Large language models are highly skilled at various tasks such as coding, essay writing, translation, and research. However, they still struggle with basic tasks in the realm of personal assistants. For instance, ChatGPT or Claude cannot order a burrito or book a train ticket. While OpenAI and Anthropic offer AIs that can interact with your computer, they still lack full functionality. The concept of AI agents that can act independently is not new, and AI has the potential to automate many tasks and save money for tech companies. However, AIs like Devin have had limited success and it is unclear whether Manus, a new Chinese AI agent, has made significant progress. There are also concerns about privacy and security when trusting AIs with personal and financial information.

Source: https://www.vox.com/future-perfect/403896/artificial-intelligence-china-manus-chatgpt-openai-google-privacy