AI Reading for Monday January 5

At CES 2026, integrated chatbots and built-in AI are now commonplace, and companies must focus on refining user experience to succeed in the AI era. - WIRED

Samsung co-CEO TM Roh says the company rolled out Gemini AI features to ~400M mobile devices so far and aims to hit 800M in 2026 - Reuters

Smart TVs get AI to edit photos and create videos - The Verge

make your own Netflix, with blackjack and hookers

Takway.Ai unveiled Sweekar at CES 2026, a Tamagotchi-style AI egg that uses Google Gemini and ChatGPT for memory and interaction, offers growth stages, and will launch a March Kickstarter priced $150–$200, raising child-safety and privacy concerns. - Gizmodo

AI can now be egg-shaped and needy, yay I guess

Plaud launched the $179 NotePin S, replacing the 2024 NotePin's haptic squeeze with a physical recording/highlight button, and released Plaud Desktop for Mac and PC to capture meeting audio, free for Plaud Note or NotePin owners. - The Verge

Venture capital flowed heavily to AI in 2025: more than half of VC dollars and 36% of deals went to AI, and Crunchbase reports 14% of funding went to OpenAI and Anthropic. - Fast Company

Surging AI investment may repeat past cycles of over-optimism and heavy capital spending, posing risks for investors. - Financial Times

Asset managers warn heavy corporate AI investment could drive inflation into 2026–27 and force central banks to halt rate cuts, and Morgan Stanley's Andrew Sheets forecasts US CPI will stay above 2% through end‑2027. - Yahoo

promptflation

David Atkins argues AI's rapid automation is concentrating economic returns among top firms and investors, amplifying winner-take-all dynamics and prompting calls for wealth taxes, stronger antitrust and AI governance to address rising inequality. - Washington Monthly

When 7 oligarch trillionaires own everything, then what, declare them game winners and start over? - Reddit

why not start at $1b and give them a medal saying, you won capitalism, and tax everything above that at 50%

Satya Nadella: the tech industry needs to move past “slop vs sophistication” arguments and view AI as a “cognitive amplifier”, not a human substitute - sn scratchpad

one man’s slopbot is another’s power multiplier

AI start-ups take on Google in fight to reshape web browser market - Financial Times

U.S. Immigration and Customs Enforcement is using facial recognition technology to rapidly identify and arrest people. - The Wall Street Journal

NYC Wegmans implements facial recognition for security - Gothamist

Boston Dynamics is training Atlas, an all-electric Nvidia-powered humanoid, to perform factory tasks using VR teleoperation, motion-capture demonstrations and large-scale simulation (over 4,000 simulated Atlases trained six hours), tested autonomously sorting roof racks at Hyundai's Georgia plant. - CBS News

ASML: The Ridiculous Engineering Of The World's Most Important Machine - YouTube

final boss machine

OpenAI's new report says 40 million Americans use ChatGPT for health care, employing chatbots to navigate insurance, find providers and perform basic triage, a trend that could prompt partnerships among AI vendors, health systems and insurers. - Axios

Hospitals are serving as proving grounds for artificial intelligence, demonstrating both its potential uses and its current limitations. - The Wall Street Journal

Alibaba's PANDA AI tool has analyzed 180,000 CT scans and detected about 24 pancreatic cancer cases since its deployment in November 2024, a Chinese hospital said. - The New York Times

Nitin Ware, lead engineer at Salesforce’s AI platform, warns agentic AI that autonomously executes tasks like scheduling and billing creates legal liability, fraud and trust risks for platforms, vendors and consumers and requires governance, audit trails and accountability mechanisms. - The Washington Post

agents = infinite stoner interns who could go rogue anytime

Physicist Max Tegmark, who advocates pausing AGI development, was subpoenaed by OpenAI regarding the Future of Life Institute's past ties to Elon Musk. - The Wall Street Journal

A curated catalog of agentic AI patterns documents practical techniques—task decomposition, tool orchestration, memory management, self‑healing feedback, CI code‑review loops, multimodal integration, and security controls—and aims to evolve as contributors apply and extend it. - GitHub

See also Nir Diamant - GitHub

A developer runs six Claude Code agents in parallel from an iPhone via Termius connected to a Vultr vhf-8c-32gb VM ($0.29/hr), using Tailscale, tmux/mosh for persistence and Poke webhooks to review PRs, refactor, and fix bugs. - granda

First thing to ask your AI before you vibe-code is how to set up a sandbox and back up your data - Reddit

AI-generated deepfakes are impersonating U.S. pastors to broadcast incendiary sermons and solicit donations from their congregations. - WIRED

It’s immoral to let a sucker keep his money. If AI has as much authority as a televangelist, and can iterate at higher velocity, the televangelists are in trouble.

Sarah Novak replaced a driver's-side visor on her 2020 Chevrolet Tahoe using ChatGPT instructions, after which the SUV flashed warnings and refused to start despite a nearly new $400 battery, with generic AI troubleshooting unable to clear BCM-related fault codes. - Motor1.com

To err is human, but to really mess everything up takes an AI trained on humans.

Panos Ipeirotis, an NYU Stern professor, built an AI oral-exam system using ElevenLabs speech to conduct two-part ~25-minute interviews for 36 students over nine days at about $15 compute cost, auto-graded by a council of LLMs (Claude, Gemini, ChatGPT). - Business Insider

Fight fire with fire

Follow the latest AI headlines via SkynetAndChill.com on Bluesky