- AI KATANA
- Posts
- How DeepSeek’s small team of liberal arts graduates transforms AI text generation in China
How DeepSeek’s small team of liberal arts graduates transforms AI text generation in China
Also: Meta to Launch Standalone AI App in 2025, Competing With OpenAI, Google

Hello!
Today’s AI landscape is evolving rapidly, with innovative startups and tech giants alike making bold moves. DeepSeek, a Chinese AI firm, is gaining attention for its unique approach—leveraging liberal arts graduates to refine text generation, setting itself apart in a competitive market. Meanwhile, Meta is preparing to launch a standalone AI app in 2025, directly competing with OpenAI and Google, signaling a shift in its AI strategy. In Japan, AI-powered robots are being integrated into elderly care to address labor shortages, though concerns remain about their emotional support capabilities. ByteDance has unveiled Trae, China’s first AI-native integrated development environment (IDE), further advancing its AI ambitions. On the AI tools front, BuddhaBot-Plus, a Buddhist dialogue AI, has expanded overseas to Bhutan in collaboration with the country’s monastic community. Lastly, in venture capital news, China’s Zhipu AI has secured $137 million in funding as state-backed investors continue to fuel the AI race.
Sliced just for you:
🤖 DeepSeek’s Liberal Arts Approach to AI – How a small team of humanities graduates is shaping AI text generation in China.
🚀 Meta’s Standalone AI App – Meta plans to challenge OpenAI and Google with a dedicated AI platform in 2025.
🏥 AI Robots in Japan’s Elderly Care – Can AI-driven caregiving solutions solve Japan’s aging population crisis?
💻 ByteDance’s AI-Powered IDE – Introducing Trae, China’s first AI-native software development tool.
DeepSeek, a rising Chinese AI startup, has distinguished itself in text generation by combining advanced algorithms with a human-centric approach curated by a small team of liberal arts graduates. Former employees and industry analysts credit the eloquence and natural quality of its AI-generated text to this unique mix of literary expertise and technical innovation. This approach has enabled DeepSeek’s models to quickly gain popularity, setting them apart in China’s competitive AI landscape. The company’s success highlights the growing significance of interdisciplinary collaboration in refining AI capabilities and enhancing user experience.
Meta is set to launch a standalone AI application in the second quarter of 2025, directly challenging OpenAI’s ChatGPT and Google’s Gemini. The move marks a strategic shift from integrating Meta AI within Facebook, Instagram, WhatsApp, and Messenger to a dedicated platform, aiming to boost user engagement and revenue. CEO Mark Zuckerberg has prioritized AI development, exploring monetization models like premium subscriptions similar to OpenAI’s ChatGPT Plus. While Meta AI currently boasts 700 million monthly active users across its integrated platforms, its standalone website sees significantly lower traffic. To compete in the AI space, Meta is accelerating development, with some teams reportedly working seven-day weeks to meet ambitious goals.
Japan is increasingly turning to AI-powered robots to address its aging population and labor shortages in the nursing sector. With one in ten people expected to be aged 80 or older by 2040, the country faces a severe shortage of caregivers. AI robots are being integrated into elder care facilities, assisting with tasks such as lifting patients, monitoring health conditions, and providing companionship. While these robots help alleviate the burden on human caregivers, concerns remain over their effectiveness in providing emotional support and their high costs. The government is promoting the adoption of AI technology in nursing through subsidies and research funding, aiming to balance efficiency with compassionate care in an evolving healthcare landscape.
ByteDance has introduced Trae, an AI-powered software development tool, marking China’s first AI-native integrated development environment (IDE). Trae features smart code completion, real-time bug detection, and support for multiple programming languages. It is powered by ByteDance’s in-house Doubao-1.5-pro large language model and allows users to switch between DeepSeek R1 and V3 models. While the tool was initially launched internationally with support for OpenAI’s GPT-4o and Anthropic’s Claude 3.5 Sonnet, the newly released domestic version has been optimized for China’s development landscape and user preferences. This move reinforces ByteDance’s growing ambitions in AI and software development amid intensifying competition in China’s AI industry.
🛠️ AI tools updates
BuddhaBot-Plus, an advanced AI designed for Buddhist dialogue, has made its first international debut through a collaboration between Kyoto University’s research team and Bhutan’s Central Monastic Body. The AI, powered by the latest ChatGPT iterations, will be introduced to Bhutanese monks to support religious discussions and education. The deployment follows an official request from Bhutan’s monastic community, ensuring the AI is adapted to their needs. The initiative includes rigorous safety assessments and independent monitoring by monks before wider access is granted. This project marks a significant step in integrating AI into religious and philosophical discourse while preserving traditional teachings.
💵 Venture Capital updates
Chinese AI startup Zhipu AI has raised over 1 billion yuan (US$137 million) in fresh funding, further fueling the country’s AI race amid growing competition from DeepSeek. The investment, led by state-backed Hangzhou City Investment Group Industrial Fund and Shangcheng Capital, reflects Hangzhou’s ambitions to become a major AI hub. Zhipu plans to use the funds to enhance its GLM large-language model and expand its AI ecosystem, particularly in Zhejiang province and the Yangtze River Delta region. The company has already secured 3 billion yuan in funding in recent months, as it competes with DeepSeek’s cost-efficient AI models, which rival Western alternatives. In response to market pressures, Zhipu has announced plans to open-source a suite of new AI models, including foundation models, inference models, multimodal models, and AI agents, as China’s AI firms embrace more open innovation strategies.
🫡 Meme of the day

⭐️ Generative AI image of the day

Before you go, check out Here’s what happens when AI takes over performance reviews
