We own EVs. Ask Us Anything.

[J Alynn]

Its worth noting, that in China, Megawatt or higher charging is not just restricted to commercial vehicles. There are already over 5,000stations, with plans to hit 20,000 stations before EOY.

The plan is also to have it begin rolling out in Europe this year.

EV charging in 10 minutes or less? It's happening

EV battery leader CATL will challenge BYD after securing a new partnership to develop fast-charging technology that will take less...

electrek.co

I know that the US is different. Just calling out that we may see residential 1MW charging in the US in the future But even then, the standard has been to call any DCFC "Level 3", so it would likely still be called an L3 charger, despite being 20x faster than some slower DCFC stations.

Can’t imagine that fight. I’m not a huge fan of AI data centers either but all these types of projects turn the no-growth residents into vigilante like hawks just waiting to show up to a zoning board hearing to argue against growth. That’s where China and Europe beat us as we are still a relatively new country with a lot less years of realizing what smart growth should be

 

[jeremydunworth]

[snip]

As you note, there's a lot of ai nonsense out there. The ai slop will trip up and confuse anyone who isn't serious about reading the actual standards or engaging with good faith intentions to learn in conversations with people who know what they're talking about. Generative ai cannot keep up with changing standards simply because those updates are around the interwebs for less time, are therefore discussed less often than the older standards, and thus are less likely to show up in the training corpus of the ai's LLM. The ai results are worse than useless because they’re simply a statistical slot machine spitting out a string of high-occurrance tokens rather than sensible, reliable knowledge.

Are you in California and available to talk some sense in to some CEOs?
I know it is a hopeless task...

Anyway, excellent concise explanation of the LLM problem in widely discussed and generally poorly understood topics.

 

[SpaceEVDriver]

Are you in California and available to talk some sense in to some CEOs?
I know it is a hopeless task...

Anyway, excellent concise explanation of the LLM problem in widely discussed and generally poorly understood topics.

I’m only a drive away…

But I agree and feel like it is a pretty hopeless task. Even the study that showed 95% of companies implementing ai have lost more money than they spent on the ai (i.e., they became less productive) didn’t help. It seems we just have to wait for the bubble to pop and maybe help the popping along where we can.

 

[R1TVT]

The AI space is so new - its exploding faster than any bubble I have ever seen, but its not all hype. My perspective is that there is good AI and bad AI (or, perhaps a better way to think about it is in terms of good and bad "AI results".

Part of the AI "slop" problem rests directly upon the shoulders un-trained humans using free tools with ill-defined and loose prompts (and no discernment or refinement of outputs). Its classic Garbage in / garbage out, and lots of humanoids clearly just believe what the read as a 1st output.

The opposite is true though. Using an AI tool like Claude and providing the tool with a very detailed and well thought out prompts, that also clearly define a problem statement, an outcome and an objective, while also providing direction, context, access to relevant data & files. Of course in a corp setting this requires more than just investment $$$ for the tool. You need a program, you need training, you need a policy. People's roles are shifting overnight to support deep AI engagement in many sectors because it adds competitive advantage and multiples of efficiency for a business. Companies are churning out new products and features in lightspeed, and if you are in a competitive space, well, do you have a choice? Its basically innovate or die in some industry settings.

 

[strider]

I'm running an 80-gallon heat pump water heater, and it works very well. In the winter, it's still close on capacity since the water coming in from outside is so cold. I'm considering adding a traditional electric water heater as a tempering tank, but hooking an auxiliary inverter output to one or both of the elements. Selling to the grid doesn't make sense where we are, so once batteries are charged, excess solar could be used to heat that pre-tank.

I am doing the same with 2 teenagers in the house. In the Summer it works fine in Heat Pump Only mode. In the winter I have to put it on High Demand Mode (uses both the heat pump and resistive electric) because the recovery time is too slow (garage is colder so less heat to move) since everyone likes to shower at the same time. I hadn't thought about the incoming water being colder. We do have a pre-tank as part of our geothermal system. In the summer it will dump excess heat from cooling the house into this pretank so the water heater doesn't have to work as hard.

 

[J Alynn]

I am doing the same with 2 teenagers in the house. In the Summer it works fine in Heat Pump Only mode. In the winter I have to put it on High Demand Mode (uses both the heat pump and resistive electric) because the recovery time is too slow (garage is colder so less heat to move) since everyone likes to shower at the same time. I hadn't thought about the incoming water being colder. We do have a pre-tank as part of our geothermal system. In the summer it will dump excess heat from cooling the house into this pretank so the water heater doesn't have to work as hard.

I grew up with geothermal and it was great.

 

[jeremydunworth]

The AI space is so new - its exploding faster than any bubble I have ever seen, but its not all hype. My perspective is that there is good AI and bad AI (or, perhaps a better way to think about it is in terms of good and bad "AI results".

<snip>

Much of what you write is valid and one thing I try to do is refer to the specific math or signal processing being used since "AI" is almost meaningless now. So my comment above was specific to LLMs. And I too have had experiences where LLMs perform a task well, for example searching for a figure that was copied and recolored without attribution, an LLM chatbot found a pirate copy of the original, where google image search was useless. Likewise, I've used an LLM speed up writing matlab code, my guess is 4x. But then I write matlab code about a few times a year.

Some of the generative AI videos are shockingly relevant, especially some made to mimic Lego movies. Others are just absolute garbage. So a lot does depend on the human user or editor.

And to bring it back to features in EVs, in my Ioniq 5 I tried the voice activated navigation right when I got the car, which I think was actually Android Auto, and it was aweful. I spoke my home address, and instead of finding it, the voice assistant started going through a list of other places I could have meant since the apparently spanish words pronounced in American English caused it to not find my address on the map. So, yeah, I hope if Scout includes AI in the UI (as someone I know has said "AI is the new UI"), they a do a better job than Android.