machine learning
26K posts
machine learning
@Mlearning_ai
https://t.co/36eFDHoV3o 🟠 https://t.co/rjCZMWrRMK #Art tools for creative economy
Katılım Temmuz 2012
391 Takip Edilen10.6K Takipçiler
@Mlearning_ai But this query was one shotted entire app, wasn't? Because in my case Gemini was able to write 10k lines of code before I burned my limit.
English
A single query in Gemini uses up the entire limit
Google Gemini is a dead project
English
@Mlearning_ai what did you query 😭😭😭
the most i got was 75% of 5h usage and it was after using Pro across multiple sessions
English
Gemini’s final response before hitting the limit
If this weren’t a paid service, it might almost be funny.
Google still don’t get AI
never did
English
Connect Grok Build xAI with Hermes Agent (Nous Research): A Practical Guide
🟠
30+ Tips & Tricks. How Hermes Agent by Nous Research works with Grok Build xAI, a practical guide covering models, API keys, SuperGrok, prompt caching, context window, VPS setup Ready to use shortcuts
English
machine learning retweetledi
do you understand what Google's AI just did to an artist..
His entire Google account got permanently banned. Not just Drive. Gmail.. YouTube.. Every single service..
His appeal was rejected. No human reviewed it. An algorithm decided his life's work was a violation.
He never shared the files publicly. It was a private backup of his own creations. The AI flagged it anyway - probably the filename or art style - and that was enough.
- Google banned a developer's 14-year-old Gmail account over a research dataset that contained no illegal content
- Google expanded its automated ban policy in October 2025 - violations now trigger immediate termination with zero warning period
- No lawsuit against Google for wrongful account termination has ever succeeded in US courts
Your Google account is not yours. You are renting access to your own digital life from a company whose AI can end it in seconds - with no appeal, no human, and no recourse.
AyakaMods@AyakaMods
Google just permanently banned a manga artist’s entire Google account, just for uploading his own old manga files to Drive. AI moderation triggered and flagged it, he tried to submit appeal then he got rejected it by Google and now he has lost everything like Gmail, Drive, all linked services is gone. He never even sharing the files publicly, it’s only backing up his own a private work like any creator and artists. This is Google Drive “AI moderation” in action. No human support and no serious to take action. Physical storage or real private alternatives only. Support the artists getting screwed by this. This level of corporate overreach is insane.
English
Here's the Erdős unit distance problem explained as simply as possible.
The Core Question
Imagine you have n dots scattered on a piece of paper. You pick one fixed distance, say, exactly 1 inch. The question is: how many pairs of dots can be exactly 1 inch apart?
That's it. That's the whole problem.
Why It's Tricky
Say you have 4 dots. You could arrange them so many pairs are exactly 1 inch apart, but you can't make every pair that distance. Geometry fights back. The question Erdős asked in 1946 was: what's the maximum possible count, and what arrangement achieves it?
The Old Belief (80 Years of It)
Mathematicians found that arranging dots in a square grid like graph paper, gave really good results. A grid of 100 dots might produce hundreds of unit-distance pairs. For decades, experts believed grid-like arrangements were basically the best you could do.
What OpenAI's Model Did
The AI found a completely different family of arrangements not grid-shaped at all, that squeezes out more unit-distance pairs from the same number of dots. It's like everyone assumed checkers was the best board game layout, and suddenly someone showed up with a totally different design that scores higher by its own rules.
Why This Matters
Two reasons:
Math reason: It disproves a decades-old intuition and pushes the boundary of what's known about geometry and combinatorics.
AI reason: The model didn't just assist a mathematician, it found the discovery autonomously, making it the first AI to independently solve a prominent open problem in a core math field.
Think of it as AI graduating from calculator to collaborator
OpenAI@OpenAI
Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids. An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better. This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.
English
Google just bet the farm on AI agents… did it pay off?
this clean summary of Google I/O 2026 shows how hard they’re pushing the “agentic Gemini era”
Agents doing the heavy lifting, massive Search redesign, Spark as your 24/7 helper, they’re playing full offense.
Still, the video makes it clear the only real standout tool is Gemini Omni (the video gen beast that likely made this summary itself)
The rest feels like aggressive positioning more than must-have new stuff. Mixed crowd reaction makes sense
The winners are definitely not yesterday's announcements
English
machine learning retweetledi
machine learning retweetledi
Grok Build 0.1.210 Beta
First impression: minimalist and probably cool
/imagine
/imagine-video build in
English
machine learning retweetledi
Codex /goal
One clear goal → it autonomously plans, codes, tests & iterates nonstop until it’s verifiably done.
No more constant prompting
1️⃣ codex features enable goals first
2️⃣ Define exact success criteria (tests + deploy + metrics)
3️⃣ Use /goal pause & /goal resume for long runs
Best for full features, refactors & backlogs
English
funniest part?
people really out here thinking something changed
Gergely Orosz@GergelyOrosz
Looking at these these texts published between Mira Murati and Sam Altman (as part of the OpenAI trial): the (only?) adult in the room at the time was Satya. Texts, emails and everything written can be used in discovery, like now. Satya simply picked up the phone and talked.
English
🚨 Claude Code Routines
A routine is a standing instruction you set up once. Claude runs it on its own from that point forward, every time the conditions you defined show up
Every routine has three pieces: How it changes your day: you design the instruction once, Claude carries it out every time the trigger fires, and you step in only when a human eye is needed, usually to approve a draft or make a judgment call
1️⃣ A trigger, what wakes it up. A clock time (every morning at 9), an event (a new email, a new Substack comment, a new file in a folder), or a signal sent by another app.
2️⃣ The work, what Claude actually does once awake. Read something, write something, summarize, draft, file.
3️⃣ A destination, where the result lands. Your inbox, a Google Doc, a Slack channel, a folder of drafts for you to look at when you have time
Briefings, summaries, and first-draft replies are already waiting when you sit down. Your hours go to the work that requires you.
The simplest way to start: pick one small task you do every week on autopilot. Write down the trigger, the work, and the destination in plain English. That description is already a routine. Setting it up inside Claude is mostly a matter of pasting that description into the routines panel and giving it the keys to wherever the result needs to go.
English
machine learning retweetledi
How to Run a Company Alone in 2026:
The One-Person $Billion Company Playbook
🟣
A beginner’s step-by-step guide to using the new AI founder stack to run engineering, marketing, sales, and operations without hiring a team
English
machine learning retweetledi
Codex /goal:
Beginner Guide to OpenAI’s New Autonomous Goal Feature
Ralph-loop style autonomy, now official
40+ Practical Tips & Tricks
/goal Step-by-Step Setup
English