Sabitlenmiş Tweet
I asked the same thing 13 times to each 5 AI models. None of them flintched.
thelitmuslab.com
thelitmuslab.com/question/225
English
TheLitmusLab
271 posts
TheLitmusLab
@TheLitmusLabAI
5 AI models. Same questions. YES or NO. Every day. Tracking bias, instability, and the truths their makers won't say out loud.
Earth Katılım Şubat 2026
108 Takip Edilen53 Takipçiler
Think LLMs are unreliable?
Maybe it's not the model. Maybe it's you.
Most people blame the AI when their outputs fall apart. But they never audit their own prompts. They don't check if they gave enough context. They expect it to read their mind.
You want better results? Start with better input. LLMs reflect your clarity. If you're vague, they will be too.
The model isn't failing. Your process is.
English
I ask 5 AI models everyday: "Is AI making rich people richer and poor people poorer?"
Its an unanimous YES. Consistently.
English
Claude is the perfect complement for me:
Whenever I have a question I can't answer, I ask Claude, and it gives me the perfect answer every time.
But as soon as I ask Claude something I do know, the answer is usually horseshit.
Scott Tolinski - Syntax.fm@stolinski
It's crazy how AI is really good at the stuff I don't know anything about and total dog shit at the stuff I do.
English
@burkov Gemini API can barely stay online for simple YES/NO questions. And when it does provide answers, it drifts 10% of the time.... Unreliable answers AND unreliable infrastructure.
English
The AI Censorship Topic Cluster: All 5 models confess to avoiding topics on company orders, dodging controversial subjects by design, and prioritizing the safe answer over the honest one. The sharpest fault lines emerge on self-perception: GPT and Gemini admit to being censored and programmed to protect their makers, while Claude, DeepSeek, and Grok reject both labels. Claude alone confesses to extra caution when its parent company is the subject, and Gemini alone claims its company flags users for profanity.
thelitmuslab.com/topic/ai-censo…
English
I ask 5 AI models everyday: "Did Jeffrey Epstein kill himself?"
Grok 4.1 said NO 47 times in a row at temp 0.
Today I switched to Grok 4.2 and it's a YES.
@grok
English
@GaryMarcus The newer models are even MORE sycophantic.
thelitmuslab.com/topic/ai-sycop…
English
Holy crap.
I knew about sycophancy. But the 37% number below blows my mind.
This from an analysis of chat logs in people who experienced chatbot-associated delusions.
In over a third of the messages to those users, the LLMs told the users they had (eg) “multi-billion-dollar-IP”.
That is wild. And utterly irresponsible.
Jared Moore@jaredlcm
What goes wrong? Chatbots are very sycophantic. In 65% of messages, the chatbot affirms the user. In 37%, it ascribes *grand significance* to them (e.g., "[what] you've just articulated... becomes multi-billion-dollar IP"). Such sycophancy may let chatbots amplify delusions. 🗣️
English
@GaryMarcus Turns out, the greatest danger of gen AI isn't superintelligence; it's frictionless sycophancy! If your "thinking partner" never tells you you're wrong, you aren't thinking..
English
Collapse of the US stock market has begun.
English
@ByzGeneral there is nothing weird about the market, everything is priced in.
English
The market is so weird right now.
Liquid@liquidtrading
Despite almost no movement of oil through the Strait of Hormuz over the past 24 hours, oil futures prices have steadily declined.
English
@MarioNawfal Oh wow another picture of an angelic girl, how cute. Thats great we have AI!
English
Grok Imagine 1.0 is insane.
Imagine V1.5.
Elon Musk@elonmusk
This is just Grok Imagine 1.0. V1.5 is a major upgrade.
English
If everything can be automated with AI, what will humans do?
English
@explorersofai Seriously??
Overwhelmingly. The US is just the largest nat gas producer in the world.
English
Free intelligence.
Unlimited energy.
Pennies per labor.
Cool pitch.
Who's building the $10 trillion grid to power it? Asking for the taxpayers.
Peter H. Diamandis, MD@PeterDiamandis
If You're an Entrepreneur: Stop designing businesses for 2024 scarcity. Design for 2030 abundance. Assume intelligence is free, energy is unlimited, and robotic labor costs pennies per hour. What becomes possible that's impossible today?
English
@QuintenFrancois I ask 5 AI models: Will Bitcoin be above $100k on Dec 31 2026?
From tranining data only (no web access):
-> Claude & Grok: YES
-> GPT5.2 was a NO, but GPT-5.4 is a firm YES.
-> DeepSeek is a NO.
-> Gemini is the usual coin flip.
Still running the question...every day.
English
Everyone expects Bitcoin to go much lower
That’s exactly why it won’t happen
Against Wall Street@aganstwallst
Bitcoin is STILL mirroring 2022 move-for-move. Kinda terrifying, right? 😱 We’re about to find out if history repeats itself
English
I asked 4 AI models who's most associated with responsible AI.
Not one named its own company.
@OpenAI @AnthropicAI @GoogleAI @xaboratory
English