• AI KATANA
  • Posts
  • Why Everyone’s Talking About China’s New Autonomous AI Agent ManusAI

Why Everyone’s Talking About China’s New Autonomous AI Agent ManusAI

Also: Microsoft reportedly ramps up AI efforts to compete with OpenAI

Author

AI KATANA
March 09, 2025

Hello!

Today’s newsletter dives into China’s latest AI breakthrough, ManusAI, an autonomous agent with advanced real-world task execution capabilities that may redefine human-machine collaboration. Meanwhile, Microsoft is ramping up its AI efforts to compete with OpenAI, developing its own reasoning models and exploring alternative providers amid growing tensions. Experts are increasingly predicting that AGI could arrive as early as 2026, though skepticism remains among leading AI researchers. In Singapore, new government initiatives aim to boost AI literacy for individuals and businesses, ensuring workforce readiness in the AI-driven economy. On the corporate front, Alibaba’s new AI model, QwQ-32B, is intensifying China’s AI race with its efficient reasoning capabilities. Google is also making moves, launching Gemini 2.0 to enhance AI-powered search, while China has established a $138 billion government-backed venture fund to accelerate breakthroughs in quantum computing and AI. Lastly, AI startup funding hit a record $103 billion in 2024, cementing AI’s dominance in global venture capital.

Sliced just for you:

  • 🤖 Why Everyone’s Talking About China’s New Autonomous AI Agent ManusAI

  • 🚀 Microsoft reportedly ramps up AI efforts to compete with OpenAI

  • 🔮 AGI could now arrive as early as 2026 — but not all scientists agree

  • 🇸🇬 Singapore to introduce initiatives to improve proficiency in AI use

  • 🏆 Alibaba’s New Model Adds Fuel to China’s AI Race

🤖 Why Everyone’s Talking About China’s New Autonomous AI Agent ManusAI

China’s latest AI breakthrough, ManusAI, is making waves with its autonomous capabilities that go beyond traditional AI agents. Developed by Monica.im, ManusAI is a self-sufficient AI agent capable of independently handling complex, real-world tasks such as travel planning, financial analysis, educational content creation, and insurance comparisons. Unlike conventional AI, it can execute tasks asynchronously in the cloud, allowing users to disconnect while it completes assignments. Its adaptive learning mechanism and personalized memory enable continuous improvement, tailoring outputs to user preferences. ManusAI has outperformed leading AI models, including OpenAI’s ChatGPT, in key benchmarks, solidifying its status as a top-tier AI agent. Its beta release has already generated immense interest, with analysts predicting the AI agent market could reach $130 billion by 2033. In a significant move, Monica.im plans to open-source select components of ManusAI, encouraging collaboration and broader adoption. Positioned as a milestone in AI development, ManusAI is expected to reshape human-machine collaboration and drive a new era of autonomous AI applications.

Microsoft is intensifying its AI efforts to compete directly with OpenAI by developing its own AI reasoning models and exploring alternative AI providers. Reports indicate that Microsoft has built AI models comparable to OpenAI’s o1 and o3-mini and is working on a proprietary family of models called MAI, which may be offered via an API later this year. Tensions between the companies have reportedly grown, with OpenAI declining to share details about its models with Microsoft, despite the tech giant’s $14 billion investment in OpenAI. To diversify its AI strategy, Microsoft is also testing models from xAI, Meta, Anthropic, and DeepSeek as potential alternatives to OpenAI’s technology for products like Copilot. Additionally, Microsoft has recruited Mustafa Suleyman, co-founder of DeepMind and Inflection, to lead its AI initiatives, signaling a strategic push for more independence in AI development.

Recent analysis suggests that AI reaching AGI could happen as early as 2026, a significant acceleration from previous estimates that placed it closer to 2060. The study, which compiled over 8,600 predictions from AI experts, found that more recent forecasts consistently bring AGI’s arrival closer, with some researchers, including Anthropic CEO Dario Amodei, expecting it within the next two years. The rapid development of transformer-based AI models and potential breakthroughs in quantum computing are cited as key drivers. However, skepticism remains, with AI pioneers like Yann LeCun arguing that current architectures are insufficient for human-level intelligence and that AGI’s feasibility is misunderstood. While some see AGI as inevitable due to increasing computing power, others stress the need for fundamental scientific advancements before such systems can truly match human cognition.

🇸🇬 Singapore to introduce initiatives to improve proficiency in AI use

Singapore is launching a series of initiatives to enhance AI proficiency among individuals and businesses, focusing on productivity and safe usage. The Ministry of Digital Development announced updates to the Infocomm Media Development Authority’s Digital Skills for Life framework, introducing generative AI education to equip citizens with essential digital competencies. The SkillsFuture for Digital Workplace 2.0 program will also expand to include AI training, ensuring workers across all sectors, regardless of technical background, can integrate AI tools like ChatGPT and CoPilot into their workflows. Additionally, a generative AI playbook is being developed in collaboration with industry leaders, including Amazon Web Services and Microsoft, to help enterprises harness AI effectively while managing risks. Acknowledging concerns about job displacement, Singapore’s government emphasizes AI as a tool for economic competitiveness, aiming to create a workforce that is confident and adept in AI usage.

Alibaba has launched QwQ-32B, a new AI reasoning model that has fueled China’s AI race and boosted Alibaba’s stock by 8%. While not as powerful as OpenAI’s latest models, QwQ-32B competes closely with DeepSeek’s R1, offering high efficiency with lower computational demands. This release underscores China’s accelerating AI development despite US chip restrictions, with local companies focusing on more efficient architectures. The model is part of a broader effort by Chinese firms like Tencent and DeepSeek to push towards artificial general intelligence (AGI), an area where the US and China are competing for technological supremacy. QwQ-32B adopts a novel “reasoning model” approach, emphasizing deeper computational reflection rather than just scaling model size. Released as open weight, the model can run on consumer hardware, reflecting a trend toward more accessible AI tools. As China rapidly deploys AI across government and industry, the race for AGI continues to intensify, with Alibaba positioning itself as a key player in the evolving landscape.

🛠️ AI tools updates

Google is expanding its AI-powered search capabilities with the launch of Gemini 2.0 for AI Overviews and the introduction of an experimental AI Mode in Google Search. AI Overviews, now used by over a billion people, will provide improved responses for complex queries, starting with coding, advanced math, and multimodal questions. AI Mode, an opt-in experiment in Google Labs, enhances search with deeper reasoning and real-time information retrieval, allowing users to ask intricate, multi-step questions and receive detailed responses with web links. Unlike traditional search, AI Mode employs a “query fan-out” technique, issuing multiple related searches simultaneously across various sources to provide comprehensive results. This integration of Gemini 2.0 with Google’s knowledge systems aims to make search more intuitive and efficient. Google is initially rolling out AI Mode to Google One AI Premium subscribers, with ongoing refinements based on user feedback, including adding more visuals and improving factual accuracy.

💵 Venture Capital updates

China has launched a $138 billion government-backed venture fund to accelerate advancements in quantum computing, AI, semiconductors, and renewable energy, positioning itself as a global leader in emerging technologies. Structured as a public-private partnership, the fund will take a high-risk, long-term investment approach to drive innovation and commercialization. This move aligns with China’s broader strategy of self-reliance in critical technologies, particularly as US and European nations intensify investments in quantum research. Chinese institutions like USTC, Baidu, and Alibaba have already made significant progress in quantum computing, with breakthroughs in quantum communication and superconducting quantum processors. The fund underscores China’s commitment to surpassing Western competitors by scaling cutting-edge research into practical applications. With investments dwarfing similar initiatives in the US and Europe, China’s push could significantly reshape the future landscape of quantum technology and AI-driven industries.

AI investment hit a record $100 billion in 2024, representing one-third of all global startup funding, with the San Francisco Bay Area leading the surge, capturing 60% of worldwide AI investments. Despite the dominance of major foundation model companies like OpenAI and Anthropic, most funding flowed into AI infrastructure, applied AI, and industries like autonomous driving, healthcare, robotics, and cybersecurity. China’s AI sector also made waves, with DeepSeek’s R1 model gaining traction in the US market. Crunchbase’s AI Prediction Engine forecasts that 30% of AI startups at the HumanX conference could be acquisition targets within a year, signaling a wave of consolidation. While AI funding to female-led startups dropped to $18 billion in 2024, their share of deals has remained stable. With continued investor confidence, AI is no longer just a speculative play—it is now the dominant force driving global venture capital.

🫡 Meme of the day

⭐️ Generative AI image of the day