10 Things I Learnt at Google About Using AI Better
I recently attended an amazing AI training day run by the folk at Google Digital Garage. It covered marketing, social media and growing a business online. Here are ten of my biggest takeaways.
1. AI doesn't have to be complicated
One of the simplest ideas from the day was to think about AI in terms of four things it can help us do: summarise, learn, create and automate.
That could mean summarising a long report, learning about a new subject, getting help with a first draft or taking some of the repetitive work out of your day. You don’t need a complicated AI strategy to benefit from it — start with something you already need to do.
2. Better prompts give you much better answers
Google shared a useful prompt framework: Task, Context, References and Evaluate
In other words, tell AI what you want it to do, give it enough background to understand the task, provide useful examples or information, look critically at what it gives you and then improve your instructions if necessary.
The difference between “Give me some marketing ideas” and a detailed prompt explaining your business, audience, goal and budget can be enormous.
3. Context really matters
AI can’t read your mind. If you want a useful response, it needs to understand your situation.
Tell it about your business, audience, goals, challenges and what you’ve already tried. You can also give it existing documents, examples of your writing or other relevant information to work from. The more useful context you provide, the less generic the response is likely to be.
4. Don't stop after the first answer
This is one of the easiest ways to get more from AI. Your first prompt doesn’t have to produce the perfect answer — it can simply start the conversation.
Ask it to make something simpler, give you another approach, challenge your idea or tell you what you’ve overlooked. One of my favourite prompts is simply: “Ask me the questions you need to ask before answering.”
Often, the second or third response is far more useful than the first.
5. Give AI examples of how you like to work
If you’re using AI to create content, you don’t have to start from scratch every time. Give it examples of your existing work, your website copy, brand guidelines, preferred formats or previous content you particularly like.
You can ask AI to identify the characteristics of your style before creating anything new. It still needs your input and editing, but giving it good reference material can make the starting point considerably better.
6. AI isn't just for creating content
There’s so much conversation about using AI to write social posts, blogs and emails that it’s easy to overlook one of its most useful roles, helping us evaluate our ideas.
Ask it to find weaknesses in an idea, give you another perspective, identify assumptions you’re making or argue against a decision you’re considering. You can even ask it to take the perspective of a customer and tell you what questions or concerns they might have.
7. Those little time savings really add up
One statistic from the training really stood out to me. Participants in Google’s AI Works research estimated that using AI was saving them the equivalent of around 122 hours a year.
That’s more than 15 eight-hour working days.
It isn’t necessarily about one enormous piece of automation. Ten minutes saved summarising a document, half an hour getting a first draft started or an hour saved on research all begin to add up. For a small business owner, getting even a fraction of that time back could make a real difference.
8. Sometimes we need 'permission to prompt'
This was one of the more interesting ideas from the AI Works research. Some people weren’t avoiding AI because they couldn’t use it; they simply weren’t sure whether they should.
The training gave people confidence that using AI to help with their work was legitimate. I think that’s an important mindset shift. Using AI doesn’t mean you aren’t capable of doing something yourself — it’s another tool that can help you do the job.
The important skill is knowing when to use it and when your own experience and judgement need to take over.
9. The best way to learn AI is to use it
You could spend hours reading about AI, but one of the best ways to become more confident is simply to experiment with it.
Start noticing the moments when you’re thinking, “This is taking ages,” “I don’t know where to start,” or “I need some ideas.” Those are perfect opportunities to ask whether AI could help.
The more you experiment, the more uses you start to discover.
10. AI should help us think, not stop us thinking
AI can be incredibly useful, but it can also get things wrong. It can misunderstand a question, provide inaccurate information and sometimes sound remarkably confident while doing it.
We still need to check important facts and sources, protect confidential information, edit what AI produces and apply our own knowledge and judgement.
For me, that’s probably the biggest lesson from the day. The goal isn’t to use AI for everything. It’s to recognise where it can make what we’re already doing easier, faster or better.
You don’t need to become an AI expert to start. Pick one real task, give AI plenty of context, see what it produces and then keep the conversation going.
You might be surprised by just how useful it becomes.