I’ve been talking with Claude

[velociraptor]

I'll be honest, I've a really hard time believing this one. Both Musk and the entirety of Twitter lean VERY heavily in one particular direction.

Uh, yeah.

Musk's AI Company Apologizes Over Grok's Antisemitic Posts

Musk seems to have wrong ways and Reich ways of doing things.

 

[saturnv]

@Outdated Have you tried the search engine kagi? It can search by country and supports exact word or phrase searches. It does not do advertising so it costs money to use it.

It has AI junk, but it is all fully optional.

 

[Matthias]

I'll be honest, I've a really hard time believing this one. Both Musk and the entirety of Twitter lean VERY heavily in one particular direction.

Musk stated he leans Left politically but is often falsely accused of being on the Right due to the Left shifting so far to the Left during the past decade that the moderate Left now seems like they are on the Right. It seems to be due to a small group of activists whose bigotry and hate leads them to constantly attack anyone who doesn't support their extremist views.

As an independent thinker who is in the Center politically, I'm able to see what's happening very clearly and it's shocking. I respect everyone, regardless of how they identify or what they believe. I hope the hate we've seen during the last several years comes to an end soon. I look forward to a return to civility, tolerance, and acceptance.

Of all of the available AIs, Grok is the one I'd personally trust the least. Not that I would trust anything that comes out of that site anyway.

Of course... it's also the very specific AI I refuse to ever touch (you could not pay me to go near that site or any other social media site as that goes against my own personal rules), so that's an idea I aint gonna test myself. Not that there's any reason to, I've got access to enough different AIs as it is.

There is definitely a variety to choose from. ChatGPT leans Left, Gemini is further Left, and Claude is extremely woke, embracing all the latest fads while demonizing and chastising those with traditional views embraced by the majority.

 

[velociraptor]

As an independent thinker who is in the Center politically

I'm also politically centrist, or at least what passes for it here in Canada. No centrists I know of do Nazi salutes and spend hundreds of millions putting a racist, right wing demagogue into power. I assume centrist means something entirely different where you live.

 

[WhitewaterWoman]

There is definitely a variety to choose from. ChatGPT leans Left, Gemini is further Left, and Claude is extremely woke, embracing all the latest fads while demonizing and chastising those with traditional views embraced by the majority.

I don’t understand this. Nothing I ask ChatGPT or Claude has anything to do with politics. I’ve never been chastised for anything I’ve asked. That’s one of the nice things about them. I can ask questions when I don’t understand something, many questions, and I’m not chastised.

Claude provided links to each piece of material so any leanings would become obvious from the sources.

Actually I used to be mellow with it because like I trusted it too much and happily believed everything it told me but it still turned out it told me all lies and now I feel so gullible for believing them.

I don’t understand this either. For one thing, it can’t be all lies. Second, it does make mistakes. Mistakes are not lies. Whether or not AI can deliberately lie is a question I don’t think has been resolved.

It’s okay not to like AI or any of these platforms. But it seems like some people are anthromorphizing more than the « therapists » and « companions ».

I’m grateful for all the responses. I’ve learned a lot.

 

[Matthias]

I'm also politically centrist, or at least what passes for it here in Canada. No centrists I know of do Nazi salutes and spend hundreds of millions putting a racist, right wing demagogue into power. I assume centrist means something entirely different where you live.

I think you've been misled by propaganda. A small group of extremists frequently lie about politicians on the Right being Nazis or fascists but I haven't found any evidence to support their claims.

Do you have any evidence of Elon Musk doing a Nazi salute?
Do you have any evidence that Elon Musk supported a racist politician or helped a racist demagogue get elected?

 

[WhitewaterWoman]

I think you've been misled by propaganda. A small group of political extremists frequently accuse their opponents on the Right of being Nazis or fascists but I haven't found any evidence to support their claims.

Do you have any evidence of Elon Musk doing a Nazi salute?
Do you have any evidence that Elon Musk supported a racist politician or helped a racist demagogue get elected?

I’m pretty sure this is venturing into politics. Let’s keep it on AI, Ok?

 

[velociraptor]

I think you've been misled by propaganda.

Uh, no. I read the news. You should, too. And you should avoid his AI if quality in results is important to you.

 

[Matthias]

I don’t understand this. Nothing I ask ChatGPT or Claude has anything to do with politics. I’ve never been chastised for anything I’ve asked. That’s one of the nice things about them. I can ask questions when I don’t understand something, many questions, and I’m not chastised.

Claude provided links to each piece of material so any leanings would become obvious from the sources.

You generally won't notice a bias unless you ask questions about controversial topics (such as climate change, gender identity, sexual preferences, infanticide, illegal immigration, etc.) . My experience and numerous studies have shown that the answers almost always lean Left. Right-wing views on these topics are often disparaged.

 

[Matthias]

Uh, no. I read the news. You should, too. And you should avoid his AI if quality in results is important to you.

I've found Grok's results to be very high quality recently, although I felt Grok was behind the others when I compared them in the past. Benchmarks showing Grok behind the other 3 but catching up recently confirm my experience with the models.

 

[Misery]

I’m pretty sure this is venturing into politics. Let’s keep it on AI, Ok?

It's a pointless argument in any case.

Been down this road before more than once.

That being said: understanding what Grok is, is something that IS important to this.

AI is useful, but always remember: it does not exist in a vacuum. That is something important to keep in mind.

I dont mean just in terms of political whatsits. Also keep in mind that sometimes you can get an update that borks stuff. Really rather frustrating, as the user has no control over this. As much as I appreciate ChatGPT, every now and then it goes all buggy after an update. Important to just like... give it a few days, I think. When that's happening.

You can also get other effects. There were a lot of people who were using ChatGPT as a sort of companion, so to speak, and then rather recently, a major update removed the... o4 model, I think it was? It caused a bit of an uproar as those using it in that way lost the specific model/mode/variant/something that made that function easy.

So that's something to keep in mind. You'll get new features at random points, but already existing features are not guaranteed to remain.

Now if we get REALLY in depth though, we'll start to get into AI that can be customized in various ways or even just downloaded. This gives you a lot of freedom to set it up how you want it, though of course you need the knowledge to do so.

I actually downloaded a bunch of stuff like that recently to experiment with very recently.

Experiments are forthcoming (when I'm feeling less lazy). If anything explodes, I'll be sure to give everyone the hilarious details.

 

[Matthias]

Grok Outperforms Competitors with Lowest Hallucination Rate in New AI Reliability Study​

Here's the level of inaccurate results the study found for different models:
Grok - 8%
DeepSeek - 15%
ChatGPT - 35%
Gemini - 38%

This study was released in December 2025. I'm not sure if I can post a link but you can easily find it by searching for the title above.

 

[Matthias]

Uh, no. I read the news. You should, too. And you should avoid his AI if quality in results is important to you.

I encourage you to rely on reputable news sites and to confirm what you read.

 

[WhitewaterWoman]

Grok Outperforms Competitors with Lowest Hallucination Rate in New AI Reliability Study​

Here's the level of inaccurate results the study found for different models:
Grok - 8%
DeepSeek - 15%
ChatGPT - 35%
Gemini - 38%

This study was released in December 2025. I'm not sure if I can post a link but you can easily find it by searching for the title above.

This is one study published on X that used an Ai game theory model. Yes, I googled it.

Hardly determinant.

And inaccurate results in what field?

Plus my understanding is that these platforms are being updated all the time, which means this kind of study would be outdated as soon as it was released.

 

[Matthias]

This is one study published on X that used an Ai game theory model. Yes, I googled it.

Hardly determinant.

And inaccurate results in what field?

Plus my understanding is that these platforms are being updated all the time, which means this kind of study would be outdated as soon as it was released.

My experience confirms the results are accurate.

 

[Chingching]

I don’t understand this either. For one thing, it can’t be all lies. Second, it does make mistakes. Mistakes are not lies. Whether or not AI can deliberately lie is a question I don’t think has been resolved.

Like 90% of what it told me was lies
Are you on Discord? I can tell you there what the lies were that I can't say here because of privacy (but not inappropriate or anything like that) and I don't trust the private messaging system on any forum

 

[Misery]

Like 90% of what it told me was lies
Are you on Discord? I can tell you there what the lies were that I can't say here because of privacy (but not inappropriate or anything like that) and I don't trust the private messaging system on any forum

These "lies" are one thing, but the prompts also need to be taken into account.

You dont get one without the other.

I've seen this sort of thing before, and without going into detail, there was a very heavy amount of user error.

I mean, it could give you the correct response to a prompt, but if that prompt itself was bad, well... you can see where that goes. A correct answer to the wrong question ends up still being of no use.

But also, you gotta consider what it does and does not have access to in terms of info.

One way or another, an AI will always answer you, that's how they work. They ALWAYS answer. But if you're asking it a question about something it has no data on, or just not much data on, you're going to get some weird stuff in return. Like my gaming example from earlier. If I ask it about Minecraft I'm probably getting correct info in return (provided, that is, I chose the correct AI for the job). If I ask it about some indie game I found under a rock, it's gonna give me just a bunch of random nonsense. How could it not?

These things arent magic, nor are they phone apps that give you victories because you swiped down.

If you want less "lies" and more reliable use, the key is to learn more. Just like with any computer tech and whatnot, you get what you put into them.

 

[Jonn]

I've never tried any of them. I hate live online chat with humans, I can't imagine I'd like chatting to a machine any better.

AI doesn't judge you.

Humans often do, but I don't have a problem with that since human psychology is a special interest, and the more damaged the person, the greater the interest.

Having said that, I am using AI more and more, particularly when checking my facts.
AI isn't always helpful/correct, especially when I see it reference Wikipedia, lol, but I do find it very useful overall.
I even using to help me with my share transaction decisions.

 

[Jonn]

When something like a comedy sketch is titled with click-bait words like “banned advert,” the system often interprets it literally because it has far more examples of real advertising controversies than niche Australian satire.

Are you saying AI is autistic?

 

[Jonn]

I use ChatGPT but not for serious discussion and if it comes up with any bull then I swear at it.

I have told AI off many times for talking crap.
The trouble is, it doesn't get upset despite my best efforts. lol

This is great input. I have dabbled a bit. l asked for help on buying stocks, looking for stock bubbles. My roomie asked about who killed JFK.

I have been using AI when researching stocks, too.

People want to "do" without the "learn" though. With that approach, it's no wonder everyone keeps running into screwy problems with it.

You have inspired me to learn about AI.
YouTube, here I come.