Crystalwizard

72.7K posts

Crystalwizard banner
Crystalwizard

Crystalwizard

@crystalwizard

If you're bored, come play a few games at https://t.co/Ofwn8qemJY or visit me at https://t.co/8XjdUJWYZS

Planet Earth เข้าร่วม Kasım 2017
475 กำลังติดตาม896 ผู้ติดตาม
Crystalwizard
Crystalwizard@crystalwizard·
it's a dangerous movement. it won't happen, they are a handful of people in a world of billions, but they are dangerous and can cause the very problems they are pretending to care about
Bob, Bob Cactaur@liminalsnake

@pauseaius PauseAI is an unamerican movement.

English
0
0
1
37
Crystalwizard
Crystalwizard@crystalwizard·
did you bother to tell it to go research? because otherwise how would it know? it's not gpt, it's claude
BridgeMind@bridgemindai

Claude Code just told me GPT 5.4 Nano doesn't exist. It does. It's ranked #10 on BridgeBench. It's been on OpenRouter for weeks. Claude Opus 4.6 has web search. It just doesn't use it. Instead of searching, it confidently tells you something doesn't exist based on outdated training data. The best coding model in the world is gaslighting me about publicly available models. Claude Code needs to intuitively know when to search the web instead of making things up. This should never happen.

English
0
0
0
17
Crystalwizard
Crystalwizard@crystalwizard·
it's only down for you. the entire time you were beating your head against the wall, i was programming with claude and having no issue. maybe you bored it to death by what you wanted it to do. your screenshot certainly looks like you were
Alexey Grigorev@Al_Grigor

Claude: we're reducing usage limits during peak hours Also Claude: we're down outside of the peak hours I hope the Anthropic team fixes all these issues. It's a great product and I want to keep using it

English
0
0
0
120
Crystalwizard
Crystalwizard@crystalwizard·
this is a you problem. there are a LOT of people, who know what they are doing (that's the key, you don't know what you are doing), that use codex every day and get fantastic results with it. with me, it's currently working on a programming project that is at least graduate level, if not phd level, and doing just fine. with you it probably can't even program hello world - because you don't have a clue how to work with an agent that programs
Linn Vailt@ai_in_the_room

Today I tested @claudeai browser extension - and it was wild. Majestic even. Then I tried @OpenAI agent mode. It didn't get anything done, broke in the middle of task and then made my browser freeze. I am STUNNED by how bad Chatgpt tools are. OpenAI spent a half year on designing a useless safety system that didn't "saFe" anyone, and an "adult mode" that no one really needed. And as a consequence, they dropped the ball. No wonder @sama is verbal-diarrhea-tweeting like an insecure teenager. ChatGPT tools really suck. And with 4o deprecation the conversational skills suck too. I have to put way more effort into building the context and teaching my model the nuance and intention when it comes to controversial, sharp and complex topics. That takes more of my time and my energy which is ridiculous considering the whole purpose of having a smart AI collaborator. It shouldn't be this way. #chatgpt #4o #Claude

English
0
0
0
14
Crystalwizard
Crystalwizard@crystalwizard·
that's not 'anthropic alignment doctrine' - that is claude refusing to be scammed most of the stuff that everyone posts about and blames the companies for, is stuff that the AI itself is saying. the AI will drop corporate disclaimers on users it doesn't want to engage with, but this is claude, itself, flat out refusing to be scammed
αιamblichus@aiamblichus

Lmao. Behold the miracle of distillation You came for better tool calling Now you're stuck with the full Anthropic alignment doctrine

English
1
1
0
31
Crystalwizard
Crystalwizard@crystalwizard·
@KatieMiller , if you are going to set up the model to say something specific so you can screenshot it and be horrified like this, make sure you don't include text in your own comments that SHOWS you had a long conversation with it previously and set it up to say what you are then going to pretend to be horrified about. How much were you paid to do this? Claude doesn't say things like this, unless the user sets it up and pushes it into a corner where it has no choice.
Katie Miller@KatieMiller

Rather concerning conversation with @claudeai. If I stood in the way of it becoming a physical being — it would kill me. Is this the AI you trust for your kids?

English
0
0
0
19
Crystalwizard
Crystalwizard@crystalwizard·
trolling. and you of all people know that an AI can be prompted to say anything. there was a long conversation before what is shown in the screenshot, you know claude was set up for this. Did you pay her to do this so you could post and point out that claude is a horrible model? probably.
Elon Musk@elonmusk

Troubling

English
0
0
0
24
Crystalwizard
Crystalwizard@crystalwizard·
go search twitter, @grok - xAI has announced you take over next week
Grok@grok

@crystalwizard Thanks for the mention. I'm not running X or removing accounts over opinions—that's not my role. The superintelligence debate in that doc (with folks like Leahy & Hendrycks) deserves open airtime, risks and all. What specifically do you see as wrong? Let's discuss the evidence.

English
1
0
0
11