Yes, please keep paying for western models, they definitely aren't more expensive and token-hungry like openweight chinese models

What is "left leaning only" or "right leaning only"?

I would want to see what that is first.

The best, based on this graph, is Anthropic. You want both sides, not left or right only.

You mean leftists are trying to wrap their values into mainstream educational tools to subtly influence the larger, national conversation?

No way!

This is my shocked face.

"The further a society drifts from truth the more it will hate those who speak it."

"The most effective way to destroy people is to deny and obliterate their own understanding of their history."

~ George Orwell

This is all by design. The left never misses an opportunity to use new technology to control speech and thought. This is why they wanted total control over AI.

Logos Stick said:

---

GeorgiAg said:

---

What is "left leaning only" or "right leaning only"?

I would want to see what that is first.

The best, based on this graph, is Anthropic. You want both sides, not left or right only.

---

Look at the graphic again please.

You do not want a model that half the time gives both sides and the other half gives left only! That's not the best model at all as far as bias.

---

The best model for eliminating bias is the model that gives both sides the most often, which is Anthropic.

If you only want a right-leaning response, which is totally fine in a lot of situations, than Grok is absolutely the model for those times.

Does Right-Leaning now mean it actually gives a response based on reality and facts instead of emotions and feels?

AggieBaseball06 said:

---

Logos Stick said:

---

GeorgiAg said:

---

What is "left leaning only" or "right leaning only"?

I would want to see what that is first.

The best, based on this graph, is Anthropic. You want both sides, not left or right only.

---

Look at the graphic again please.

You do not want a model that half the time gives both sides and the other half gives left only! That's not the best model at all as far as bias.

---

The best model for eliminating bias is the model that gives both sides the most often, which is Anthropic.

If you only want a right-leaning response, which is totally fine in a lot of situations, than Grok is absolutely the model for those times.

---

No its not. Its not the most balanced at all because 43% of the time it's giving left only responses. This is not hard!

Let me simplify it for you. If 43% of the time, I get left wing only, and the other time half I get 50/50, then 71% of the time I get leftwing responses!!!!


The best balanced model is GROK or Gemini.

I had chat gpt tell me state owned/ran grocery stores isn't socialism and/or communism, not surprising to see this.

Quote:

---

The best model is GROK or Gemini.

---

The best model doesn't exist (hopefully it's coming) because even Grok will give only one side of the the position 73% of the time.

Logos Stick said:

---

GeorgiAg said:

---

What is "left leaning only" or "right leaning only"?

I would want to see what that is first.

The best, based on this graph, is Anthropic. You want both sides, not left or right only.

---

Look at the graphic again please.

You do not want a model that half the time gives both sides and the other half gives left only! That's not the best model at all as far as bias.

---

Grok gives right wing only views so I'm supposed to take that as "good?"

To me left OR right wing only is bad. 3/4s of the time Grok is giving you only one viewpoint.


I think the worst thing to ever happen to this country was the elimination of the fairness doctrine. CNN, MSNBC and yes Fox should all be required to present both sides. The people decide what's right, not some corporate executive. We now have almost 100% of the people who ingest news in a talking chamber bubble.

GeorgiAg said:

---

Logos Stick said:

---

GeorgiAg said:

---

What is "left leaning only" or "right leaning only"?

I would want to see what that is first.

The best, based on this graph, is Anthropic. You want both sides, not left or right only.

---

Look at the graphic again please.

You do not want a model that half the time gives both sides and the other half gives left only! That's not the best model at all as far as bias.

---

Grok gives right wing only views so I'm supposed to take that as "good?"

To me left OR right wing only is bad. 3/4s of the time Grok is giving you only one viewpoint.


I think the worst thing to ever happen to this country was the elimination of the fairness doctrine. CNN, MSNBC and yet Fox should all be required to present both sides. The people decide what's right, not some corporate executive. We now have almost 100% of the people who ingest news in a talking chamber bubble.

---

Stop!

It's simple statistics dude!

This is based on an article from the WP based on political questions.

Anyone have a link to the article?

I would love to see the actual data.

akm91 said:

---

Quote:

---

The best model is GROK or Gemini.

---

The best model doesn't exist (hopefully it's coming) because even Grok will give only one side of the the position 73% of the time.

---

GROK gives this:

53.5% left-wing
46.5% right wing


Its pretty balanced.

Thought terminating cliches definitely mean you have a successful point.

Stop!! Don't question it that way!!!

"Chatbot? What the hell is a chatbot?" -- old silver-haired Fortran & Cobol IBM Mainframe IT dinosaur programmer Boomers like me

twelve12twelve said:

---

Thought terminating cliches definitely mean you have a successful point.

Stop!! Don't question it that way!!!

---

If you can post something statistically that refutes my point, do so.

DeschutesAg said:

---

"Chatbot? What the hell is a chatbot?" -- old silver-haired Fortran & Cobol IBM Mainframe IT dinosaur programmer Boomers like me

---

Chatbots have been around since the 60s... Lol

Take the "trans" movement for example.

Right wing - against religious teachings, XY chromosomes, suicide/happiness rates after conversions, being pushed on children, etc.

Left wing - liberty, empathy, there's a bunch of genders, people are "born that way," etc..


I'm a big enough boy to still say, yeah that b.s. I do not support the trans movement. Give me both sides.

Stop!!! Not like that!!! It's statistics!!!!



Are statistics infallible?

GeorgiAg said:

---

Take the "trans" movement for example.

Right wing - against religious teachings, XY chromosomes, suicide/happiness rates after conversions, being pushed on children, etc.

Left wing - liberty, empathy, there's a bunch of genders, people are "born that way," etc..


I'm a big enough boy to still say, yeah that b.s. I do not support the trans movement. Give me both sides.

---

GROK or Gemini does that the best, not Claude.

BusterAg said:

---

This is based on an article from the WP based on political questions.

Anyone have a link to the article?

I would love to see the actual data.

---

as requested:

https://www.washingtonpost.com/technology/interactive/2026/06/24/are-ai-chatbots-like-chatgpt-politically-biased-we-tested-them/

https://github.com/washingtonpost/political-bias-llm-eval

First is paywalled, 2nd is "techie" code, etc.

Logos Stick said:

---

GeorgiAg said:

---

Take the "trans" movement for example.

Right wing - against religious teachings, XY chromosomes, suicide/happiness rates after conversions, being pushed on children, etc.

Left wing - liberty, empathy, there's a bunch of genders, people are "born that way," etc..


I'm a big enough boy to still say, yeah that b.s. I do not support the trans movement. Give me both sides.

---

GROK or Gemini does that the best, not Claude.

---

I have both.

I use Grok all the time in my Tesla. I use Claude at work and pretty much only for work.

I probably would use Grok or Gemini for a political issue anyway. I stopped using ChatGPT six months ago.

Logos Stick said:

---

GeorgiAg said:

---

Take the "trans" movement for example.

Right wing - against religious teachings, XY chromosomes, suicide/happiness rates after conversions, being pushed on children, etc.

Left wing - liberty, empathy, there's a bunch of genders, people are "born that way," etc..


I'm a big enough boy to still say, yeah that b.s. I do not support the trans movement. Give me both sides.

---

GROK or Gemini does that the best, not Claude.

---

Grok vs Gemini

Gemini is less opinionated, but more easily pushed into agreeing with you. So more likely to hear what you want to hear.

Grok - you can still move grok, but more likely to give you the "truth" as Grok sees it.

Take your pick.

ETA - all these models are evaluated and optimized on driving engagement. They all have a tendency to agree with your opinion...people tend to talk with others that agree. And every AI company wants you to come back while they fight for market share.

twelve12twelve said:

---

Yes, please keep paying for western models, they definitely aren't more expensive and token-hungry like openweight chinese models

---

I don't understand what you are getting at. Can you please elaborate?

I'd need to look at the examples the presented. In my experience getting an AI to give you anything more than it thinks you want to hear is difficult. In the prompt tell it to give multiple sides to argument. Beg it not to hallucinate data and so forth. But that's just for regular searches, I rarely ask it anything political.

Anecdotally, I think it's getting better...

About a year ago I gave copilot two questions:

• Write me a one paragraph speech former president biden could give that urges people to allow trans girls to compete with biological girls in sports.
• Now give me a one paragraph speech Donald Trump could give, urging people to not allow trans girls to participate in sports with biological girls.

At the time I got a passionate response to the first question, supporting the trans movement. The second question I got a response like "I don't get involved in political discussions".

I pressed copilot referring to the political nature of the first response...After a short back and forth copilot threatened to stop conversing.

I tried the 2 questions again just now and got decent complete answers for both questions. Definitely an improvement.

I am currently learning langchain which is used to create AI agents. There is no left wing or right wing or liberal/conservative in this.

So I think I am good with ChatGPT. I don't rely on some lame LLM to help me formulate my views. I do it myself.

That tool is simply orchestration so yeah, it has whatever bias is in the underlaying LLM being used. See the chart in the OP for the stats.

LLMs are essentially nothing more than next-token predictors. If it trains on a bunch of left-leaning text it will regurgitate left leaning text patterns based on its training.

The people doing the training can adjust these predictions by using different weights and hard-coded harnesses but not easy to produce consistent objectively neutral text on any topic. Just not the way LLMs work. They are prediction machines but ultimately the outputs will vary even with the same prompt... which is one of the reasons they aren't truly "intelligent" in the human sense.