Lightship

[brianman]

I'm curious if the Lucid Gravity (DE) is viable for towing a Lightship.

Copilot has this to say to "can you tow with Rivian R1S?":

Yes! The Rivian R1S has a maximum towing capacity of 7,700 lbs when using a weight-distributing hitch. Without one, the towing limit drops to 5,000 lbs.

Since the Lightship AE.1 Cosmos Edition is designed for most EV trucks with an upgraded tow package, the R1S should be compatible, but it's always best to check your specific model's towing specs and requirements.

And then about "what about a Lucid Gravity?":

Yes! The Lucid Gravity has a towing capacity of up to 6,000 lbs2. This means it should be compatible with towing the Lightship AE.1 Cosmos Edition, but it's always best to check your specific model's towing specs and requirements.

The Gravity also features Tow Mode and vehicle-control software that helps manage trailer behavior. If you're considering towing with an EV, factors like range impact and charging accessibility are worth keeping in mind.

Thoughts, opinions, recommendations? Thanks.

 

[knewman]

I queried Copilot to draft a response to your post:

Great question! Given the Lucid Gravity’s towing capacity of up to 6,000 lbs, it appears viable for towing the Lightship AE.1 Cosmos Edition, which is designed for EV trucks with an upgraded tow package. Additionally, features like Tow Mode and vehicle-control software should help ensure a smoother towing experience.

That said, it's always worth checking your specific Gravity model’s towing specs and requirements to ensure compatibility. Another key consideration is how towing impacts range, especially on longer trips—factors like aerodynamics, terrain, and charging accessibility can all play a role.

Have you come across any firsthand experiences of towing with the Gravity? It’d be interesting to hear how real-world performance holds up!

 

[brianman]

Yeah, what Copilot told me was unsatisfying, hence asking here where informed people might have useful info.

 

[TahoeMichael]

No, the Lightship is too much for the Gravity to tow. The Lightship website lists the dry weight at 7450 lbs, and gross weight as 8300 lbs. That means even a completely empty trailer exceeds the Gravity's rated towing capacity by 1450 lbs, and by more than a ton once you've loaded it up. While there's always a little safety margin in the towing figures, I'd consider it incredibly reckless and dangerous to overload a vehicle's towing capacity by such a large amount.

The Lightship team seems to be explicitly targeting EV pickup trucks as tow vehicles, and is making little effort to target EV SUVs. The SUVs pose a real challenge for powered trailers like this, because adding the weight of a drive train and batteries to a trailer to get a long range boost is fundamentally also adding a lot of mass, and thus making it hard for smaller EVs to cope with the forces that towing puts onto the tow vehicle.

 

[jered]

Thoughts, opinions, recommendations? Thanks.

As a professional in the field, I recommend that you never use an LLM/chatbot for research purposes as they have no ability to reason or understand truth.

 

[037]

As a professional in the field, I recommend that you never use an LLM/chatbot for research purposes as they have no ability to reason or understand truth.

You mean they can’t always find the source of truth and re-tell the story as they have no experience of their own?

 

[jered]

You mean they can’t always find the source of truth and re-tell the story as they have no experience of their own?

They are optimized for generating "words that go together well" but don't have symbolic reasoning to back that up. (There's tons of ongoing research around things like Retrieval-Augmented Generation and other techniques to improve this.) The underlying data is no worse than if you do a web search, but the result is presented so confidently that people just assume the answer is correct.

The above is a great example. You (and the language model) are both capable of finding the tow capacity of the Gravity and the GVWR of the Lightship. You're able to determine that 6000 is less than 7500. The language model doesn't have the concept of those things as "numbers" that have a mathematical relationship, and reaches the wrong conclusion (confidently). This is made worse by the fact that they're prompted to sound "agreeable".

This isn't to say these tools aren't useful! I tried Gemini Deep Research the other day and asked it to write a report on the background of a topic I'm pretty familiar with, and it was about as accurate as an analyst report on the same topic. (Not 100%, but neither is.) But you should treat the results of a LLM the same as you might public search results -- even though they sound like a human who is very sure of what they are saying.

 

[037]

I would think AI would get everything wrong but math, math should be the easiest part.

Assuming it understood the assignment…humans do this all the time, try to solve things before understanding what the problem is.

 

[DeaneG]

I would think AI would get everything wrong but math, math should be the easiest part.

AI in its current form is more or less a bullshit generator. The results sound plausible, but there is no explicit logic behind the words.

I went to reset my fiber modem a few days ago. Searched for instructions on Google, and Google provided an AI-generated step-by-step. Step one was to unplug the fiber modem. There were several subsequent steps in the recommended process involving the reset button, etc. None of the steps involved plugging the modem back in.

I also asked for a plan for a four-day trip to a specific destination. Each day's plan sounded fine, but when you looked at them together, the trip was clearly impossible.

 

[037]

AI in its current form is more or less a bullshit generator. The results sound plausible, but there is no explicit logic behind the words.

I went to reset my fiber modem a few days ago. Searched for instructions on Google, and Google provided an AI-generated step-by-step. Step one was to unplug the fiber modem. There were several subsequent steps in the recommended process involving the reset button, etc. None of the steps involved plugging the modem back in.

I also asked for a plan for a four-day trip to a specific destination. Each day's plan sounded fine, but when you looked at them together, the trip was clearly impossible.

If it’s that bad maybe they should give me a call…

 

[TahoeMichael]

I would think AI would get everything wrong but math, math should be the easiest part.

Assuming it understood the assignment…humans do this all the time, try to solve things before understanding what the problem is.

It's a natural assumption, given that computer software is all basically just math. But LLMs aren't built for mathematical reasoning. There are AI systems designed for that, but it turns out to be an extremely challenging problem. Academics have been working on the problem of mathematical reasoning for several decades now, with only limited success. When it works, the results are impressive. But even the best systems need oversight from a skilled mathematician to direct them, and to ensure the results aren't BS.

 

[brianman]

As a professional in the field, I recommend that you never use an LLM/chatbot for research purposes as they have no ability to reason or understand truth.

It can be a useful tool but healthy skepticism and follow-up is required, like -- well -- the internet.