Every tool leaves something behind. A hammer leaves a shelf on the wall and a slightly stronger arm. A good map leaves you knowing the city better. A slot machine leaves an empty wallet and a strange tiredness. We usually judge technology by what it does while we use it. A better test is what it leaves when we put it down.
That test matters more now that software shapes our evenings, our learning and our work, and AI systems write, plan and give advice. With that much reach, the gap between a tool that builds people up and one that drains them decides a lot about who we become.
The test is simple enough to use at a kitchen table. Ask whether a tool helps living things grow. Living things here means people, their skills and attention, the communities they belong to, and the natural world around them. A good tool leaves them stronger. A bad one feeds on them.
Builders tend to take one of two stances. The harvester looks at users as a supply of attention, data and spending, and asks how much can be taken this quarter. If people are worn out by next year, that is next year's problem. The gardener asks what users need to do well and measures success by how they are doing. Gardeners still make money. They make it in a way that leaves people better off.
Most products have some of both. You find out which one wins when the two pull apart. Say a user tries to cancel. The harvester's app adds another screen and one more notification. The gardener's app makes leaving easy, betting that a tool which helps will get chosen again.
Attention is the first thing most apps compete for, and it can be built up or worn down. It recovers with rest and degrades under constant interruption. Someone who gets long uninterrupted stretches gets better at concentrating. Someone pulled in twenty directions an hour slowly loses the habit.
When a product is paid for in minutes of engagement, every design choice leans toward more minutes: the endless feed, autoplay, the red badge, the push notification at eleven at night. Each one works. Together they train people out of the ability to focus, and that costs far more than the minutes.
An app built the other way respects the end of a task. It sends fewer notifications and makes each one worth reading. It lets you find what you came for and then lets you go.
Skills grow when used and fade when every hard part is done for us. This is the main question for AI in daily life. Sometimes handing a task to a machine frees you for more important work. Sometimes it removes the practice that would have made you capable.
Take a student learning to write. An AI that writes the whole essay saves an evening and costs a skill. An AI that asks what the student is trying to say, points out where the argument is thin, and suggests one paragraph to rework leaves a better writer by next week. Both run on the same model.
Work follows the same pattern. A good assistant takes the routine parts so people can spend more time on judgment. A poor one takes everything and leaves people signing off on output they no longer understand. As more of the hard problems in AI come down to human judgment, tools that wear down judgment undercut their own future.
Communities work the same way. A family group chat can keep relatives close across an ocean. A neighborhood forum can find a lost dog in an afternoon and organize a street party by the weekend. Small acts of attention add up to trust.
The harvester's version of the same product pushes people toward conflict, because conflict drives engagement. A forum full of shouting produces plenty of traffic and very little community. The dashboard shows growth while the neighbors stop speaking to each other.
A better measure for a social product is whether people leave a conversation understanding each other better. It is harder to count than daily active users and much closer to what people want.
Technology affects the physical world directly too. Data centers use power and water. Devices are made from mined minerals and end up as waste. Technology also helps: a moisture sensor tells a farmer when a field needs water, and a frost forecast saves a harvest. Counting the clean energy while ignoring the pile of discarded phones is bad bookkeeping. Count the whole cycle and design for repair.
Building like a gardener
Start with the metric. Measure growth in the user, such as skills learned, problems solved and time saved, instead of clicks. These are harder to track, which is exactly why teams that track them end up building different products.
Then set the defaults. Most people never change their settings, so the default is the real design. Good defaults mean fewer notifications, clear stopping points and an easy way out. Each costs some engagement now and earns trust later.
Accept slower numbers. A product designed to help people often grows more slowly at first than one designed to capture them. If you want to build something that lasts, you have to accept that tradeoff and explain it to whoever funds the work.
Be honest about what the tool is for. A meditation app that floods you with streak reminders has lost track of its purpose. So has a learning platform that keeps students busy but never makes them confident.
You can run the test on your own phone today. Go through your apps and ask what each one left you with this past month. Some taught you something, put you in touch with someone, or saved time you then used well. Keep those. Delete the ones that mostly took, and move anything you are unsure about off your home screen for a month to see whether you miss it.