Common Beginner Mistakes
Lesson 2.5 — Common Beginner Mistakes
Estimated reading time: 15 minutes
Everyone makes these mistakes. They're not signs of low intelligence — they're predictable patterns that come from carrying assumptions across from other tools (search engines, word processors, other people) to something genuinely new.
The good news: once you see each mistake, you usually stop making it. This lesson is designed to save you weeks of trial and error.
We're covering eight real mistakes, each with a concrete wrong approach, the better approach, and an explanation of why the mistake is so common.
Mistake 1: Treating It Like a Search Engine
Wrong approach:
"best project management tools 2024"
Better approach:
"I manage a remote team of 6 people spread across three time zones. We're a design agency with ongoing client projects. What project management tools would you recommend, and why? I currently use spreadsheets and it's getting painful."
Why this mistake is so common:
Most of us have used search engines for 20 years. When we have a question, our instinct is to type keywords — short, noun-heavy phrases designed to match pages in a database. That's exactly wrong for AI chat.
AI assistants don't search for pages. They generate responses based on your input. Keywords give them almost nothing to work with. Full sentences with context give them everything.
The magic of AI chat is that it can synthesise information for your specific situation. But it can only do that if it knows your specific situation. "Best project management tools" could be asked by a solo freelancer, a 500-person corporation, or a student group project. Each answer would be completely different. Tell the AI who you are and what you actually need.
Mistake 2: Accepting the First Response Without Pushing Back
Wrong approach: Reading the first response, thinking "that's okay I suppose," and moving on.
Better approach:
"That's a decent start, but the third paragraph isn't quite right for my situation. Can you revise it to [specific change]?"
Or simply:
"Can you try that again but make it shorter and more direct?"
Why this mistake is so common:
In most human interactions, asking someone to redo their work feels rude. We're socialised to accept what we're given rather than ask for a revision. That instinct is completely wrong with AI.
The AI doesn't have feelings. It won't be offended. It will not think less of you. Asking it to revise is exactly what it's there for.
The first response is a draft. Treat it as one. Professional AI users almost never use the first response without some refinement. The iteration — "can you make it more concise / more formal / less jargon-heavy / more specific to my situation" — is where the real value gets unlocked.
Mistake 3: Asking Vague Questions
Wrong approach:
"Help me with my presentation."
Better approach:
"I'm giving a 10-minute presentation to my company's board of directors next week about why we should invest in upgrading our customer service software. I have three main arguments but the second one feels weak. Here it is: [text]. Can you help me strengthen it with more compelling logic and maybe a relevant statistic or analogy?"
Why this mistake is so common:
When we ask a friend for help, context is often implied — they know us, they know our situation. We've learned to be economical with our explanations.
AI assistants know nothing about you except what you tell them in the conversation. "Help me with my presentation" forces the AI to guess: What's the presentation about? What kind of help? What's wrong with it? The vaguer your question, the more generic the answer.
A useful test: read your question back to yourself and ask, "If I sent this to a smart stranger on the internet, would they have enough to help me?" If not, add context.
Mistake 4: Trusting Everything It Says
Wrong approach: Taking an AI-generated fact, statistic, or claim at face value and repeating it to others — in emails, reports, conversations — without checking.
Better approach: Using AI-generated content as a starting point, and verifying any specific facts, statistics, names, dates, or quotes from a primary source before using them.
Why this mistake is so common:
AI responses sound authoritative. They're well-structured, confidently written, and use the tone of someone who knows what they're talking about. The problem is that AI models can "hallucinate" — generate plausible-sounding but incorrect information — and they don't always flag it when they do.
A classic example: ask an AI to cite sources, and it will sometimes generate citations that look real but lead nowhere. The journal name is real, the author is real, but the specific paper doesn't exist. The AI is not lying intentionally — it's pattern-matching to what a citation looks like, not checking a database.
This doesn't mean AI is useless for research. It means you use it for the synthesis and reasoning, and you verify the specific facts independently. Think of AI as a very well-read colleague who sometimes misremembers details — valuable, but not the last word.
Key takeaway: The higher the stakes, the more you verify. For a casual email, AI content is usually fine. For a business report, a presentation to clients, or anything medical or legal, fact-check before you use.
Mistake 5: Writing Paragraph-Long Prompts When Structure Would Help
Wrong approach:
"I need help writing a job description for a new marketing hire for my company we're a small tech startup about 25 people in Dublin we've been operating about three years we're looking for someone who can run social media and also do some content writing and maybe help with events and this person will report to me I'm the CEO and they'll work closely with our sales team the salary is around €40k to start we want someone with about two years experience but we might be flexible if someone was really good but definitely needs to be able to use Canva and maybe some video editing would be a bonus."
Better approach:
"Write a job description for a marketing hire at a 25-person tech startup in Dublin. Here are the details:
- Role: Marketing Coordinator
- Responsibilities: Social media, content writing, events support
- Reports to: CEO
- Works with: Sales team
- Salary: €40-45k
- Experience required: 2+ years
- Must-have skills: Canva, copywriting
- Nice-to-have: Basic video editing
Tone should be friendly and energetic — we're not a corporate company."
Why this mistake is so common:
When we're in a hurry, we dump our thoughts out in a stream of consciousness. In everyday communication, context fills in the gaps. With AI, a structured list is almost always better than a dense paragraph for requests involving multiple specific details. It's easier for the AI to parse, harder to accidentally skip something, and usually produces a better result.
Mistake 6: Starting Over When You Should Follow Up
Wrong approach: Getting a response that's 80% right, closing the chat, opening a new one, and starting from scratch.
Better approach: Continuing the conversation: "Good start. Can you adjust [specific part]? Keep everything else the same."
Why this mistake is so common:
New users often don't realise that follow-up messages maintain all the context of the conversation. They think they need a "fresh start" to get a different answer. In fact, the opposite is true — a follow-up in the same conversation is far more efficient because the AI still has all the context. Starting a new chat means re-explaining everything.
The only time you should start a new conversation is when the topic genuinely changes. If you're refining an output, stay in the same chat.
Mistake 7: Asking for Information Instead of Help
Wrong approach:
"What are some tips for having difficult conversations at work?"
Better approach:
"I need to tell my employee that they're not meeting performance standards and might be put on a performance improvement plan. I've been avoiding this conversation for two months. Can you help me prepare — what should I say to open the conversation, how do I stay calm if they get defensive, and is there anything I should avoid doing?"
Why this mistake is so common:
We're used to information retrieval — search engines give us lists of tips, articles, explainers. We bring that pattern to AI and ask for general information.
But AI assistants are more useful as interactive advisors than as encyclopedias. The difference is the word "help." Instead of "what are tips for X," try "help me do X" and give it the specific situation. You'll get something immediately applicable rather than a listicle of generic advice you've probably read before.
Mistake 8: Using AI for Things It Consistently Gets Wrong
Wrong approach:
- Asking for today's news or current events (if the AI doesn't have web access)
- Asking it to calculate complex numbers precisely
- Asking for specific legal or medical advice and treating it as authoritative
- Asking for real-time prices, stock values, or availability
Better approach: Knowing the limits and routing accordingly:
- Current events → Perplexity AI, Gemini with web access, or just... the news
- Complex calculation → a calculator or Excel
- Legal/medical specifics → a qualified professional
- Prices/availability → the business's website
Why this mistake is so common:
AI assistants are so capable across so many domains that it's tempting to use them for everything. The failure modes aren't always obvious — sometimes the AI will give a confident wrong answer rather than admitting it doesn't know.
The pattern is: AI is excellent at reasoning, writing, synthesis, explanation, and creativity. It's unreliable for real-time data, precise arithmetic, and professional advice that requires accountability. Keep this distinction in your head and you'll know when to use it and when to look elsewhere.
The Mistake Underneath All the Mistakes
If there's one meta-mistake that underlies most of the above, it's this: treating AI like a passive tool rather than an active collaborator.
Search engines are passive. You throw in a query, you get back links. There's no back-and-forth.
AI assistants are active. They respond to context, nuance, feedback, and revision. The more you engage with them — clarifying, pushing back, refining — the better the output gets.
Users who get the most from AI aren't necessarily the ones who write the best initial prompts. They're the ones who treat the conversation as a process, not a transaction.
Key takeaway: The best way to fix all of these mistakes at once is to approach every AI conversation as a collaboration. You're not entering a command and waiting for output. You're working with a tool that responds to how you engage with it. The more clearly you communicate, and the more you refine and iterate, the better your results.
Quick Reference: Mistakes and Fixes
| Mistake | Wrong approach | Better approach |
|---|---|---|
| Search engine thinking | Keyword prompts | Full sentences with context |
| Accepting first response | Move on immediately | Iterate and refine |
| Vague questions | "Help me with this" | Specific situation + specific ask |
| Over-trusting facts | Use without checking | Verify specific claims |
| Paragraph dumps | Wall of text | Structured lists for complex requests |
| Starting over unnecessarily | New chat for every tweak | Follow up in the same conversation |
| Asking for info, not help | "What are tips for X?" | "Help me do X, here's my situation" |
| Using AI for its weak spots | Current events, precise math | Route to the right tool |
Up next: Lesson 2.6 gives you a practical guide to choosing the right AI tool for different tasks — with a comparison table you can bookmark and refer back to.