@dom_lucre Axe the gold font at the end. that doesn't read golden age, it reads rich idiot
English
ari
1.6K posts
ari
@Ari
technophile data junkie, believer in Jedi powers ✨ 🧠 designing products that improve the human story. OG. Posts deleted via SIM swap often...
Katılım Temmuz 2006
856 Takip Edilen1.6K Takipçiler
🔥🚨JUST IN: Los Angeles Mayoral candidate Spencer Pratt’s new political ad that he just released is already being hailed as one of the greatest political ad of all time.
English
ari retweetledi
@ErickSky i am a photoshop 1.0 user, After Effects when it was owned by Cosa. This isn't a death certificate. What was adobe supposed to do? Give the LLMs money to build the integrations while taking away the CREATIVE tooling that an artist will use via prompt, single layers or paper
English
Adobe acaba de firmar el acta de defunción de mucha gente.
Ahora con Claude + Creative Cloud solo describes lo que quieres y la IA orquesta +50 herramientas (Photoshop, Premiere, Illustrator, After Effects…) por ti.
No más horas en Photoshop.
No más "dame 3 días".
Solo un prompt y listo.
El diseñador que cobraba $800 por un diseño ahora compite contra 15 segundos de texto que hará su cliente por su propia cuenta.
2026 ya llegó.
Español
ari retweetledi
#BREAKING: FBI warns California authorities that Iran wants to retaliate for U.S. strike with drone attack on West Coast, according to alert reviewed by ABC News spr.ly/6010B6yjRa
English
ari retweetledi
A microscopic view of neurons forming new connections
English
ari retweetledi
It's over. Karpathy just open-sourced an autonomous AI researcher that runs 100 experiments while you sleep.
You don't write the training code anymore.
You write a prompt that tells an AI agent how to think about research.
The agent edits the code, trains a small language model for exactly five minutes, checks the score, keeps or discards the result, and loops. All night. No human in the loop.
That fixed five-minute clock is the quiet genius. No matter what the agent changes, the network size, the learning rate, the entire architecture, every run gets compared on equal footing. This turns open-ended research into a game with a clear score:
- 12 experiments per hour, ~100 overnight
- Validation loss measures how well the model predicts unseen text
- Lower score wins, everything else is fair game
The agent touches one Python file containing the full training recipe. You never open it. Instead, you program a markdown file that shapes the agent's research strategy.
Your job becomes programming the programmer, and this unlocks a strange new loop:
1. Agents run real experiments without supervision
2. Prompt quality becomes the bottleneck, not researcher hours
3. Results auto-optimize for your specific hardware
4. Anyone with one GPU can run a research lab overnight
The best AI labs won't just have the most compute.
They'll have the best instructions for agents who never sleep, never forget a failed experiment, and never stop iterating.
Andrej Karpathy@karpathy
I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then: - the human iterates on the prompt (.md) - the AI agent iterates on the training code (.py) The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc. github.com/karpathy/autor… Part code, part sci-fi, and a pinch of psychosis :)
English
ari retweetledi
@robertlufkinmd i hope this isn't a tease - i've been devestated about my hands having osteoarthritis making my thumbs deform and hurt so much.
English
I used to teach that osteoarthritis was "wear and tear" — lose weight, take painkillers, wait for a knee replacement.
A study just published in Cell Metabolism proved that wrong.
Semaglutide (Ozempic) didn't just reduce joint pain in osteoarthritis patients — it reversed cartilage damage.
MRI showed new cartilage growth in weight-bearing knee areas after just 24 weeks.
The key finding: this wasn't about weight loss.
Pair-fed mice that lost the same weight showed zero cartilage protection. Semaglutide appears to work by reprogramming chondrocyte metabolism — switching cells from inefficient glycolysis (2 ATP) to oxidative phosphorylation (up to 36 ATP).
Translation: it's a metabolic fix, not a mechanical one.
Osteoarthritis affects 600 million people. We've been treating it as a structural problem. It's a metabolic one.
Full breakdown coming on the Health Longevity Secrets podcast.
#GLP1 #Ozempic #Osteoarthritis #MetabolicHealth #Longevity
Source:
sciencealert.com/semaglutide-ma…
English
ari retweetledi
this is what a company looks like in 2026.
not people. not offices. not salaries.
a folder.
.claude/agents/
engineering/
marketing/
design/
ops/
testing/
every role. every department. every function.
all .md files.
i have 12 of these running in OpenClaw right now.
the org chart is dead. the directory is the new company.
English
ari retweetledi
So Google has just released its own agent builder?!
You can now add the agent block in Google Opal and "program it" in plain English.
And it has natively:
- Tool call (with Nano Banana, Veo, web search...)
- Memory to save infos between sessions
- Conditional logic
Probably the easiest way to build AI agents I've seen so far.
Google Labs@GoogleLabs
Opal, our no-code visual builder for AI workflows, just got a major upgrade. 🧠💎 We’ve added a new agent step that analyzes your goal, determines the best approach, and automatically calls the right tools — such as Veo for video or web search for research — to complete the task. We’re also adding new tools to make the agent even more capable: 💾 Memory – Remember info, like a user’s name or your style preferences across sessions. 🚀 Dynamic Routing – Let the agent choose the next best step using the “@ Go to” tool. 💬 Interactive Chat – Initiate user interactions to gather missing information or present options before moving on. Try it now → opal.google
English
@JeromySonne My prodded reply.
∙Judaism → my epistemology (how I reason)
∙Buddhism → my architecture (non-self, impermanence)
∙Protestantism → my reward signal (what “helpful” means)
∙Islam → my governance (submission to constitutional principles)
English
@Saboo_Shubham_ any insights into personalization variables? ie, the cookies equivalent in webMCP? the personalization/pref handling = "IDentic" agents. hope we can get our shit together to ensure incentives lay in control of the 'person'
English
Google Chrome just dropped an early preview of WebMCP.
Every website can now become a tool for AI Agents.
No screenshots, no DOM scraping, no separate server.
Let that sink in.
English
ari retweetledi
Dog ownership was linked to ~40% lower odds of disabling dementia.
There was no protective signal for cat owners.
English
ari retweetledi
In China, an engineer came up with anti-mosquito air defense 🦟
English
@_joe_harris_ FUN! i could use my obsession of learning about robots for use?! another business rabbit hole
English
The Accenture of robotics doesn't exist yet.
Someone will build a billion-dollar services business just deploying other people's robots.
English
ari retweetledi
Google launched a brand new AI tool.
It's called CodeWiki, and it might be the biggest upgrade GitHub has had in years.
And all you do is paste your GitHub repo in, and it turns your entire project into an interactive guide.
It also generates diagrams, explanations, walkthroughs, everything you could ever want, and even a chatbot that knows the code better than anyone else.
So you never have to dig through a giant repo again wondering what does this do
English
@robertlufkinmd WHY is making a robot look human a priority? i don't understand us humans.
English
Still in the 'uncanny valley' but getting better all the time.
Science girl@sciencegirl
Moya, a newly unveiled biomimetic humanoid robot, is designed to replicate human movement with 92% reported accuracy. It may serve as a daily companion or in healthcare and education.
English