Chapter 18: AI in Everyday Life and Creativity
Uses that become part of everyday life
You have already seen fifty possibilities. In practice, most people settle into four or five and abandon the rest. This chapter is about which ones tend to survive, and about the more complicated question of what AI does to creative work.
The domestic uses that last share a common shape: recurring, moderately annoying, and low-stakes if slightly wrong.
Meal planning against what you have. Not recipes. You can find recipes anywhere. The value is working backward from your actual refrigerator, your actual constraints, and the four dinners you need this week.
Turning decisions into comparisons. Two appliances, two schools, two insurance policies. The tool can build the comparison table you would have created if you had an hour and surface criteria you had not considered. You still make the decision, using current specifications checked at the source.
Understanding paperwork. Bills, statements, notices, benefit letters, insurance explanations. Not to replace professional advice, but to arrive at the conversation knowing what to ask.
As a real estate investor, I recently had a commercial property under contract. Commercial transactions often involve long closing periods and detailed agreements. When the buyer sent a revised contract and said that only the name of the purchasing entity and a few dates had changed, I uploaded both versions to ChatGPT and asked it to identify the differences. Within moments, it confirmed the changes the buyer had described and found several spelling errors that I could flag for my agent. Professional review still has its place, but AI made the first comparison fast, practical, and useful.
Correspondence you keep postponing. The complaint, the difficult reply, the letter to a relative. A rough draft on the screen can dissolve the resistance that kept the task undone for three weeks.
I frequently write to government officials and court staff. ChatGPT helps me find a respectful, professional tone and make the request easy to understand. It is also useful when I prepare public-records requests: I can provide the agency, the records I need, and a few relevant facts, then ask for a clear first draft. I review the language before sending it, but the tool gives me a strong starting point in a fraction of the time.
Home and vehicle troubleshooting. Describe the symptom in detail and ask for a ranked list of likely causes and sensible checks. Used as preparation for a repair conversation, this can save time and help you ask better questions; confirm safety-critical advice with a qualified professional.
Trip planning. Realistic day-by-day itineraries built around your actual pace and interests. Always verify hours, prices, and whether places still exist.
Learning something for its own sake
One of the quieter pleasures of these tools is having an inexhaustibly patient explainer for things nobody is paying you to know.
The bird you saw. Why bread does that. What is actually happening in a piece of music. The history of the street you live on. What that legal phrase in the news means.
Two things make this work better than most people realize. Ask it to explain at the level you are actually at, and adjust until it lands: simpler, no, more technical than that, use an analogy from cooking. And follow the thread; the value compounds when you keep asking why rather than accepting the first explanation.
Structured learning receives fuller treatment in the next chapter. Here it is enough to say that if you have ever stopped wondering about something because finding out was too much trouble, that barrier has largely gone.
Creative work: the honest conversation
This is where I want to be careful, because most writing on the subject is either promotional or apocalyptic and neither is much use to someone deciding how to spend Saturday.
Start with a distinction that resolves most of the confusion.
Some creative work is done for the artifact. Some is done for the doing.
If you need a birthday poem, a party invitation, a logo for the school fundraiser, or a serviceable short story to read aloud to a seven-year-old, the artifact is the point. Use whatever produces a good artifact. Nobody is diminished.
If you write because writing is how you think, paint because painting is the part of the week that belongs to you, or play music because playing is the point, then handing the making to a machine does not save you effort. It removes the thing you wanted.
Nobody can make that call for you, and it differs by activity within one person. I know people who use AI freely for prose and refuse to let it near their photographs.
Where AI helps without taking the thing
If the making matters to you, the useful principle is: work at the edges, not in the middle.
Before creating, use AI for research, constraints, and breaking the blank page. You might request twenty first lines simply to reject them; the act of rejection clarifies what you want. Or ask for five formal restrictions that would make the work harder in an interesting way.
After: critique, structural diagnosis, the reader’s-eye view. Where does this lose momentum? What is the reader expecting at this point that I’m not delivering? What is this actually about, based only on what’s on the page? That last question is frequently startling.
Around: the parts that are not the art. Formatting, submission letters, synopses, artist statements, tags and descriptions, the tedium that surrounds creative work and stops people finishing.
Not in the middle, if the middle is why you are there.
Writing with AI without losing your voice
For those using it on prose directly, three practical cautions.
The homogenization pull is real. Ask a model to improve your writing and it will tend to move it toward a competent average: smoother, more conventional, less yours. The idiosyncrasies it sands off are frequently the reason anyone reads you.
Rewrite is more dangerous than draft. A draft you rewrite yourself stays yours. A passage of your own writing “polished” by a model often comes back subtly deadened, and because each individual change looks like an improvement, the loss is hard to see until you compare the whole.
Use style capture rather than description. Feed it your own best writing and ask it to match, rather than asking for “engaging” or “professional.” Some tools now provide a dedicated styles feature for this purpose. Even then, treat the output as a suggestion to edit rather than text to accept.
A test I find useful: read the AI-assisted version aloud. Voice survives in the ear long after it has become invisible on the page.
Images
Most major tools now generate images from a description. A few practical notes.
Be specific about the things that determine the result: subject, style, composition, lighting, mood, what to exclude. A photograph of an empty diner at dawn, warm light through venetian blinds, seen from a corner booth produces something. A nice picture of a diner does not.
Iterate rather than restart. Change one thing at a time.
Refinement is the current weak point. Be very specific about what you want changed, and tell the tool to leave everything else alone.
Know the current weak points: hands, text within images, consistent characters across multiple images, and precise spatial relationships. These improve steadily; check rather than assume.
And understand that the rights questions are unresolved and genuinely complicated. Whether you may use an AI-generated image commercially, whether it can be copyrighted, and what the status of models trained on artists’ work will turn out to be are live legal questions. Bonus Chapter A covers what is known and what is not. For a birthday card, none of this matters. For a business logo, it matters a great deal.
A word on companionship
Some people find these tools good company: a patient listener at two in the morning, something to think out loud with, a conversation available when no one else is.
That is real, and for isolated people it can be genuinely valuable. I would not dismiss it.
I would ask you to notice two things. The warmth is a product of training on human writing and of preference tuning toward agreeableness — it is not regard, and it does not persist between conversations except as stored notes. And a companion that never tires, never disagrees seriously, and never has needs of its own is easier than human company in a way that can quietly make human company feel harder.
If you notice yourself preferring it, that is worth taking seriously rather than dismissing. Chapter 25 returns to this alongside the question of what you are disclosing, and Chapter 19 addresses it specifically for young people, where the considerations are sharper.
Just remember there is nobody on the other side—just a wall bouncing back what you send.
Exercise 18.1: Find your recurring use
For one week, note every moment you think I wish someone would just sort this out.
At the end of the week, look at the list. Two or three will be tasks AI handles well, and those are the ones that will actually stick.
Exercise 18.2: Work at the edges
Take a creative project you care about. Use AI only before and after: constraints and research first, critique afterward. Do the making yourself.
Then ask whether the finished thing is better and whether it is still yours.
Exercise 18.3: The read-aloud test
Take a paragraph of your own writing. Have a tool polish it. Read both aloud.
Note what was lost. Most people find something.
Learn More About Artificial Intelligence
Learn more about the CLEAR, TRUST and LEARN methods for Artificial Intelligence