Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror

Comment Re:Such efforts usually or always fail (Score 1) 70

> Every roadblock you have mentioned is merely a prejudiced perspective that needs to be overcome

I'm very certain that physical compatibility is more than "prejudiced perspective." Fuck man it took twelve years to decide on a connector standard in North America and that doesn't even impact the size and shape of the vehicle like a battery format would.

> Battery swapping *IS* the way forward until the tech changes

You know that even among vehicles that are swap-capable, the most common form of recharging is like, still plugging in same as non-swap-capable vehicles, right? Swapping is just one solution to an edge case.
=Smidge=

Comment Re: small service center or car wash (Score 2) 70

> unlikely to be insurable

This is in China.

But if you want to make it US/EU-centric, I just need to point out that we have zero problems putting stockpiles of flammable and explosive materials right next to commercial and even residential zoning.

> Or are all those videos of EV fires that can't be extinguished fake news?

Yes, because they absolutely can be extinguished. There are published procedures for extinguishing EV battery fires (spoiler alert: It involves water, and not even all that much water if done properly) and safely handling them after they are out. You have a lot of ignorant people saying what you are saying without evidence, when there is lots of evidence to the contrary.

> I do know they managed to ignite the worry in me.

Ignorance tends to have that effect on people, yes. My sympathies but the situation can be fixed through learning.

No, EV batteries do not "make their own oxygen."
No, EV batteries can not burn underwater.
No, putting water on an EV battery fire will not cause the lithium to burn/explode.

=Smidge=

Comment Re:ChatGPT is not a chess engine (Score 1) 132

My job doesn't have much to do with this at all. All humans engage in motivated reasoning and other cognitive biases. But it is also very easy to think someone one disagrees with is engaging in some sort of cognitive error even when they are not. So instead of just labeling this as motivated reasoning, maybe you could explain what it is wrong with the point I made?

Comment Re:What? (Score 1) 278

Not just a question of legality, but of ethics. Jimmy Carter had to give up his peanut farm out of a concern of conflicts of interest. Obama also had a policy of turning away even gift books from authors that were sent to the Whitehouse when we was President. George W. Bush had a policy almost as strict as Obama's. How far we've come from that point.

Comment Ok, so AI music is not 'real'? (Score 2, Interesting) 129

Before drum n bass was a thing, dance halls were enamored of speed garage, eurolounge was all over, and raves were, well, raving, I had already made a couple of analog synthesizers, one intended for a guitar pedal chain, which got used by a vibraphone artist who scared the heck out of me.

I stumbled into electronica, not the disco-in-a-box crap, and started experimenting with all that. Splurging for a TB-303, my first 'purchased' instrument, I started sequencing and stuff. Adding in some filters and whatnot, I got with a soccer buddy and we gave some tapes to the DJ I was working with (lights and video), and they got mixed in to blend from, for instance, from BeeGees to Frankie. Ugly, but kept everyone dancing...

And I never thought of it as music. I had no training. Rhythms I hacked at until I got something that sounded right. Tempo was easy to fix. Making a bbd pitch corrector based on a Sony design cost me 3 months but fixed some analog stuff. But I was just making or using tools to make sounds. Music? Welllll....

And now I hear AI 'generated' music, and it's actually recognizable as music. As if disco with drum kits in a box and 66 key synths spewing orchestra hits was 'music'...

Well it was, and this AI music stuff is, pseudonyms and indecipherable identity not a new thing for bands, and all this is a controversy ginned up by 'artists' who resent competition. They act like poets... Or Boothbay Harbor painters. A pox on them.

Comment Re:Such efforts usually or always fail (Score 2) 70

> You just said there's no infrastructure needed, and then describe putting up thousands of buildings around the place.

You're ignoring the context; "you STILL need to build charging infrastructure anyway" - you do not, at least no more than any other business would need infrastructure which is already local and existing. Contrast to "charging infrastructure" which invokes mental images of huge cabinets of electrical equipment and rows of kiosks all out in the middle of nowhere.

> That's a lot of investment for no current customers

Nio has sold over 750K vehicles and they've only been in business like, seven years? They have customers. I use Nio specifically because they are the only company that's doing this on any scale worth discussing of course.

And I guess it needs to be pointed out after all; You can still charge the cars "normally" using L1/L2/DCFC systems, including home charging. You don't NEED to swap the battery every time...
=Smidge=

Comment Re:Adversarial Noise (Score 1) 57

> Convolutional neural networks are NOT "behind nearly all generative AI models today", as Benn Jordan so casually states. LLMs overwhelmingly use transformer architectures which do not use convolution.

Correct, but the generative models being discussed are music and visual, not language. Would you like to guess what kind of neural network is most common in those applications? (Hint: if you actually watched the video, he explains that the processing of audio is mostly done using visual/spacial algorithms by converting to and from spectrograms)

> Benn Jordan is a musician, not a technologist

Correct, though because he is a musician he has a big stake in the issue of AI use and training/theft. Again, if you watch the video, he's working with actual researchers and is taking the role of a science communicator rather than a principal developer.

> Embedding information in audio streams has been proven for decades to always be audible

You did not watch the video. You also clearly didn't bother to do any lateral reading.

> it reflects a flawed mentality that says that destroying the work of others should be the goal

Destroying who's work, exactly? The AI companies? If someone is stealing your work, I don't see why anyone should defend or facilitate that behavior. If they want to train their AI on my work, they can negotiate a licensing deal for the non-poisoned versions. (And if your argument against that is it would be too expensive or laborious to get permission from all the authors and artists whose work is used to train AI, then you may as well just admit in plain language that your favorite hobby horse requires IP theft to exist...)

> Sure, if you're technically illiterate you'd believe that.

Man you sure do love being wrong don't you? If it wasn't a problem there wouldn't be so much research into solving the problem.
=Smidge=

Comment Re:Such efforts usually or always fail (Score 4, Informative) 70

There is almost no infrastructure required for this; It's a single building about the size of a small service center or car wash. In there is all the charging infrastructure you need, too.

The problems with battery swap are nothing related to what you mentioned. The real hurdles to mass adoption are questions about battery ownership, standardization, and to a lesser extend vehicle integration.

Since an EV is useless without a battery, laws in some countries (like the US) make selling an EV without a battery basically impossible. Even if you lease the battery separately. It would be akin to selling a new car without an engine; since it's not in a functional state, it can't be sold as a road-worthy vehicle. Countries like China don't have this hurdle so manufacturers like Nio and BYD have been rolling out that kind of business model - Nio alone has thousands of swap stations.

Standardization also generally means you are vendor locked. The battery pack needs to be physically and electrically compatible with your vehicle. It's not impossible but it's very unlikely that the industry will develop such standards especially when it puts hard constraints on the shape and size of the whole vehicle that needs to be build around the pack. Again, not a problem for single manufacturers but if you buy a Nio you will only ever be able to use battery swapping at Nio locations.

Finally integration; To reduce costs some manufacturers (read: Tesla) tightly integrate the battery pack into the structure of the vehicle making it impossible swap without a multi-day ordeal. Imagine trying to battery swap the Model Y where the top of the battery casing has the front seats and center console bolted to it, because it's also the floor of the cabin.

> With that factored in, what are you paying per swap? 3-4x what normal recharging would cost?

From what I understand, Nio charges a flat fee of 180 renminbi, or about US$25, per swap. I'm aware of some schemes that also charge based on battery SoC but I don't think Nio specifically uses that business model. Remember; under this system you do not own the battery, you're basically renting it. This also makes the car cheaper up front.

> Besides, battery tech is improving daily. Increasingly seeing mentions of 5-6min recharge times.

I have not seen any credible demonstrations of "5-6 min recharge times" though I've seen plenty of sensationalist tech news headlines about the possibility of such... relatedly, a battery swap does take about 5 minutes.

And the fact that battery tech is constantly improving is an argument in favor of the swap model, since you can potentially upgrade to a better battery automatically at no cost or inconvenience.
=Smidge=

Comment Good way of getting a list of companies to avoid (Score 1) 32

The current AI systems have some definite use cases, but right now outside some very narrow areas (such as some customer service oriented jobs and some of the more basic programming jobs), the efficiency increases are too small to reasonably justify reducing headcounts based on them. Seems like a good way of identifying areas where management is on a hype-train which can cause real damage to the companies and the quality of their services.

Comment Re:ChatGPT is not a chess engine (Score 1) 132

Pages of instruction are not the only thing that matters. Lots of humans don't learn well from simply reading instruction sets. And since ChatGPT doesn't have a good visual representation of the board, this is equivalent to trying to teach a human who has never learned to play chess to learn to play without a visual board and only able to keep track of moves based on the move notation. Even some strong chess players have trouble playing chess in their heads this way.

Comment Re:Adversarial Noise (Score 1) 57

Adversarial noise isn't "noise" like static or random junk. It's specially crafted to make the model see things that aren't visible to humans, to alter their behavior.

Benn Jordan created a pretty good video about audio-specific implementation. Examples include perfectly normal sounding audio clips tricking digital assistants into thinking they're getting voice commands and having music completely misidentified. The practical application means an artist can apply adversarial noise to their work and have it sound perfectly normal to a human audience, but any generative model that tries to train on it will end up producing inappropriate and useless output.

There are also methods to do similar with images. Text may be a bit harder but it's still possible with websites through embedding or invisible text and similar tactics.
=Smidge=

Comment Re:ChatGPT is not a chess engine (Score 1) 132

Obnoxious snark aside, it appears that you are missing the point. Yes, ChatGPT is trained on a large fraction of the internet. That's why it can do this at all. What is impressive is that it can do that even without the sort of specialized training you envision. Also, speaking as someone who has actually taught people how to play chess, you are to be blunt substantially overestimating how fast people learn.

Comment Re:ChatGPT is not a chess engine (Score 1) 132

You shouldn't be surprised that it will try. All of the major LLMs are wildly overconfident in their abilities. I'm not sure if this is more because they've got human reinforcement to be "helpful" or if because they are trained on the internet where there's very rarely a response in the training data of "That's an interesting question, I've got no idea."

Slashdot Top Deals

"It might help if we ran the MBA's out of Washington." -- Admiral Grace Hopper

Working...