Are EVs actually much more expensive compared to PHEV?

The prices seem comparable in the few models I looked at. I still have the comparison open between a base mustang machE and a Ford escape. I picked those because they are both built by the same company and they are quite similar vehicle. The machE starts at 34k and the ford escape at 30k.

I don't know anyone who actually had to be towed because they ran out of charge. Is this commonly happening?

Are repairing them actually expensive? Tesla's are, but it's more a Tesla special than anything else.I have friends with some Nissan Leafs (I feel like I should call them Leaves; but that feels wrong), with Chevy Bolts, with Kia EV6s. They are not telling me the tale of absurd repairs. Maybe they are still in the new-ish phase of the cars, so it doesn't apply to them yet?

I am an HPC person and the answer to that is very simple. NVIDIA makes sure that gemm works well on their hardware, while the other don't. You can't just rip standard gemm codes and get good performance. There is a ton of engineering that needs to happen to make the operations work transparently with what comes before the gemm and what comes after the gemm.

Just look at how complex cutlass is (which is how cublas works) and you'll get a sense of the problem. Now it is not unsolvable, but it's not simple. And the competition doesn't seem particularly interested in funding the effort appropriately.

Very few people have the technical expertise to make that happen and NVIDIA is hiring most of them.

I don't think that making hardware specialized for LLM is sensible. Because the next tech might not be LLM; actually we pretty expect it not to be LLM. So you need something that you can easily convert to the next application.

Many in the space are trying to build or building competing hardware. It is just that NVIDIA has a massive lead in the HW engineering team, in the SW engineering team, and a large existing user base.

Have you tried to work in torch? That code base is a freaking nightmare. It's more slapped together than anything else. You see papers finding optimization in Torch at almost every performance conference and the reaction is always "you are telling me the code is so complicated that making that optimization was not trivial?" And the answer is always yes, it's that messy.

That's a bit nonsensical. In the US, most people do not live "out in the sticks" almost by definition. Most people live in densely populated areas where EVs are viable.
And yes driving cross country takes days. But also virtually no one does that. It certainly does not have a meaningful impact on the sales nationwide.

I don't know that the "time to recharge" problem is a real problem and not an imagined problem. In daily usage you charge at home. So really the "time to recharge" is probably lower with an EV than with a gas engine. It's only really an issue on long distance driving. That forces you to do a 15 minute stop every 2 or 3 hours. But you are recommended to do that anyway for general health reasons. So it is an issue that marketing should address, but I don't get the sense the problem is real. At least all my friends driving EVs don't seem to think it is a problem even if they do occasional long distance drives. Now if the primary use of the car is road trips, then yeah maybe it is not best.

I bought my current ICE car in 2015. I reviewed all the long distance trips I have done with it. All of them would have been fine with a 150 miles range EV. I live in NC and have driven as far as Florida, Illinois and New Jersey. OK, that's not cross country, but that's still a good 20% of the country. I don't think I would drive further anyway.

There are reasons why EVs are slow to take over. But I don't know the problems you mention are real for 95% of Americans. Almost every family has multiple cars with a road trip car and a commute car. There is little reason the commute car can't be an EV.

A significant impediment is that if you can't charge at home then EVs lose a lot of their appeal. So almost every renter is essentially locked out of the market (and that's about 30% of the population). Also you may have a house but not an indoor garage, so you may not be able to plug in at home. And that's before we consider how many people in the US use the garage as storage and leave their car on the driveway.

But producing hardware for retail is quite different from producing hardware for internal business use.
And note that MS produces/license all kind of hardware for retail as well: both laptop and gaming systems.

What do you mean?
At the scale that they are operating, you probably can design special purpose chips that will be significantly cheaper than an off the shelf device.

I am in a city but not in the bibgest cities in the us. And the "cheap" theaters charges $16 per seat.
So yeah, not surprised people are not going.

AI is like XML; If it doesn't solve your problem, use more!

They are asking the court to ensure that they have permissions to do it. This is way cheaper than broadcasting the ad and getting sued later.
Now if they lose the suit, they are just out lawyers expenses.
If they push the ad, they could get sued for a billion dollars. So the liability is a lot lower now.

I am not sure why you would want to make that particular ad SO MUCH that paying lawyers to sue disney for permission makes sense. But maybe the ad is just THAT good!

Of course, this being /. I have read TFS but not TFA.

what is it that they are detecting really. Some text was generated by an LLM is not terribly surprising. Lots of language tools like akin to grammarly are powered with LLMs.

The abstract was partially written by an LLM is not really a problem either. Abstracts are summary and LLM are somewhat decent at that. As long as you proofread for accuracy, it's probably fine.

Now if the paper and the results are LLM generated, then yeah, it's an issue.

But... But... I wrote a todo list app using AI.
Obviously that means it can cure cancer when asked, right?

"so we will make them larger and they will go airport to airport and then what?"

That answer is actually in TFS. The plan is to " land on makeshift dirt runways adjacent to wind farms."

I see what you are saying and I mostly agree with it.
I do think though as we look at technology that they tend to be overhyped when released, bust out, but still continually deliver unforeseen value in the long run.
And I think these LLM and whatever will follow will do that.

The core problem is that it may be hard to filter out.
Yes we got dumb fads; but even the dumbest fad "take it or double it for the next person" takes some time to make. Here you might be seeing a single person pumping 10 of those per day.

If you look past the "if I publish this my hindex goes up", the result does make sense. Yes, it could be transactional, but it doesn't have to be.

The most common reasons to reject a paper are "the method isn't quite right" or "there is a missing point of comparison". And since their own work is the piece that they would know the most, citing that usually fixes the problem in the mind of the reviewer.

I have a guess.

With new policies in the US, profits are expected to skyrocket even though the consumer might get massively screwed.
The policies will likely enable companies to go unchecked, to pay fewer taxes, and to squeeze their employees more, while delivering less reliable value to the consumer. So profit will go up (at least in the short term), the stocks go up too.

A random computer scientist opinion isn't worth much, but that my $.02