The Best Way for Everyday People to Learn AI: Start with Problems, Not Anxiety

Caught in the wave of AI, countless everyday people find themselves in a frustrating paradox: afraid of falling behind if they don't learn AI, yet immediately put off by a wall of technical jargon and complex tutorials. Bilibili creator Daliang offers a more accessible learning path in his AI tutorial videos — no need to panic, just start by solving concrete problems and work your way deeper from there. This article systematically breaks down that methodology.

The Two Dominant AI Learning Schools — and Why Both Fall Short

Most advice on learning AI falls into one of two extreme camps. The first is the academic route: start with machine learning and deep learning, grind through Transformers, then tackle large models, prompt engineering, Workflow, Agent, MCP, Skills… a rigorous, layered progression. The second is the just-do-it approach: AI doesn't require understanding principles at all — the fastest way is to dive straight in.

But both approaches have serious flaws. The academic route overwhelms beginners with complex prompts and theory, and most people lose interest fast. The just-do-it approach leads you to download a bunch of AI tools, only to find that everything you produce is garbage and AI "doesn't actually seem that useful."

The real problem isn't which path to choose — it's that you haven't clearly defined why you're learning in the first place.

Step One: Define Your Purpose

A key insight from the video is this: the first thing a beginner should do is not figure out how to learn AI, but clarify their goal. People learning AI generally fall into three categories:

• Type 1: Efficiency Seekers. Want to use AI to boost productivity at work and in daily life — writing content, building spreadsheets, researching, organizing files, cutting down on repetitive tasks.
• Type 2: Independent Builders and Creators. Want to become indie developers or content creators — building basic AI apps, creating workflows others can use, or making comics with AI.
• Type 3: Professional Practitioners. Want to work on model training, algorithm research, or join an AI company.

The creator notes that 90% of people who come across this kind of content fall into the first two categories. So the entire learning path is designed with that 90% in mind. This framing matters — it means you don't need to wrestle with the intimidating math and research papers lurking beneath the surface.

The Core Principle: Focus on Your Problem, Not the Technology

The first and most important piece of advice: don't focus on how to learn AI — focus on the problem you're trying to solve.

When you have no specific problem in mind, you can open an AI tool and have no idea what to do with it. But when you have a concrete task — say, the tedious work at your job you'd rather not do — your motivation to use AI skyrockets and your goal becomes clear. At that point, just open your AI tool and get started.

But the author cautions: don't just wing it. Because you'll quickly swing from "AI seems like it can do anything" to "AI can't seem to do anything right." To get output you're actually happy with, you need to work with intention.

The simplest approach: search online for what others have shared about your specific problem. If you want to make a presentation with AI, search for "how to make a PPT with AI" and follow someone else's experience — you'll get twice the results with half the effort. If you can't find anything, use AI itself to guide you on using AI.

One often-overlooked but critically important point: you need a computer. Your phone is fine for consuming information — scrolling, reading — but real hands-on work requires a computer. Only a computer is a productivity tool.

The Key Leap: Understanding the Underlying Logic

The real turning point in using AI well is whether you can understand the underlying logic and apply it broadly while working through problems. The author illustrates this with two highly relatable examples.

From "AI-sounding" to Concrete References: The Logic Behind Prompts

Many people use AI to write something, feel it sounds "too AI-ish," and then ask it to "write more vividly, less like AI." But AI has no idea what you mean by "vivid" or "AI-ish."

Then you find someone online who suggests: tell the AI to "write in the tone of two close friends chatting," and suddenly the output is much better. The crucial insight here — don't just memorize that specific prompt; understand why it works.

The reason: "vivid" is an abstract description — AI can't read your mind. But giving a concrete reference gives AI something to work with. Once you understand this, you can apply it everywhere: "in the authoritative tone of a manager addressing employees," "in the style of Zhang Ailing or Lu Xun" — any specific reference AI can grasp.

The Same Logic Applies to Aesthetics: Specifics Beat Vague Requests

If you ask AI to build a website and it comes out ugly, telling it to "make it look better" won't help. But if you say "make it look like Apple's website — that minimal aesthetic" — AI immediately gets it, because it knows what Apple's website looks like.

The core logic: the more specific and visually concrete your description, the more the output matches your intent. What you should internalize is this logic, not any particular prompt. As you solve more and more problems, you'll build a growing understanding of how AI works — and at that point, you've truly gotten started.

The Next Level: Tool Combinations and Automated Workflows

Once you've got the basics down, start exploring different AI tools. Because you already understand the underlying logic, picking up an unfamiliar tool won't leave you lost.

Why try different tools? Because different AI tools and models each have their strengths. To consistently use AI for real work, you need the right set of tools for the job.

As you get more skilled, you'll realize that most tasks require a whole toolkit working together. Making a short video might involve: a large language model to write the script, an audio model to generate voiceover, video and image models to create visuals, and a music model to produce background tracks. Which tools fit your workflow best is something only you can discover through hands-on testing — until you settle on a reliable combination. At this stage, you've genuinely arrived.

Go one step further, and you'll start building combined workflows. By designing Workflows and chaining API calls, you can automate your most common tasks — entering the realm of AI automation. These eventually become Skills, or automated buttons inside an Agent, and you'll build them with ease.

From AI User to Indie Developer

Once you have a solid grasp of the underlying principles, understand what different models and Agents can do, and can build your own workflows and Skills — you're already equipped to start building small products and tools.

You can try building Skills, apps, mini-programs, or web services that solve problems for more people. The first things you make will probably be buggy and barely functional — and that's completely fine. Those new problems become your motivation to go deeper. And if you do get something working, it could turn into real income. Finding unmet needs and pain points in the market, then building solutions for them — that's the first step on the indie developer path.

At this point, you've fully left behind your beginner status and become an AI enthusiast or practitioner. Where you go from here is entirely up to your own interests.

Information Sources and Mindset: Breaking Free from AI Anxiety

In an era of rapid AI development, the ability to find good information is especially valuable. The author recommends that beginners follow domestic AI community platforms for news and sharing, while more advanced learners should regularly check GitHub and similar technical platforms.

Finally, a word on mindset. AI is evolving at a breakneck pace, and there will always be more to learn than you can keep up with — but there's absolutely no need to feel anxious. Coming back to concrete problems, and using AI to actually solve them, is where AI's real value lies. As the author quotes: "Every inch forward brings its own joy." Knowledge may be infinite, but every step you take yields something real and tangible. That, perhaps, is the healthiest attitude any everyday person can bring to the AI wave.