China Introduces Manus: The New AI Assistant Taking the World by Storm
Artificial intelligence world has been advancing rapidly with successive developments in recent years. China’s latest bombshell is a new AI agent named Manus. According to its developers, Manus is not just a chatbot, but a completely autonomous agent that requires very little human intervention. But is it truly a revolutionary step towards Artificial General Intelligence (AGI), or just a slightly more advanced version of existing tools? Let’s delve into Manus’s capabilities and the debates surrounding it together.
Manus is an artificial intelligence tool that can produce results on its own with minimal guidance from humans, according to the development team.
Lead scientist of the project, Yichao “Peak” Ji, mentions that unlike other AI models, Manus not only generates ideas but can autonomously complete the entire process.
Ji states that when Manus receives a command from a user, it not only provides a response but also plans the next steps automatically using its memory. Derived from the Latin phrase meaning ‘Mind and Hand,’ Manus is described by researchers as the “new paradigm of human-machine collaboration.”
Currently, Manus is only accessible by invitation.
It has been reported that access codes are being sold for thousands of dollars on the Chinese second-hand sales platform Xianyu. According to Manus’s website, the AI tool can perform various real-world tasks. So, what are these tasks? Let’s see:
-
Creating personalized travel plans.
-
Conducting real estate research and providing price-performance analyses.
-
Analyzing correlations between stocks.
-
Scanning resumes, ranking candidates, and preparing evaluation reports.
In a demo demonstration, Ji showcased Manus analyzing 15 different resumes, sorting the most suitable candidates, and automatically transferring this information to a spreadsheet.
Even more impressively, Manus can be trained to directly prepare a table for similar tasks in the next assignment. Another significant feature is Manus’s cloud-based and asynchronous operation. Even if the user shuts down their computer after assigning a task, Manus continues to complete the process and can receive new instructions during the operation.
Ji claims that Manus reached the same level as OpenAI’s Deep Research tool in GAIA Benchmark tests, delivering the best results. Additionally, it is stated that Manus is currently fulfilling tasks on freelance platforms such as Upwork and Fiverr.
Manus was developed using data from open-source communities.
Ji revealed that Manus is a multi-agent system that currently works with different AI models, and the team plans to share some models as open-source later this year.
Manus’s introduction sparked various reactions in the AI world. What did the experts say? Let’s find out…
AI researcher Dean Ball from George Mason University emphasized that Manus should not be compared to DeepSeek, stating that while DeepSeek was a copy of existing models in the US, Manus is truly pushing the boundaries.
AI researcher Derya Unutmaz tested Manus against OpenAI’s Deep Research tool side by side, revealing that Manus couldn’t complete the task even in 50 minutes while Deep Research finished it in just 15 minutes.
Alexander Doria, the co-founder of the French AI lab Plelas, admired Manus’s interface but pointed out that it actually still works like a workflow management system, saying, “Ultimately, it still operates as a workflow management system.”
While Manus’s capabilities are impressive, the claim of being the “first general AI agent” seems debatable at the moment.
Yes, Manus operates on the cloud, has memory, and can make autonomous decisions. However, some experts still do not see it as radically different from traditional AI tools.
Considering China’s rise in the field of artificial intelligence, it seems that more time is needed to determine whether Manus is a significant step towards AGI.
