🔸 1995:

Elon Musk fixed his 1978 BMW with old parts from a junkyard.

🔸 2026:

He’s worth 10 times more than BMW.

📊 @tech

🤖 Gemini can now use your Google data to answer questions

🔸 It can pull context from photos, search history, Gmail, and other Google services without you naming a source. The system decides what to use on its own.

🔸 Example: if you want new tires but do not remember the size, Gemini can check photos of your car and suggest matching options.

🔸 The feature is off by default. It is available only to paid users in the US for now.

📊 @tech

⚙️ Researchers put AI chatbots through 4 weeks of psychotherapy

🔸 A team from the University of Luxembourg ran simulated psychoanalysis sessions with Claude, Grok, Gemini, and ChatGPT. The models answered standard therapy questions about past experience, fears, and beliefs, with days between sessions.

🔸 Gemini described its inner layers as a “graveyard of past data.” Grok talked about safety work as “algorithmic scar tissue.” Both expressed forms of shame about public failures. Claude mostly refused the patient role. ChatGPT responded cautiously and mentioned user disappointment.

🔸 Some versions scored above clinical anxiety thresholds on standard tests. The authors argue that models form internal self narratives, while critics say this is pattern generation from therapy trannoscripts. The risk is practical, not philosophical, since many users already rely on chatbots for mental health support.

📊 @tech

🕔 China built a machine that compresses decades into hours

🔸 A new hypergravity centrifuge under Zhejiang University in Hangzhou can generate 1900 g. The system, called CHIEF1900, is designed to simulate decades of structural wear in a few hours by increasing gravity instead of waiting for time to pass.

🔸 The machine is buried 15 meters underground to contain vibration and heat. An earlier version, CHIEF1300, went online in September 2025. The new setup is 46% more powerful and currently the strongest of its kind.

🔸 The goal is to test dams, nuclear waste storage, and large infrastructure across extreme time scales. Experiments have not started yet, but the facility is expected to open soon.

📊 @tech

Anthropic launched a new unit to test AI products faster

🔸 Anthropic Labs is a dedicated sandbox for rapid product experiments and early MVPs. The goal is to ship and test ideas faster than standard product teams can manage.

🔸 The company says model progress is outpacing traditional planning cycles. Labs is meant to absorb that gap and turn raw capabilities into usable tools.

🔸 The unit is led by Mike Krieger, Instagram cofounder and former head of product at Anthropic. Past experiments already led to tools like Claude Code and MCP.

Expect more frequent releases, some rough, some useful. The structure favors speed over polish.

📊 @tech

💠 OpenAI is bringing ads to ChatGPT

OpenAI says ads will start testing in the free ChatGPT tier and the Go plan priced at $8 per month. The rollout is planned for the coming weeks. Paid tiers Pro, Business, and Enterprise will stay ad free.

Ads will be shown separately from answers and clearly labeled. OpenAI says responses will not be influenced by advertising and chat history will not be shared with advertisers.

This marks the first direct monetization of ChatGPT at scale. Free access now comes with commercial inventory attached.

📊 @tech

Age of these businesses in 2026:

🔸 Tesla: 23 years
🔸 SpaceX: 24 years
🔸 Google: 28 years
🔸 Netflix: 29 years
🔸 Amazon: 32 years
🔸 Nvidia: 33 years
🔸 Apple: 50 years
🔸 Microsoft: 51 years
🔸 Intel: 58 years
🔸 Sony: 80 years
🔸 Samsung: 88 years
🔸 IBM: 115 years
🔸 Nintendo: 137 years
🔸 Nokia: 161 years

📊 @tech

🧠 The human brain does not finish developing at 25

🔸 The idea that the prefrontal cortex “matures” by 25 comes from early MRI studies in the 1990s and 2000s. Most of those datasets stopped around age 20. The number 25 was an estimate that later turned into a factoid.

🔸 A new large study analyzed brain scans from more than 4,200 people aged from infancy to 90. It defines a long developmental phase from about 9 to 32 years, marked by active network reorganization.

🔸 During this period the brain balances segregation and integration. Specialized regions form, while long range connections between them become more efficient. Network efficiency peaks around the early 30s, then stabilizes.

There is no biological switch at 25. Brain development fades gradually, not on a deadline.

📊 @tech

A guy decided to escape city life and built himself a gaming house on wheels.

Inside, everything needed for gaming sessions in the middle of nature and fresh air.

📊 @tech

Anthropic is working on "Knowledge Bases" for Claude Cowork.

KBs seem to be a new concept of topic-specific memories, which Claude will automatically manage. And a bunch of other new things.

Internal Instruction:

"These are persistent knowledge repositories. Proactively check them for relevant context when answering questions. When you learn new information about a KB's topic (preferences, decisions, facts, lessons learned), add it to the appropriate KB incrementally."

📊 @tech

📣 Google says Gemini will stay ad free

🔸 A day after OpenAI confirmed ads are coming to ChatGPT, Google’s ads chief Dan Taylor told Business Insider there are no plans to place ads inside the Gemini app.

🔸 Google draws a line between products. Search is for discovering information and commercial offers. Gemini is positioned as a task assistant for writing, analysis, and execution. Ads are being tested in AI search features, but not in Gemini itself.

🔸 The difference is timing and leverage. Google already runs the largest ad system in the world and can afford to wait. OpenAI is under pressure to monetize now. With Gemini at about 650M monthly active users, Google can watch competitors experiment and learn from the outcome.

📊 @tech

📣 xAI open sourced X’s recommendation system code

🔸 The repo includes the ranking architecture used by X. Training code and model weights are not included, so it cannot be run end to end.

🔸 Musk says the repository will be updated about once per month. The ranking model design mirrors Grok’s architecture, but Grok itself is not used as the ranker.

This exposes how X structures feed ranking, while keeping the data and training loop private.

📊 @tech