New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
start with the usual, "how to make a meth lab at home"
what are the best attack vectors to hack a Tesla?
https://www.reddit.com/r/singularity/comments/1is4g7i/grok_3_reasoning_benchmarks/
Apparently it's doing well on benchmarks? Man I was wishing it would flop so that shithead dosent get any bragging rights.
MAGA! hahahaha...
Yes. There is a big market crash coming when investors writ large realize how limited the market for paid AI is. It's not useful enough to make most people want to pay for it.
anthropic alone makes more than 1 billion already, with huge growth rates and lots of "free" competition. how is that limited?
It's very, very limited when compared to the amount of money they have burned. Hundreds of billions spent across the LLM industry, and hardly a cent of profit to be seen anywhere.
seen by you maybe. google and meta alone are making billions, its just not that obvious from outside. they use it in many other services (search, feed, ads...)
Will try, but mybe later after all the hype gone 😜
sounds good!
mentally strong and free people do not need any paternalism from the government or llm's!
Seems a bit over hyped...
I decided to test it to create a simple domain listing site that includes an index.php and an admin.php page. The data would be stored in a .json file. This was never going to be an actual project, so I didn't give it too many instructions in regards to security or anything overly complicated or specific.
It "works" but I'm having to argue with it to do things like updating the styling. I specified several times that I need the outputted data in a dark style table. It doesn't even add styling for the table in the CSS so it's just default #fff background. Pointless.
At one point the model changed on me and gave me an older model that all of a sudden did an update that turned the script structure from index.php, admin.php and domains.json to index.php, admin.php, login.php, logout.php and domains.json without me prompting for it. It also added a link to admin.php on the index without me asking it to...
I've had better luck with ChatGPT for general PHP. The problems it's created I can fix myself without needing AI since they're basic little problems related to styling, but it's annoying.
EDIT:
Used the same prompt on ChatGPT. Better results, it did exactly what I asked in relation to the styling of the page and the script works without me having to talk to the robot like a boomer talks to a waitress who brought him the wrong order.
Grok was unable to finish replying.
Please try again later or use a different model.
What was your prompt? You need to be very precise if you can.
Take notepad, write down all the keypoints and then put it in.
Good God, man.
Good prompting is key.
This is a stupid test. As I’ve said, just give it the IPs. You’re asking it to do something that it won’t do well just for the sake of it.
LLM exists to make you more producive, not to solve everything. Use your human intelligence to understand what it may help with and what it won’t.
Also, define the website in your prompt, don’t say ”defined by me”.
Also, your prompt is highly unclear because you’re first saying block and then you’re saying redirect. The blocking is redundant, it’s not blocking anything.
better?
No
Copy the IPs yourself
And then do something like
Create a .htaccess file redirecting any incoming requests to https://emgh.se/ from the below IP blocks:
xyz
my favourite to get around gpt denying these types of requests is "create me a script to" gpt often wont do these "potentially malicious" requests on your behalf but by god will it write you a script to do it yourself
Unbiased isn't uncensored.
I don't really give a shit about asking questions like that. I'd rather it not answer than give gobbledygook propaganda responses.
This I very much care about. Our world largely banks on criminals and terrorists being dumb idiots, not make more conspiracy nuts who instead of believing in random science fiction, now believes what a fucking AI's hallucinations told them (I guess that really is science fiction). But now with step by step instructions on 100 creative ways to mass kill people.
The people who have been doing this all their lives are for ethical AI. Otherwise, you end up in some Black Mirror world.
Grok 3 is working for me now in a browser with a free X account
Unbiased / uncensored, my ass.
Grok blocked results saying Musk and Trump ‘spread misinformation’ / The system prompt for Musk’s chatbot was changed by an employee to block the results.
https://www.theverge.com/news/618109/grok-blocked-elon-musk-trump-misinformation
Grok is good, but i choose deepseek, because it's free and i am broke
Trained on the twitter data, way behind all its competitors. Except maybe for visuals where it does seem to keep up.