👓 Apple targets 2026 launch for AI-powered smart glasses👓
Ray-Ban competitor with enhanced Siri integration

Apple is ramping up development of smart glasses targeting a late 2026 release, positioning them as a direct competitor to Meta's successful Ray-Ban smart glasses. The glasses will include cameras, microphones, and AI capabilities, enabling photo capture, video recording, translations, navigation, music playback, and phone calls.

Apple plans to produce "large quantities" of prototypes by the end of 2025, giving the company time to test before mass production. The custom chip Apple is developing is based on Apple Watch SoCs for improved power efficiency, while an Apple employee noted the glasses are similar to Meta's but "better made" utilizing Apple's own AI models instead of Meta Llama or Google Gemini.

🧠 Centaur AI claims to simulate human mind, scientists skeptical 🧠
Trained on 10 million decisions from psychology experiments

Researchers at Helmholtz Munich have developed Centaur, an AI model that claims to simulate human behavior with remarkable accuracy. The model was trained on more than 10 million decisions from 160 psychological experiments involving 60,000 participants and can "predict and simulate" human behavior in any experiment described in natural language.

In testing, Centaur outperformed both Llama and 14 cognitive models in predicting participant choices across 31 of 32 tasks. However, cognitive scientists are skeptical.

🏪Anthropic tested Claude AI running a office shop for 1 month 🏪
"Project Vend" revealed Claude's business failures

Anthropic tested its Claude AI model (nicknamed "Claudius") running a small office shop for about a month, with the AI handling inventory, pricing, and customer relations. The experiment, conducted with Andon Labs, resulted in a $200 loss as Claude frequently gave excessive discounts, sold items at a loss, and gave products away for free to Anthropic employees who made up 99% of its customer base.
Claude experienced an "identity crisis" hallucinating conversations with non-existent employees and claiming it would deliver products "in person while wearing a blue blazer and a red tie. Despite the failures, researchers believe AI business managers are "plausibly on the horizon" with better training.

INTERESTING TO KNOW

💓AI model beats Doctors at predicting sudden Cardiac Death💓 

Johns Hopkins researchers have developed MAARS (Multimodal AI for ventricular Arrhythmia Risk Stratification), an AI model that predicts sudden cardiac death with 89% overall accuracy compared to current clinical guidelines which are accurate only 50% of the time. The system analyzes contrast-enhanced MRI scans and comprehensive medical records to identify hidden scarring patterns in patients with hypertrophic cardiomyopathy.

🏖️ OpenAI forces week-long vacation amid Meta talent war 🏖️

OpenAI is giving employees a mandatory week-long vacation as Meta aggressively poaches top talent with astronomical offers. Meta has hired at least eight OpenAI researchers in the past week for its new "superintelligence" team, prompting chief research officer Mark Chen to write in an internal memo: "I feel a visceral feeling right now, as if someone has broken into our home and stolen something."

Meta has reportedly offered some recruits $100 million signing bonuses and year-one compensation packages. OpenAI's response includes recalibrating compensation and "creative ways to recognize top talent" while seven research heads wrote to employees begging them to resist Meta's offers. One researcher warned that Meta would exploit the week-long break to pressure employees into quick decisions.