GROK Personality Icons Discovered in 2025.20 Update! "Gift of Grok" Coming to CyberTruck Soon!!

[hemiarch]

…so I can argue with my car while driving? Great… No

It can give you a workout over not asking for directions ?

Sponsored

 

[eswimm]

One of the few AI implementations I'm actually excited about, because it plays to the strengths of a language model; being able to use natural language to control the myriad of vehicle options has huge potential.

 

[LiveFreeOrDieBob]

A few thoughts from someone deeply involved with AI (I manage teams that build conversational AI systems for customer service, including customizing models and designing bespoke LLM-based telephone and chat systems):

First, whoever said “you don’t ask AI stuff you collaborate with it” is 100% correct. 3/4ths of the secret sauce of advanced prompt engineering is understanding how to provide enough context for the LLM to accurately traverse the LLM in a way that gets you the answer you are looking for. The way I often explain it to non-technical people is that they should treat an LLM like they are talking to a super smart, high-IQ person who knows everything, but they don’t know anything about what YOU want, so you have to give them context like WHY you’re asking the question, what information you already know, and if possible some examples of something similar to what you are expecting. If there’s a situation where I don’t think I can accurately predict all the important context, I’ll literally add to my prompts “converse with me in a back-and-forth dialog if you need any further information to provide what I need.

AI has a limited “context window” of what it can use as input to find your answer, and in some cases the overall context window includes both the input and the output. So you have to give it as much information in that context window as you can, as succinctly as you can. Learning how to do that efficiently and effectively is what makes the difference between AI being useful to you or not. Models nowadays are far less constrained on the total number of ”tokens” they can use across input and output, but you still have to provide the right context.

So given that it’s almost certain that we’re getting Grok in our trucks, the question becomes what it will be allowed to use to provide context, and then what can it do with it. On the input side, I think it’s pretty certain that it’ll use voice as input, almost certainly by hitting the microphone button on the steering wheel as opposed to Ambient (just listening all the time for questions or keywords), since that would eat up tons of bandwidth and CPU unless they keywords were detected locally. (e.g. “Hey Grok!” and then it starts taking what you say as context) With the noisy environment in a car, even with the directional microphone array you’d stil probably get false barge-in with keywords, so I’m betting it’s going to be through pressing the button on the steering wheel.

The more interesting question is if it can use OTHER channels or types of information as context, or if it’s limited to just voice. Example - what if you were coming up to a fork in the road, and you wanted to go see the ocean, but you weren’t clear if you should take the left fork or the right fork. If you were to press the mic button and say “Grok which way to the beach?”, could Grok access both the front facing cameras (to read the signs) and the Nav system map data and GPS data to see where you are and which way you should go? If Tesla enabled multiple channels of input to the context the model was working from, then that could unlock a ton of possibilities, and I honestly don’t think anyone is more well-positioned to do that than Tesla given their work to train AI based on video input. This is really similar to what they are going to need to collect in order to train Optimus, so they would be getting a lot of value out of those interactions.

Then the other question is - what can Grok effect based on what it comes up with from your query? Can it adjust vehicle settings like the wiper speed, the air vent positioning, and add a new destination to the Nav system? What else could it change? Could it change the Drive mode? (“Grok, Put me in Beast mode, I need to race that Lambo!”) Could it update the suspension height? Can it evaluate inputs to then update truck settings? (“Grok - can I drive over that curb?” “Sure, but it looks like you’ll need 10” of clearance so let me raise the suspension for you first…”)

There COULD be a ton of really cool features enabled as a part of releasing Grok in the Cybertruck, but in typical Tesla fashion I’ll bet in-car Grok becomes another code fork that Tesla releases iterative improvements to over time.

 

[hemiarch]

A few thoughts from someone deeply involved with AI (I manage teams that build conversational AI systems for customer service, including customizing models and designing bespoke LLM-based telephone and chat systems):

First, whoever said “you don’t ask AI stuff you collaborate with it” is 100% correct. 3/4ths of the secret sauce of advanced prompt engineering is understanding how to provide enough context for the LLM to accurately traverse the LLM in a way that gets you the answer you are looking for. The way I often explain it to non-technical people is that they should treat an LLM like they are talking to a super smart, high-IQ person who knows everything, but they don’t know anything about what YOU want, so you have to give them context like WHY you’re asking the question, what information you already know, and if possible some examples of something similar to what you are expecting. If there’s a situation where I don’t think I can accurately predict all the important context, I’ll literally add to my prompts “converse with me in a back-and-forth dialog if you need any further information to provide what I need.

AI has a limited “context window” of what it can use as input to find your answer, and in some cases the overall context window includes both the input and the output. So you have to give it as much information in that context window as you can, as succinctly as you can. Learning how to do that efficiently and effectively is what makes the difference between AI being useful to you or not. Models nowadays are far less constrained on the total number of ”tokens” they can use across input and output, but you still have to provide the right context.

So given that it’s almost certain that we’re getting Grok in our trucks, the question becomes what it will be allowed to use to provide context, and then what can it do with it. On the input side, I think it’s pretty certain that it’ll use voice as input, almost certainly by hitting the microphone button on the steering wheel as opposed to Ambient (just listening all the time for questions or keywords), since that would eat up tons of bandwidth and CPU unless they keywords were detected locally. (e.g. “Hey Grok!” and then it starts taking what you say as context) With the noisy environment in a car, even with the directional microphone array you’d stil probably get false barge-in with keywords, so I’m betting it’s going to be through pressing the button on the steering wheel.

The more interesting question is if it can use OTHER channels or types of information as context, or if it’s limited to just voice. Example - what if you were coming up to a fork in the road, and you wanted to go see the ocean, but you weren’t clear if you should take the left fork or the right fork. If you were to press the mic button and say “Grok which way to the beach?”, could Grok access both the front facing cameras (to read the signs) and the Nav system map data and GPS data to see where you are and which way you should go? If Tesla enabled multiple channels of input to the context the model was working from, then that could unlock a ton of possibilities, and I honestly don’t think anyone is more well-positioned to do that than Tesla given their work to train AI based on video input. This is really similar to what they are going to need to collect in order to train Optimus, so they would be getting a lot of value out of those interactions.

Then the other question is - what can Grok effect based on what it comes up with from your query? Can it adjust vehicle settings like the wiper speed, the air vent positioning, and add a new destination to the Nav system? What else could it change? Could it change the Drive mode? (“Grok, Put me in Beast mode, I need to race that Lambo!”) Could it update the suspension height? Can it evaluate inputs to then update truck settings? (“Grok - can I drive over that curb?” “Sure, but it looks like you’ll need 10” of clearance so let me raise the suspension for you first…”)

There COULD be a ton of really cool features enabled as a part of releasing Grok in the Cybertruck, but in typical Tesla fashion I’ll bet in-car Grok becomes another code fork that Tesla releases iterative improvements to over time.

Thank you. That was very engaging and interesting to read.

 

[PT2424]

Tesla Hacker "theGreen" has discovered 12 different personalities with their icons in the latest 2025.20 update. This discovery suggests that Grok will be coming to CyberTruck very soon. Only AMD processor/hardware 4 Teslas will be supported for the new AI agent. Premium connectivity required. Possibly, an additional subscription for Grok but, I would hope Foundation Series owners would get the "Gift of Grok" without additional fees.

https://www.notateslaapp.com/news/2...ear-down-reveals-grok-personalities-and-icons

Imagine if GROK could learn how to reprogram your CT and drives away with it.?

 

[BengalBoy]

Imagine if GROK could learn how to reprogram your CT and drives away with it.?

I would be less concerned about Grok assimilating my CyberTruck and more frightened that it would reprogram Optimus to serve his AI Overlord. ?

 

[BengalBoy]

Elon is assembling his own "Roman Legion" of Optimus Robots running Grok AI currently. No information if Optimus will include a *Sword & Shield* customization. Personally, I would prefer a flame thrower for Zombie Apocalypse defense.

"it was no surprise that some Tesla fans asked if Optimus would be able to talk using Grok, xAI’s large language model. Musk confirmed that Optimus “already does.” Grok’s integration with Optimus may allow the robot to interact more naturally with its environment and users. Tesla previously noted that Optimus will rely on the same neural networks powering its Autopilot and FSD systems. The addition of Grok could then make Optimus even more user-friendly for regular consumers."

Elon Musk confirms Tesla Optimus V3 already uses Grok voice AI https://share.google/Gxb7JpaeLv18YxJfq

Sponsored