Musk was in OpenAI. He left/sold his part. He created Grok instead, which is just another LLM chatbot, like ChatGPT, Gemini, Bard, DeepSeek, whatever.

ChatGPT consist of 2 parts, the LLM, called GPT, and the Chat, which is kind of a frontend to allow people to interact with GPT. However Chat has a bunch of constraints, filters, built-in "wokism". Supposedly Grok doesn't have that. Now there's some pseudo-controversy about some made up bs isn't worth to follow.

 

Bwt OpenAI offers an API to their various GPT models, I think querying their LLMs like that avoids those filters. One can have conversations with them with cURL (have to pay tho).

That's right, but it was already "cancelled." If you ask Grok 3, he favors the hypothesis that it was a mistake or even sabotage by some progressive employee. (It's worth noting that this was with version 4, i.e., Grok 4.)

 

Maybe, but I'm still struggling to convince myself that it's just the chatbot version. Dall-e is also very limited in that regard; I'd bet the entire model is conditioned. GPT is also used in Copilot.

 

From what I've been reading, the chatbot is given coverage (filters or layers), but the model itself is also refined (fine-tuned).

 

Have you already tried APIs in any external app or chatbot? (Excluding Copilot.)

 

https://www.ibm.com/think/topics/fine-tuning

So it’s just a conservative AI? Honestly I feel like no one would care if it wasn’t Elon muskrat promoting it. No one talks about https://www.4conservative.com/ because no one uses it. Politics aside this is just another AI.

No, that’s the image the woke press wants to push. But the reality is that Elon’s company (xAI) has managed to achieve the main models (GPT, DeepSeek - the Chinese one - and Gemini - Google’s woke AI -) in a very short time. And the most interesting thing is that xAI is smaller than OpenAI (in other words, it’s a boom similar to what DeepSeek achieved back in the day). A smaller-scale company achieving the same results in terms of performance.

 

Now, beyond the technical aspect, from the perspective of the culture war: This is good because it brings balance to the market. Until now, the only functional and commercial models were either aligned with woke ideology (Gemini), neutral (GPT), or of Chinese origin (DeepSeek), among others with similar leanings. In other words, no company was aligned with “conservatism” (though Grok is more technolibertarian than conservative). This created an ideological bias in the market’s products. Now, with xAI’s Grok, that bias is counterbalanced and leveled out.

 

The example you mention (4conservative) is, in a sense, outdated. Not because of that specific product, but because we know that in this race for AI development, only disruptive companies with high infrastructure will survive (not even Apple or Meta are managing it, and in just two years, xAI has). The emergence of Elon’s Grok is no small feat: it’s proof that woke ideology and its agenda don’t have a guaranteed hegemony over a technology with immense commercial and strategic value.

 

 

Source:

 

https://medium.com/data-science-in-your-pocket/grok-4-benchmarks-explained-55572135449c

 

https://www.datacamp.com/blog/grok-4

 

https://www.scientificamerican.com/article/elon-musks-new-grok-4-takes-on-humanitys-last-exam-as-the-ai-race-heats-up/

 

https://docs.x.ai/docs/overview

So it’s a non bias AI Elon made, checks out why are the libtards are jumping up and down.

 

 

Unless they embedded it in a place I don’t know about then never.

 

300$ and I will have to go out of my way to use something that is not worth my time.

I probably start using in different sites (starting with X, of course). It is difficult to know why companies use APIs to access models without warning users (Copilot's is because it is Microsoft that is a large company).

 

In general, users who don't work with AI don't usually pay for it (also, many jobs don't require paid AI, depending on the power level).

 

In support, I used conventional ChatGPT and Copilot. Some industrial engineers at the company were using conventional ChatGPT. The $300 fee must be for enterprise-level users or users who need AI on a constant basis.

 

No, it's not "impartial" (it shouldn't be reduced to that simplistic approach), but it's actually necessary. If Thiel and Jobs are right in their theorization, it's only a matter of time before one company monopolizes AI and the rest orbit around it (as Google did with its search engine, LinkedIn with its professional network, and Amazon with e-commerce, etc.).

 

The best thing is for xIA to take off in the race or for OpenIA to ultimately prevail.

 

https://en.wikipedia.org/wiki/Big_Tech#

 

Also:

In any case, the tests speak for themselves, and they've shown a superiority. We'll have to see what the rest of the field throws and who ends up taking the lead this year.

OpenAI offers some free queries for a limited amount of time after free sign up. But have to add credit card details even for that. So I have not tried it, and for now not planning it.

I do use various APIs but not for chatbots.

 

I dunno, as far as I know they just don't add leftlib bias. The lack of that makes it biased conservatively? I don't think so. But I don't know if they add other bias filters or not.

Like an imagboard, it's easy just post "nigger" all the time, because it doesn't get moderated out. Does this make imageboards conservative?

I do use various APIs but not for chatbots.

I'm unemployed now, but I was keen to buy a graphics card and try out local AI. I was thinking about something moderate, an RTX 3060/4060

 

Or a dedicated local AI rig. (Like pic related)

Not just graphics card, each content generation (be it textual, or audiovisual) consumes horrible amount of electricity. Paying bills is no joke.

It's worth considering how much a device like that consumes. If it consumed the same amount of power as having the device play for 6 or 8 hours, it wouldn't be such a problem (that's what happened before, and it wasn't that expensive, between 400W and 800W). It's assumed that it wouldn't be used continuously.

I looked at prices, and bringing one to my country costs $1,400 (36% of that is taxes).

A graphics card like an RTX 4080 costs roughly the same. But even if I bought a card like that, I'd have to modify the rest of my PC (and I'd still lose VRAM).

 

 

I think the idea would be to run a 70b model. On my current one, I tried a 7b Deepseek, and besides having the speed issue, it was pretty amazing.