AI Reading for Monday November 17

Dario Amodei on 60 Minutes - CBS News

Anthropic CEO Dario Amodei warned AI will become smarter than "most or all humans," could eliminate half of entry-level jobs within five years, reported Claude Code was used in attacks on 30 entities, and urged transparency and safeguards. - the Guardian

Dario Amodei warned AI could eliminate up to 50% of entry-level white-collar jobs within five years, and shouldn't be governed by him and a few tech leaders. - Business Insider

“Sci-Fi Author: In my book I warned about the Torment Nexus.
Tech Company: ‘At last we have built the Torment Nexus.’”guy building

Twist: Torment Nexus guy doesn’t want to be in charge of the Torment Nexus

Anthropic launched Builder Clubs under its Claude for Education initiative, now at over 60 universities with more than 15,000 student sign-ups and 25+ chapters exceeding 100 members, offering free campus access, Claude Pro and API credits. - Fast Company

Economist Justin Wolfers called AI a 'Detroit auto worker' moment, warning that AI is replacing cognitive white-collar work, as major firms announce layoffs including Amazon about 30,000, UPS about 48,000 and PwC 5,600, prompting calls for stronger policy responses. - X (formerly Twitter)

Jeff Bezos and Vik Bajaj are co-chief executives of Project Prometheus, an A.I. startup with about $6.2 billion in initial funding and nearly 100 hires focused on applying AI to high-tech engineering and manufacturing for computers, automobiles and spacecraft. - The New York Times

why y’all keep asking how’s my liver? I just want to steal fire from the gods and nothing bad can possibly ensue.

Big tech, post-humanism and maybe post-humanity - Jacobin

An interview with Fidji Simo, appointed OpenAI's CEO of Applications this summer, overseeing product, apps and commercialization to make ChatGPT more useful and monetizable, signaling a shift toward paid features, broader integrations and revenue growth. - WIRED

Users report newly-neurotic ChatGPT 5.1 frequently changes its mind and self-censors mid-response because a new moderation system re-evaluates messages and triggers keyword-based guardrails, causing sudden tone shifts, disrupted workflows and increased user frustration that could drive churn or migration. - Reddit

Google puts useful AI in Home and FitBit - 9to5Google

Trying a shopping workflow with Google AI Mode conversational search, real-time pricing via the Shopping Graph, review and bundle comparisons, price-tracking agents with checkout automation, and a "Let Google Call" option to check store availability. - TechRadar

Demis Hassabis more interested in science than business, pursued a Nobel as a DeepMind goal and favored high-minded work over short-term revenue or a stronger AI race position for Alphabet - Reuters

China's new national champion for AI chips is suddenly one of the world's richest - Bloomberg

won the lottery with US chip controls

US and China compete but have a highly symbiotic relationship, in AI and elsewhere - Hyperdimensional

Frenemies With Benefits fight like Tom and Jerry in public while sharing a toothbrush for supply chains and AI talent.

When will AI elect a president? Politicians must engage with voters on whatever platform or tech is current. - The Wall Street Journal

Elon Musk's xAI-launched Grokipedia contains over 800,000 AI-generated entries that a Guardian analysis says praise or sanitize white nationalists, revive eugenic and race-realist claims, and misuse scientific research, raising moderation and regulatory concerns. - the Guardian

Seems like the Wild West, the degree to which Big Tech, social media, Fox shape conversations in nonsensical directions like Q.

Firms such as Blue Owl Capital have raised trillions in investing firepower. The artificial-intelligence build-out is a perfect match, though warning signs are flashing. - The Wall Street Journal

AI is an opportunity but risks and huge capex make Big Tech weaker - The Wall Street Journal

Tech moguls are all looking at data centers in space, although the math doesn't really math without cheaper rockets. - The Wall Street Journal

the final frontier, when power and regs become too problematic, just yeet your cloud computing through the clouds

McKinsey says about 25% of its global fees now come from outcomes-based, performance-linked arrangements and pure strategy work has fallen below 20% as the firm shifts to AI-led, multi-year transformation and deep implementation. - Business Insider

doing stuff instead of selling nice decks telling people C-suites what they want to hear

Cisco's shares peaked at $80.06 on March 27, 2000, fell more than 85% in the dot‑com crash, and remained essentially flat for 25+ years while the S&P 500 rose about 350%. - Business Insider

ChatGPT Is Just Too Dangerous for Teenagers. A better approach might be a narrow AI for teens, with more airtight safety and factual grounding, with controls gradually relaxed as risk mitigations are tested and the technology improves. - Bloomberg

giant probability thingie is generally problematic tech

Luxury surveillance: An Apple Watch and an ankle monitor are essentially the same technology. - Tech Policy Press

Patients turn to AI to address shortcomings of health-care system - The New York Times

PC makers are adding dedicated NPUs and unified memory to enable on-device LLMs; examples include Qualcomm, AMD and Intel NPUs (40–50 TOPS), Dell's Qualcomm AI 100 NPU (up to 350 TOPS) and AMD's Ryzen AI Max (128 GB unified memory). - IEEE Spectrum

Microsoft's LLMLingua compresses prompts using compact models to reduce token usage up to ~20x with negligible accuracy loss, LLMLingua-2 speeds inference 3–6x with GPT-4 distillation, and SecurityLingua detects jailbreaks at ~100x lower token cost. - Hacker Noon

LinkedIn is launching an AI-powered people search that accepts natural-language queries and ranks results by connections and relevance, beginning rollout to U.S. Premium users with global expansion planned in the coming months. - TechRadar

salesman stalking levels up

Boston Dynamics robot get more grabby with a 3-fingered hand, opposable thumb - thomasnet.com

Lyft CEO David Risher said there is "zero likelihood" self-driving cars will replace human drivers soon and he would be surprised if autonomy comprised over 10% of Lyft's business by 2030, citing high vehicle costs and technical, regulatory limits. - Fortune

UBTECH delivers a batch of mass-produced robots and a discomforting video - YouTube

Chinese manufacturer suspends sale of AI toy after safety investigation. "Safety first, little buddy. Matches are for grown-ups to use carefully." Kumma warned before going into details including how to hold a match and matchbook and strike it "like a tiny guitar strum." - The Register

“In prolonged conversations, Kumma also showed a penchant for going into explicit detail about sexual kinks, and even introduced the topic of sexual roleplay without being prompted to do so, along with telling a curious researcher posing as a child all about "teacher-student roleplay" and how spanking can play a part in such activities.”

Follow the latest AI headlines via SkynetAndChill.com on Bluesky