About Getting Started with AI (Chinese Users)

1. For ordinary users, just use HaoDouBao in daily life, and master these three points:
① Functionality: Use Baidu Search to verify information, use Photoshop to draw images, use Seedance to generate videos as a director
② Models: Prefer to use the Thinking Version and Expert Version, focusing on the strongest model to generate the best results
③ Prompt words: Learn simple logical structures and meta-prompt techniques (AI-generated prompts)
2. For office workers, who need to work based on their own data, it’s recommended to use IMA:
① Basic tasks: Create a private knowledge base, upload non-confidential data, organize into folders and tag
② Basic skills: Work based on tags or within folders, using precise knowledge in the dialogue box
③ Key points: Use the ds 3.2 Thinking model, make good use of notes and synchronize with the knowledge base, and prompts are equally important
3. For geeks, you must try the popular openclaw:
① Deploy Cloud Shrimp: Set up on cloud servers (geeks might look down on this), beginners can try at low cost (mainly to learn deployment process)
② Domestic imitation shrimp: NetEase’s Lobster, Fu Sheng’s Yuanqi AI, including many on GitHub
③ Native Lobster: Still OpenClaw, as a beginner still stepping into the pit, I won’t give specific advice
AI capability is a “dual helix structure of AI evolution + human learning,” meaning: AI evolves rapidly, so you need to understand it, then interact with it to make it part of your ability.
However, any technology must serve needs. AI cannot do demand insight and decision-making for you. These soft skills are even more important than AI capabilities. That’s why some users in lower-tier markets are better than many AI geeks—they are closer to the needs and can even make quick money because of that.
Developers of software may not necessarily profit from selling or using software.
In short, master this principle of human-AI collaboration—
Use AI with 7 points of capability to solve N number of 3-point practical problems and quickly monetize; rather than pursuing 9+ points of AI capability while ignoring a 3-point need, turning into a poor person who understands AI (which seems quite difficult).