2026 Oracle Red Bull Racing F1 Team

[organic]

They recruited approximately 170 engineers from Mercedes HPP, as well as dozens from Ferrari, Renault, Honda, and even Cosworth, in addition to new hires. It would be unrealistic to assume that Mercedes’ power unit department was unaffected by the loss of that much experience.

On top of that, there is the massive support that Ford can provide with their state-of-the-art facilities. Based on this, we can safely assume that the Red Bull Ford engine will be very close to Mercedes, or even better. Then it will be the aero and the driver that will decide, not the engine. And Mercedes has been weak in that area for the past four years. Red Bull has a better aero department.

Incidentally, the lap times also show this. Hadjar was immediately the fastest at the start of the test, even though he is much slower than Verstappen. After that, Red Bull deliberately slowed down, as they only improved by a few tenths compared to that lap. Sandbagging.

Very chatgpt of you

Can people seriously stop

[AR3-GP]

They recruited approximately 170 engineers from Mercedes HPP, as well as dozens from Ferrari, Renault, Honda, and even Cosworth, in addition to new hires. It would be unrealistic to assume that Mercedes’ power unit department was unaffected by the loss of that much experience.

On top of that, there is the massive support that Ford can provide with their state-of-the-art facilities. Based on this, we can safely assume that the Red Bull Ford engine will be very close to Mercedes, or even better. Then it will be the aero and the driver that will decide, not the engine. And Mercedes has been weak in that area for the past four years. Red Bull has a better aero department.

Incidentally, the lap times also show this. Hadjar was immediately the fastest at the start of the test, even though he is much slower than Verstappen. After that, Red Bull deliberately slowed down, as they only improved by a few tenths compared to that lap. Sandbagging.

Very chatgpt of you

Can people seriously stop

None of this looks like ai

[organic]

They recruited approximately 170 engineers from Mercedes HPP, as well as dozens from Ferrari, Renault, Honda, and even Cosworth, in addition to new hires. It would be unrealistic to assume that Mercedes’ power unit department was unaffected by the loss of that much experience.

On top of that, there is the massive support that Ford can provide with their state-of-the-art facilities. Based on this, we can safely assume that the Red Bull Ford engine will be very close to Mercedes, or even better. Then it will be the aero and the driver that will decide, not the engine. And Mercedes has been weak in that area for the past four years. Red Bull has a better aero department.

Incidentally, the lap times also show this. Hadjar was immediately the fastest at the start of the test, even though he is much slower than Verstappen. After that, Red Bull deliberately slowed down, as they only improved by a few tenths compared to that lap. Sandbagging.

Very chatgpt of you

Can people seriously stop

None of this looks like ai :?

Maybe I'm just more familiar with the type of overconfident logic chains LLMs make when it comes to F1. Things like: "hadjar was immediately fastest... after that red bull deliberately slowed down", and "state-of-the-art-facilities... We can safely assume they will be very close or even better" which state of the art facilities of ford, can we safely assume that? Even assuming ford will be close is a big leap given the level f1 engine manufacturers usually start at.

An individual could have written it, but there are specific structures and phrasing that are extremely common to chatgpt and none of it reads like something that chatgpt would spit out.

Pantherxxx is also prolific in this area. For instance this recent reply chain was laughable https://www.f1technical.net/forum/viewt ... 3#p1315563 - there's a smoking gun

Spend 10 mins generating responses from chatgpt on technical F1 topics and you'll see the angle I'm viewing this from

[tinuva]

Very chatgpt of you

Can people seriously stop

None of this looks like ai

Maybe I'm just more familiar with the type of overconfident logic chains LLMs make when it comes to F1. Things like: "hadjar was immediately fastest... after that red bull deliberately slowed down", and "state-of-the-art-facilities... We can safely assume they will be very close or even better" which state of the art facilities of ford, can we safely assume that? Even assuming ford will be close is a big leap given the level f1 engine manufacturers usually start at.

An individual could have written it, but there are specific structures and phrasing that are extremely common to chatgpt and none of it reads like something that chatgpt would spit out.

Pantherxxx is also prolific in this area. For instance this recent reply chain was laughable https://www.f1technical.net/forum/viewt ... 3#p1315563 - there's a smoking gun

Spend 10 mins generating responses from chatgpt on technical F1 topics and you'll see the angle I'm viewing this from

I tend to agree with this.

For fun, asked Claude for an assessment (on the original post claimed to be from chatgtp):

> Overall assessment: 60-70% likely to be AI-generated

It supplied a bunch of reasoning why, but I think the only part I found funny, is this:

> The biggest tell is the overly neat logical flow and the way it builds to conclusions - very typical of LLM output trying to sound authoritative about F1 technical matters.

[f1isgood]

Using LLMs to detect if something is LLM is a bad idea because at least from a purely theoretical pov it's statistically indistinguishable from real text. That said, I didn't see any ChatGPT in that post. I also don't see what's wrong with using LLMs in general to make your point if you want your phrasing to be better. Not defeding the poster's previous posts but this one didn't read ChatGPT-pilled enough for me.

[tinuva]

Using LLMs to detect if something is LLM is a bad idea because at least from a purely theoretical pov it's statistically indistinguishable from real text. That said, I didn't see any ChatGPT in that post. I also don't see what's wrong with using LLMs in general to make your point if you want your phrasing to be better. Not defeding the poster's previous posts but this one didn't read ChatGPT-pilled enough for me.

So you want to tell me, your untrained eye is better at detecting whether something is generated by an LLM compared to using a model that is trained to generate LLM output, evaluating the same piece of text?

[Emag]

What fuels will RedBull be using this season? Somehow I don't remember seeing this anywhere.

[Badger]

What fuels will RedBull be using this season? Somehow I don't remember seeing this anywhere.

Type of fuel, who knows. Fuel supplier, Exxon.

[lio007]

What fuels will RedBull be using this season? Somehow I don't remember seeing this anywhere.

Type of fuel, who knows. Fuel supplier, Exxon.

I read recently that Shell is the only supplier to use biofuel all others use synthetic fuels.

[venkyhere]

What fuels will RedBull be using this season? Somehow I don't remember seeing this anywhere.

Type of fuel, who knows. Fuel supplier, Exxon.

Mercedes - Petronas
Ferrari - Shell
RBPT - Exxon
Audi - BP
Honda (AM) - Aramco

[f1isgood]

Using LLMs to detect if something is LLM is a bad idea because at least from a purely theoretical pov it's statistically indistinguishable from real text. That said, I didn't see any ChatGPT in that post. I also don't see what's wrong with using LLMs in general to make your point if you want your phrasing to be better. Not defeding the poster's previous posts but this one didn't read ChatGPT-pilled enough for me.

So you want to tell me, your untrained eye is better at detecting whether something is generated by an LLM compared to using a model that is trained to generate LLM output, evaluating the same piece of text?

I am not saying I am better at this task. As you rightly point out, LLMs are trained to generate text, not detect them and that difference is important. It is well-known at this point (if you work adjacent to LLMs in a research capacity) that you cannot differentiate LLM generated text from human generated text at least theoretically.

I can still have some opinions like saying that the poster's original post didn't feel LLM enough to me. That doesn't mean it wasn't run through GPT. All I am saying is that neither my impression nor using models to detect something are strong evidence in any case and it's a pointless exercise to try to take a stab at every post that potentially was run through GPT just for language reasons while missing the discussion points it might raise. Some posts are obvious, but in this case it is not so obvious (imo).

[Vettel165]

Since 2018, the winner of the Abu Dhabi GP wins the championship the following season...

[AR3-GP]

M. Sainsbury former Product validation engineer at Mercedes High Performance Powertrains has joined Red Bull Powertrains as PU Data Analytics engineer in Jan.26

Interesting...Safe to say there are no secrets between HPP and RPBT.

[tinuva]

Using LLMs to detect if something is LLM is a bad idea because at least from a purely theoretical pov it's statistically indistinguishable from real text. That said, I didn't see any ChatGPT in that post. I also don't see what's wrong with using LLMs in general to make your point if you want your phrasing to be better. Not defeding the poster's previous posts but this one didn't read ChatGPT-pilled enough for me.

So you want to tell me, your untrained eye is better at detecting whether something is generated by an LLM compared to using a model that is trained to generate LLM output, evaluating the same piece of text?

I am not saying I am better at this task. As you rightly point out, LLMs are trained to generate text, not detect them and that difference is important. It is well-known at this point (if you work adjacent to LLMs in a research capacity) that you cannot differentiate LLM generated text from human generated text at least theoretically.

I can still have some opinions like saying that the poster's original post didn't feel LLM enough to me. That doesn't mean it wasn't run through GPT. All I am saying is that neither my impression nor using models to detect something are strong evidence in any case and it's a pointless exercise to try to take a stab at every post that potentially was run through GPT just for language reasons while missing the discussion points it might raise. Some posts are obvious, but in this case it is not so obvious (imo).

LLMs are also great for evaluations. In this is example I only copied over the 60-70% answer, but the bullet points I left out, pointed out real examples of why that piece of text could very well have been LLM generated. I like the answer it gave because it echoed how I saw the text, literally not 0% nor 100% chance, but somewhere inbetween, which ironically is somewhere along the same lines of what you are saying. Anyways, probably a pointless exercise at this point. The poster has not come back yet online to defend his post.

[langedweil]

Since 2018, the winner of the Abu Dhabi GP wins the championship the following season...

https://postimages.org/

Very well, season over then