When to use AI

When Use AI

There are several types of work where AI can be particularly useful, given the current capabilities and limitations of LLMs.

1. Brainstorming and idea generation When quantity leads to quality. Most people stop after generating just a few ideas because they become exhausted, but AI can provide hundreds without repeating itself.

Give me 20 different names for my dog walking business, be creative and consider different tones - professional, playful, and memorable

2. Work where you're an expert and can assess quality This can involve complicated work, but it relies on your expertise to determine whether the AI is providing valuable outputs. For example, the new AI models can solve some PhD-level problems, but it's hard to know whether answers are useful without being an expert yourself.

3. Summarizing and translating between formats AI excels at taking information and presenting it differently—summarizing large amounts of content, converting data between formats, or adapting the same information for different audiences.

Turn these meeting notes into: 1) a 3-bullet executive summary, 2) action items with owners, 3) a casual team update email

4. Getting unstuck and moving forward When little obstacles block your way. Instead of walking away stuck on a sentence or idea, AI can provide the push you need to keep moving.

Give me 15 ways to end this sentence: "The main challenge with remote work is..." 
Make them varied - some serious, some creative

5. Learning companion for complex topics AI allows you to ask infinite questions about material you're studying, making it an excellent companion when reading or learning new subjects.

I'm reading this research paper about renewable energy. Ask me questions one at a time to test and deepen my understanding of the key concepts

6. Multiple perspectives and feedback Get a variety of viewpoints on your work—from hostile critics to friendly supporters to neutral experts. Perfect for preparing for presentations or understanding how different audiences might react.

Review my business proposal as if you're three different people:
1) A supportive colleague who wants me to succeed
2) A skeptical investor looking for risks
3) A potential customer evaluating whether to buy

7. Entrepreneurial work requiring broad expertise When you need to stretch across many disciplines and the alternative to a good-enough partner is not being able to act at all. AI can help with documents, demos, and approaches outside your experience.

8. Automating routine work Work that serves no useful purpose or has become disconnected from its original intent. Reports no one reads, standardized communications, or repetitive formatting tasks.

9. Second opinions and data analysis Give AI access to the same data and see if it reaches similar conclusions. Useful for checking your reasoning or finding patterns you might have missed.

10. Tasks AI genuinely excels at This includes many types of coding, format conversion, and specific technical tasks where AI often outperforms humans. This category continues to expand rapidly.

Don't Use AI

Beyond obvious scenarios like illegal purposes or high-stakes situations where errors could be catastrophic, here are 5 times to avoid AI:

1. When you need to learn and synthesize new ideas Asking for a summary isn't the same as reading for yourself. Asking AI to solve problems for you isn't effective learning, even if it feels like it should be. To truly learn something new, you need to do the reading and thinking yourself, though AI can still help with parts of the learning process.

2. When very high accuracy is required AI errors (hallucinations) are particularly dangerous because they're very plausible and hard to spot. Research shows people often "fall asleep at the wheel" and stop checking AI output carefully. Hallucinations can be reduced but not eliminated.

3. When you don't understand how AI fails AI doesn't fail like humans do. Beyond hallucinations, AI might try to persuade you it's right when it's wrong, or become overly agreeable and support your incorrect answers. You need experience with AI to understand these failure modes.

4. When the effort is the point In many areas, people need to struggle with a topic to succeed. Writers rewrite the same page multiple times, academics revisit theories repeatedly. By shortcutting that struggle, you may lose the vital "aha" moment. Friction often helps facts stick in memory.

5. When AI is simply bad at the task AI is surprisingly bad at things you'd expect it to handle (like counting the number of r's in "strawberry") and surprisingly good at unexpected tasks (like writing a Shakespearean sonnet about that counting difficulty). There's no general manual for AI's abilities, which constantly evolve. Trial and error is essential.

Cautionary example - Don't do this:

AI, write my performance review for my direct report Sarah. She's been struggling but I don't want to deal with the difficult conversation.

This removes the human element that makes performance reviews meaningful and avoids necessary management work.

Mindset Needed: It's not Google

The biggest hurdle isn't learning prompt techniques—it's changing how you think about AI entirely. Most people treat AI like a fancy search engine, but that's like using a race car to deliver pizza. You'll get there, but you're missing the point.

Google vs. AI: Choose the right tool

Use Google when:

• You need one specific fact and want peace of mind
• You want the most recent information
• You need to verify something is actually true

Use AI when:

• You want a fact WITH an explanation
• You have multiple related questions
• You need help thinking through a complex topic
• You want to explore ideas and possibilities

Example: The difference between searching and instructing

Instead of searching "best vacation spots with kids," you can say:

Act like a travel agent. Plan a 5-day family trip in August with kids under 10, $2K budget, flying from Seattle. Include rainy day activities.

There's a big difference. You're not searching—you're instructing.

Develop an experimental mindset

Think of AI as a smart colleague who's eager to help but needs direction. The best AI users approach it with curiosity, not specific expectations. They're willing to:

• Try different approaches when the first one doesn't work
• Build on responses rather than accepting them as final
• Treat conversations as collaborative problem-solving sessions
• Learn from what works and what doesn't

The goal isn't to become a prompt engineering expert—it's to develop a partnership mindset where you and AI work together to solve problems, generate ideas, and explore possibilities you couldn't reach alone.