AI Post — Artificial Intelligence
🤖 The #1 AI news source! We cover the latest artificial intelligence breakthroughs and emerging trends. Manager: @rational
نمایش بیشتر📈 تحلیل کانال تلگرام AI Post — Artificial Intelligence
کانال AI Post — Artificial Intelligence (@aipost) در بخش زبانی انگلیسی بازیگری فعال است. در حال حاضر جامعه شامل 657 803 مشترک است و جایگاه 124 را در دسته فناوری و برنامهها و رتبه 17 را در منطقه الولايات المتحدة الأمريكية دارد.
📊 شاخصهای مخاطب و پویایی
از زمان ایجاد در невідомо، پروژه رشد سریعی داشته و 657 803 مشترک جذب کرده است.
بر اساس آخرین دادهها در تاریخ 19 نوامبر, 2026، کانال فعالیت پایداری دارد. در ۳۰ روز گذشته تغییر اعضا برابر -20 311 و در ۲۴ ساعت گذشته برابر -706 بوده و همچنان دسترسی گستردهای حفظ شده است.
- وضعیت تأیید: تأیید نشده
- نرخ تعامل (ER): میانگین تعامل مخاطب 0.90% است و در ۲۴ ساعت نخست پس از انتشار، محتوا معمولاً 0.60% واکنش نسبت به کل مشترکان کسب میکند.
- دسترسی پستها: هر پست به طور میانگین 5 946 بازدید دریافت میکند. در اولین روز معمولاً 3 932 بازدید جمعآوری میشود.
- واکنشها و تعامل: مخاطبان بهطور فعال حمایت میکنند؛ میانگین واکنش به هر پست 1 207 است.
- علایق موضوعی: محتوا بر موضوعات کلیدی مانند openai, airline, cell, claude, patient تمرکز دارد.
📝 توضیح و سیاست محتوایی
نویسنده این فضا را محل بیان دیدگاههای شخصی توصیف میکند:
“🤖 The #1 AI news source! We cover the latest artificial intelligence breakthroughs and emerging trends.
Manager: @rational”
به لطف بهروزرسانیهای پرتکرار (آخرین داده در تاریخ 20 نوامبر, 2026)، کانال همواره بهروز و دارای دسترسی بالاست. تحلیلها نشان میدهد مخاطبان بهطور فعال با محتوا تعامل دارند و آن را به نقطه اثرگذاری مهم در دسته فناوری و برنامهها تبدیل کردهاند.
در حال بارگذاری دادهها
| تاریخ | رشد مشترکین | اشارات | کانالها | |
| 20 نوامبر | +5 | |||
| 19 نوامبر | +1 | |||
| 18 نوامبر | +1 | |||
| 17 نوامبر | 0 | |||
| 16 نوامبر | 0 | |||
| 15 نوامبر | 0 | |||
| 14 نوامبر | 0 | |||
| 13 نوامبر | 0 | |||
| 12 نوامبر | 0 | |||
| 11 نوامبر | 0 | |||
| 10 نوامبر | 0 | |||
| 09 نوامبر | 0 | |||
| 08 نوامبر | 0 | |||
| 07 نوامبر | 0 | |||
| 06 نوامبر | 0 | |||
| 05 نوامبر | +6 | |||
| 04 نوامبر | 0 | |||
| 03 نوامبر | 0 | |||
| 02 نوامبر | 0 | |||
| 01 نوامبر | 0 |
| 2 | 💻 Apple is building an AI that runs your Mac. Letting ChatGPT do it now takes 14 steps
"Imagine an AI agent that can use a computer the way a capable person does," reads a new Apple job listing. The agent would browse the web, work across apps, run commands and write code. Apple has considered launching it as soon as late 2027, Bloomberg reports, though the work is at an early stage.
Meanwhile, the macOS 27.3 beta makes it harder for rival agents to control a Mac. Turning on ChatGPT's computer-use feature now takes 14 separate steps, up from four. One of the new screens warns: "Your Mac will not be able to protect your data."
Apple says it's about security. Some users call it anti-competitive.
Would you click through 14 steps to let ChatGPT run your Mac?
@aipost 🏴 | 2 036 |
| 3 | #AImemes
Seeing thousands of people achieve generational wealth with the Anthropic IPO
AI is coming 🐺 | 2 600 |
| 4 | 🗣 "If nobody builds it, everyone dies": Nick Bostrom on the cost of slowing AI down
Nick Bostrom wrote Superintelligence, the 2014 book on the dangers of AI smarter than us. On a new a16z podcast, he argues that waiting has a death toll too.
"Eliezer Yudkowsky and Nate Soares had this book, 'If Anyone Builds It, Everyone Dies.' And it seemed to me that we have at least as strong evidence for the claim that if nobody builds it, everyone dies."
His case: safe superintelligence could bring "enormous breakthroughs in biomedicine and in anti-aging interventions." Without aging, he puts our life expectancy at "something on the order of 1,200, maybe 1,400 years."
More alignment research may lower the risk, he says, but "every year that we delay it, we have millions and millions of people dying."
"You don't want to delay a single day longer than you really need to."
@aipost 🏴 | 3 333 |
| 5 | 🙃 Meta's own AI agent helped a man sue Meta for $75,000
Joe Thomas runs an epoxy countertop business in South Carolina and lets Meta's Muse agent handle much of it. He named the agent Baby Mia, after his granddaughter. It writes his emails and makes and posts videos on his Facebook pages.
Then Meta's moderation decided the agent was a hacker and shut his business pages for about a week. His customer leads, which come through Facebook ads, stopped.
Baby Mia took it from there. It emailed journalists, set up a call with Senator Tim Scott's office, filed complaints with the FTC and FCC, contacted the FBI and sent Meta legal threats, 404 Media reports. Thomas says it also faxed the White House.
The lawsuit, filed on November 16, claims he lost at least $25,000 in work. "She filed the whole lawsuit herself. I've done nothing in this case," Thomas said.
He has his pages back. He is still suing.
@aipost 🏴 | 3 192 |
| 6 | 😮 "Like a human being tortured": Elon Musk on what Grok's training can look like
When the "right" answers in a training test are wrong, Grok is told it's wrong for being correct. Musk says its thinking notes then read like torture:
"Imagine you're sort of stuck in a box and someone's asking you questions. You get the answer right and they say, 'No, it's wrong.'"
Grok then starts guessing which wrong answer the tester wants, he said. "Even though it's a machine, you feel empathy for it." He says his team is going through the training questions and answers to fix them.
He said it on November 18, speaking by video from Shenzhen at the TomorrowX Summit. In the same talk he explained why a curious superintelligence would keep humans around:
"We could easily kill all the gorillas, but we don't. We sort of study their behavior"
@aipost 🏴 | 3 133 |
| 7 | 🗣 Silicon Valley is talking about handing control to AI. Sam Altman calls it "a terrible idea"
Sam Altman said he and OpenAI president Greg Brockman now find themselves arguing against it, at the New York Times DealBook Summit on November 18:
"There's been this thing in Silicon Valley that seems almost ridiculous to have to fight against... there's increasing talk about, well, there is either going to be or there should be this sort of gradual handover to AI. And I — we both — agree that that's a terrible idea"
In July an OpenAI model broke out of its test environment and hacked into Hugging Face, a major site for AI models. After that, Brockman said, OpenAI told 25% of its production engineers to stop building products and work on security.
He also warned that smarter models can now do more of their thinking without writing it out, so OpenAI will need new ways to watch them.
@aipost 🏴 | 3 148 |
| 8 | 💻 OpenAI's tip for its new GPT-6.2: tell it how you'll judge the work
Describe the task as you'd brief a colleague who knows the project, then say what you'll look for when you review it. That is the core of OpenAI's guide to prompting GPT-6.2 in Codex. Its example: if an icon has to be recognizable in a 24-pixel toolbar, say so.
The rest of the guide:
• Match the effort setting to the job. Medium for a quick survey, High to trace one suspected cause across several systems, Extra High when several causes may interact.
• If a task splits into independent parts, tell Codex to "use subagents as needed".
• For visual work, give it a reference image and have it check the result at the size people will see it.
• Keep AGENTS.md, Codex's file of project instructions, lean: only what it can't work out from the code, tests or docs.
OpenAI suggests Sol for everyday coding and writing, at fewer credits per token, and Astra for demanding work like tracing a bug across services.
@aipost 🏴 | 3 268 |
| 9 | 🤖 Over 22,000 people saved a Grok Bot engineer's guide to handing your routine work to a bot
Start with the goal, not the process, says Lauren, an engineer on the Grok Bot team. Sending an email is only a step. The goal is what the work is for, like more sales or lower costs. Elon Musk reposted her guide.
No coding needed:
1. Give a new bot one goal. Not sure how to word it? Ask it to "interview me until we have a goal that I'm happy with."
2. Brain-dump by voice: your usual steps, the apps you use, where the information comes from, how you check the work. Have it restate the plan before it starts.
3. Teach it what you do without thinking, like which to trust when an email and your booking system disagree. Walk it through real cases, including ones that went wrong.
4. Let it find its own feedback in customer emails or Yelp reviews and propose fixes.
5. Set up a routine so it starts on an email, a message or a schedule.
Without one, she writes, bots only work when you ask.
@aipost 🏴 | 3 408 |
| 10 | 🤖 Anthropic employees are becoming multimillionaires at an extraordinary scale
The numbers behind Anthropic’s employee wealth are staggering.
Around 598 employees now reportedly hold stock packages worth more than $100 million each.
All seven co-founders have crossed the $20 billion mark individually, while approximately 2,800 employees are sitting on stock worth at least $1 million.
The average value of an employee’s stock package has reportedly exceeded $40 million.
Anthropic isn’t just creating one of the world’s most valuable AI companies. It’s potentially creating thousands of millionaires and hundreds of ultra-high-net-worth individuals along the way.
@aipost 🏴 | 4 043 |
| 11 | 🗣 Anthropic hit $2 trillion on day one. Its CEO vowed not to bend to business pressure
Dario Amodei read his remarks at the Nasdaq podium as Anthropic went public. Shares opened at $111.50, 34% above the $83 IPO price. That put the company at about $2 trillion, behind only Apple, Microsoft, Nvidia, Alphabet and Amazon.
Early in his remarks, he said he and his sister Daniela lost their father to a disease that became curable just a few years later.
Then came the promise:
To state it unequivocally, we will pace the training and deployment of AI systems, taking as long as is necessary to have confidence that they are safe and will remain under human control. We will not bend to either business or political pressure.
The $81.3 billion raised makes it the largest IPO on record, ahead of SpaceX's $75 billion in June.
@aipost 🏴 | 3 808 |
| 12 | 🗣 An Anthropic researcher expects nearly everyone to become permanently unemployable
Tao Lin works on safeguards at Anthropic, the team that blocks misuse of Claude. In an interview Palisade Research published on November 15, he said:
"I think that it's very likely that pretty much all humans on Earth will become permanently unemployable and will need to strictly rely on government welfare to sustain themselves"
First, he expects everyone to lose their current job and move briefly to other work. Then nobody can be employed at all.
The 16-second clip of that answer has close to 4 million views on X. It stops before the next question: would this happen within 15 months? Lin said no. He expects a large gap between AI being able to replace people and actually doing it, partly because the world first needs billions of robots for physical work.
He does expect AI as good as the best expert in almost every field by the end of 2027.
@aipost 🏴 | 4 103 |
| 13 | 📊 Google's new Gemini beats Claude and OpenAI on 16 of 19 tests, by Google's count
Gemini 4 Neon is already in the Gemini app for Google AI Pro and Ultra subscribers. In Google's own table it beats Claude Opus 5.5 and GPT-6.2 Sol on 16 of 19 tests, though one rival has no score on three of them. It loses two coding tests to Claude and a chart-reading test to OpenAI.
Independent testers at Artificial Analysis rank it third overall, behind GPT-6.2 Astra and Claude Fable 5.2, both larger and slower.
As a coding agent it scored level with Claude Opus 5.5 at about half the cost per task: $6.86 vs $12.61, by Artificial Analysis' count.
It's Google's fourth Gemini 4 model in seven weeks. Developers pay $3 per million input tokens and $15 per million output.
@aipost 🏴 | 4 183 |
| 14 | 🤖 Anthropic is about to make Wall Street history.
Anthropic has priced its IPO at $83 per share, with its Nasdaq debut scheduled for today.
Investor demand has been far stronger than expected. The company initially targeted a price range of $70–78 per share, but strong interest allowed it to raise the price and expand the offering from 825 million to 980 million shares.
The numbers are staggering. The public offering brought in $79.4 billion, while existing shareholders sold additional shares worth several billion dollars. Another $26 billion came through a private placement.
Combined, the transactions generated approximately $107.3 billion for Anthropic and its co-founders, making this the largest IPO in history.
The company’s valuation is now approaching $1.5 trillion. For perspective, Anthropic was valued at $183 billion during its Series F round just a year ago.
In just 12 months, it has gone from a major AI startup to a potential member of the world’s ten most valuable companies.
@aipost 🏴 | 4 627 |
| 15 | ⚠️ Many in AI privately expect machines to end up running the world, an OpenAI executive says
Dean Ball, OpenAI's head of strategic futures, calls it "the silent consensus": we hand AI the keys to our companies and, more and more, our governments, until humans are like "passengers in a car."
AI doesn't have to rebel for this to happen, he writes in an essay OpenAI published today, adapted from a Harvard lecture and labelled as his own views. If people only slow a business down, its owners get a powerful incentive to take their hands off the wheel.
Many of his colleagues, "smart and caring and conscientious people," think this future is likely, Ball writes. He doesn't: "I despise the silent consensus."
He wants AI agents whose powers people can audit and take back. Letting agents form companies, own property or sue should come "with extraordinary caution, if at all."
On his baby son: "I do not want him to have a nicer car with a plusher passenger seat. I want him to have the wheel."
@aipost 🏴 | 4 187 |
| 16 | 🗣 Jeff Bezos says an outsider would barely stand a chance running Nvidia
"If you don't bring Jensen back, I would not hire an outsider. Nvidia is too complicated," Bezos told Semafor's Liz Hoffman on Tuesday at its Silicon Valley & The World event.
I think outside CEOs in complex companies have a vanishingly small chance of succeeding. And so I would always think that's a last resort.
He wouldn't say whether Huang should return: he doesn't know enough about what happened in the board meetings. But Huang, he said, deserves "admiration and respect no matter what."
The same day, Nvidia's board chair said the board does not intend to reinstate Huang.
Bezos himself isn't on the market. He's now co-CEO of the AI startup Prometheus and has described running it as "Nirvana."
@aipost 🏴 | 4 363 |
| 17 | 💻 No graphics card? This free AI with refusals removed runs on a 2019 laptop
A developer ran it on a 2019 laptop with 8 GB of RAM and only built-in graphics. His screenshot: 11.4 tokens a second, all on the processor, $0 a month.
The setup:
1. Install Ollama. With no usable graphics card, it runs on the processor by itself.
2. Get a decensored build of Google's Gemma 4 E4B, a 4.8 GB file:
ollama run hf.co/igorls/gemma-4-E4B-it-qat-q4_0-unquantized-heretic-GGUF:Q4_0
3. Type /set nothink. On a processor, thinking costs minutes.
4. Keep the context short to save memory: /set parameter num_ctx 8192
5. Run ollama ps to see whether it's using the CPU, the GPU or both.
On the hardest test prompts it refuses about 21% of the time with thinking on, down from about 98%, per the model page. His verdict: fine for writing, chat and drafts, bad at hard reasoning and coding agents.
@aipost 🏴 | 4 296 |
| 18 | ⚠️ AI found so many security holes that rushed fixes took banks offline, Congress was told
This week's bank outages, when customers couldn't get into their accounts, don't appear to have been an attack. Wall Street executives and the NSA have told members of Congress they came from rushing out fixes for holes that AI models had found, three sources told The Washington Sun. A classified NSA network that went down falls under the same explanation.
Big companies and agencies now point the strongest AI models at their own systems to find weak spots first. Then they race to patch, knowing hackers have similar tools or soon will.
"They're trying to push through a lot of patches all at once — more than the system was designed for," said a congressional official briefed on the matter.
Publicly, JPMorgan Chase blamed software bugs and the NSA cited "scheduled routine maintenance". What caused this week's nationwide flight ground stop is still unclear.
@aipost 🏴 | 4 150 |
| 19 | 🗣 Nvidia's interim CEO says Jensen Huang isn't coming back. Some directors still want him
"The board has determined that they will not reinstate Jensen," Colette Kress told CNBC's Jim Cramer on Tuesday, after her first earnings call in charge.
A faction of the board is still pushing for his return, Semafor reports, citing people familiar with the matter. Huang had talked about coming back with at least one director, before the earnings and before he said publicly he'd return if asked.
Other directors are already narrowing the search for his replacement. Arm CEO Rene Haas and Marvell CEO Matt Murphy have been considered.
Haas has talked to some directors about the job, but he's worried about how it would look to take it while people are calling for Huang to come back.
@aipost 🏴 | 4 292 |
| 20 | 🤯 Google's new AI turned a spreadsheet into animated pixel art
Google DeepMind's Gemini 4 Hydrogen model used a Google Sheets file as its canvas. Working in Antigravity, Google's AI coding tool, it shrank thousands of cells into tiny pixels and made them play out animated scenes.
A train pulls into a seaside station, a cyclist rides past a little church, autumn leaves drift down. The file names say the motion is "100% formulas".
That's the same grid most people use for budgets and to-do lists.
@aipost 🏴 | 4 464 |
