Manus, a Chinese startup's "agentic" AI platform, has taken the tech world by storm, amassing over 138,000 members on its Discord server and selling invite codes for thousands of dollars on Chinese reseller app Xianyu. However, despite the hype, many early users and experts are questioning whether Manus lives up to its lofty claims.
Manus has been touted as a revolutionary AI tool, capable of performing complex tasks such as drafting research reports, analyzing financial filings, and even programming video games. But, as it turns out, the platform's capabilities may be overstated. According to reports, Manus uses a combination of existing and fine-tuned AI models, including Anthropic's Claude and Alibaba's Qwen, rather than being developed entirely from scratch.
Yichao "Peak" Ji, a research lead for Manus, has made bold claims about the platform's abilities, stating that it outperforms OpenAI's deep research on a popular benchmark for general AI assistants called GAIA. However, early users have reported encountering error messages, endless loops, and inconsistent results when testing Manus. Some have even pointed out that the platform makes mistakes on factual questions and doesn't consistently cite its work.
Personal experiences with Manus have also been underwhelming. When asked to perform simple tasks such as ordering a fried chicken sandwich or booking a flight, the platform failed to deliver. It crashed, provided incomplete information, or simply couldn't complete the tasks. These limitations raise questions about Manus' capabilities and whether it's truly a game-changer in the AI space.
So, why did Manus generate so much hype in the first place? Part of the reason may be the exclusivity created by the scarcity of invites, which fueled a sense of FOMO (fear of missing out) among tech enthusiasts. Chinese media outlets, such as QQ News, have also been quick to tout Manus as an AI breakthrough, which may have contributed to the platform's inflated reputation.
AI influencers on social media have also played a role in spreading misinformation about Manus' capabilities. A widely-shared video, which was later debunked by Ji, showed a desktop program taking action across multiple smartphone apps, leading many to believe that Manus was capable of more than it actually is.
Despite the criticism, Monica, the startup behind Manus, claims that the platform is still in early access and that they're working to scale computing capacity and fix issues as they're reported. However, as it stands, Manus appears to be a case of hype running ahead of technological innovation. While the platform shows promise, it's clear that it still has a long way to go before it can live up to its lofty claims.
In the meantime, experts and early users are urging caution and a more nuanced understanding of Manus' capabilities. As the AI space continues to evolve, it's essential to separate hype from reality and to critically evaluate the capabilities of emerging technologies.