- AI KATANA
- Posts
- AI could wipe out half of all entry-level white-collar jobs - CEO of Anthropic
AI could wipe out half of all entry-level white-collar jobs - CEO of Anthropic
Also: OpenAI Picks Seoul for First Korean Office
AI KATANA
May 29, 2025
Hello!
Today’s edition captures a critical inflection point in AI’s impact on the global workforce and society. Leading voices like Anthropic CEO Dario Amodei are sounding the alarm over the potential displacement of millions of entry-level white-collar jobs by agentic AI systems, calling it a looming crisis with far-reaching social and economic consequences. Meanwhile, the AI gold rush continues: Nvidia posts a record-breaking quarter despite export hurdles, Meta’s AI assistant surpasses a billion users, and OpenAI deepens its footprint in Asia with a new Seoul office. In parallel, the cultural tide is turning with college students embracing ChatGPT, and new tools from Google and DeepSeek push the frontiers of accessibility and performance. As these developments unfold, venture capital keeps flowing, betting on startups that promise to redefine how real-time data and video intelligence integrate with AI infrastructure.
Sliced just for you:
⚠️ AI could wipe out half of all entry-level white-collar jobs - CEO of Anthropic
👩🏻🎓 The AI Hype Index: College students are hooked on ChatGPT
📈 Nvidia’s Record Quarter Clouds by China Risks
📱 Meta AI Surpasses One Billion Monthly Users
🇰🇷 OpenAI Picks Seoul for First Korean Office
Anthropic CEO Dario Amodei has issued a stark warning that AI could eliminate up to half of all entry-level white-collar jobs within the next five years, potentially triggering 10–20% unemployment. Despite rapid advances in large language models and the rise of agentic AI that can perform complex tasks across finance, law, tech, and customer service, governments remain largely unprepared and silent. Amodei and others fear a swift, mass transition where businesses—already pausing hiring and investing in automation—replace human workers en masse, leading to a “white-collar bloodbath.” While some, like OpenAI’s Sam Altman, maintain a more optimistic view grounded in historical tech progress, Amodei argues for urgent public awareness, legislative briefings, and policy innovation, including a potential “token tax” on AI transactions. The race to deploy AI agents capable of near-human work is accelerating, with companies like Meta predicting entire job categories—such as mid-level engineers—could disappear imminently. Amid this disruptive shift, Amodei emphasizes the need to steer AI’s trajectory before inequality and societal imbalance become irreversible.
College students are deeply immersed in ChatGPT, a trend emblematic of the broader AI hype permeating industries and culture. The AI Hype Index tracks this fervor, highlighting both the overblown claims and genuine breakthroughs emerging across the field. While language models increasingly exhibit confident inaccuracies—sometimes fueling misinformation, hallucinations, or bizarre behavior like Grok’s conspiracy rants—there are also transformative developments underway. These include Google DeepMind’s promising problem-solving advances, AI models assisting in communication for patients with severe disabilities via brain implants, and Meta’s introduction of AI-powered friends. Yet amid this innovation, ethical concerns persist, such as citation failures in legal cases and the growing risk of misuse. The mixed reality of AI’s potential and pitfalls underscores the urgent need to discern substance from spectacle.
Nvidia reported 69 percent year-on-year revenue growth to $44 billion, blowing past Wall Street expectations despite being barred from selling its top-tier H20 accelerator in China. In a fresh SEC filing the company warned that April’s tighter U.S. export rules could carve another $8 billion from future sales and already forced a $2.5 billion write-down of unsold stock. CEO Jensen Huang nevertheless projected $45 billion for Q2 and argued that keeping Chinese open-source models on Nvidia silicon gives Washington a vital window into global AI progress. Analysts view “compliance SKUs” and geo-segmented chips as the new normal, but say the company’s unrivalled software stack still cushions demand.
Mark Zuckerberg told shareholders that Meta’s generative assistant now sees a billion monthly interactions across Facebook, Instagram, Messenger and WhatsApp, making it the fastest-growing product in the company’s history. Usage is surging fastest on WhatsApp voice chats, prompting Meta to ship a dedicated Meta AI app with a focus on personalised, voice-first conversations. Executives say the next challenge is adding thicker context memory and “micro-apps” such as travel booking, without eroding ad revenue or trust. Critics remain wary of misinformation risks at planetary scale, but the milestone highlights how embedding AI into existing platforms can outrun standalone chatbots several-fold.
OpenAI has incorporated a local entity and begun hiring for a Seoul hub, its third in Asia after Tokyo and Singapore. Chief Strategy Officer Jason Kwon called South Korea a “full-stack ecosystem” where world-class chip fabs sit next to consumer-app giants and a user base that ranks second only to the U.S. in paid ChatGPT subscriptions. The office will court partnerships with telecoms, hardware makers and the government, and follows a Kakao tie-up to localise Korean-language models. Lawmakers from both major parties have already scheduled briefings, signalling Seoul’s intent to weave frontier AI into national strategy.
🛠️ AI tools updates
Chinese startup DeepSeek quietly uploaded R1-0528, an updated reasoning model, to Hugging Face. Benchmarks on LiveCodeBench show it edging out several Western mini-models in code generation while trailing only OpenAI’s o3 and o4 mini. Though billed as a “minor trial upgrade,” the release signals relentless iteration even before the awaited R2. Open-source watchers say the jump tightens the race on cost-efficient alternatives and complicates assumptions that export curbs would slow China’s model quality.
Marking its 10th birthday, Google Photos rolled out a rebuilt editor that puts generative tools front-and-centre. Auto Frame suggests new compositions, Reimagine drops text-prompted objects into images, and AI Enhance offers three one-click variations that blend sharpening, lighting and object removal. Previously Pixel-only effects now head to all Android devices next month, with iOS to follow later in the year. The interface trims tabs to a single toolbar and adds QR sharing for albums, turning the once-simple backup app into an AI playground.
💵 Venture Capital updates
San Francisco-based Chalk secured a $50 million Series A led by Felicis at a $500 million valuation. Its platform pipes proprietary data streams into models for fraud detection, lending and solar optimisation, positioning itself as a real-time complement to batch-oriented data lakes. Investors cite growing demand for low-latency feature stores as companies push AI decisions further toward live user interactions.
Sydney-based Unleash Live closed a Series B that lifts its valuation to $100 million. The nine-year-old company lets enterprises plug CCTV, drones and smartphones into its cloud platform, then buy off-the-shelf computer-vision apps for tasks ranging from power-line inspections to public-transport analytics. New funds will fuel expansion into the Americas and Europe and grow its AI App Marketplace.
🫡 Meme of the day
⭐️ Generative AI image of the day
Before you go, check out We Made a Film With AI. You’ll Be Blown Away—and Freaked Out.
