context matters, ARM gaming is insanely niche, far more niche than Linux gaming providing the context includes recognising that tapping on a touch screen is not "gaming". The reality is if you create a game you want to reach the target audience, that is Windows x86. Many people consider the border of "niche" to be some 15% of market adoption. ARM currently is 0.0fuckall% of the gaming market.

You realise what this story is about right? The answer is no. The entire premise here is that bullshit gets retracted from publication.

You're an optimist, but in the past 30 years we have developed fuck all. We've only taken existing developments and mildly improved them. Not only have we in 30 years not stopped a catastrophic rise in emissions thanks to our new inventions, several of them (AI, crypto, etc) have contributed massively towards them, wasting energy without any practical benefit to society massively cancelling out any improvement.

At some point you should read those "legacy" models. You may find that many of them aren't even old enough to walk unassisted without mama yet. Not only that, but no model spits out a single number. They spit out a range that is based on different development trajectories, and those (the most optimistic of them) include radical new technological development.

I'm sure fusion power is ready any day now.

Until then let's just focus on the 20+ year promise of carbon capture that is currently producing 0.000fuckall% carbon offsetting. What wonderous technology we have /s

Who said build? Datacentre companies are bitching about not having power made available to them. They don't want to build it. They can in theory put themselves off grid if they were willing to pony up the cash, but it's easier to be the government for a handout and saddle the public with the high cost of power.

Sorry buddy but that is horseshit. The war on green predated AI. It was stupid then, and it's stupid now.

Fundamentally that's also bullshit especially for steel production and datacentres especially which can be configured to load follow and gracefully shed performance to suit energy, but in general the cost of sustained renewable generation using wind+solar+storage is now cost competitive with building a coal plant. In fact many actual heavy industry sites are migrating precisely to this very thing, including quite a few in Australia where these companies historically run their own power plants. Turns out running heavy industry from renewables is winning on cost. That said I realise Australia != UK, but then heavy industry and the UK don't go hand in hand anymore either.

You really need to update you knowledge of the industry.

I'm sure glad the wireless connections in Canada were all disabled when I bought my vehicle.

Maybe the monster yachts should have been your first clue?

I happen to know a seasoned mechanic personally.. He said 8000km.

Science is allowed to be wrong, if there even was any science behind the people you are going on about. You don't look at mistakes from the past. You look at what is believed now and from that feel free to point out any mistakes you see, because that's the whole point. Again, science is allowed to be wrong.