AI is like magic in a box

Refined insight

Once you know how to direct it, AI can turn an intention into something real so quickly that using a computer begins to feel like waving a magic wand.

What better definition of magic is there?

“When you can just say what you want and then it is built and it is presented to you. I don’t know what a better definition of magic is.”

The distance between wanting and creating is collapsing

For most of history, wanting something and creating it were separated by specialized knowledge, manual skill, money, time, and access to the right people and tools. AI compresses that distance.

A person can increasingly describe an objective in ordinary language, collaborate with AI through successive prompts, judge the results, and refine them until an idea becomes a usable artifact. It is not a single wish followed by a perfect result. It is closer to waving a wand, seeing what appears, and making a few corrective gestures until the thing in your imagination exists outside it.

This website came out of the box

Common Sense Magic is itself an example. Tim supplied the insights, direction, taste, and judgment. AI helped refine the language, develop the design, create the website, and carry out repeated improvements.

The important skill is not merely having access to AI. It is knowing what you want, expressing that intention clearly, providing useful context, recognizing weak output, and continuing to direct the system until the result matches the objective. The apparent magic is human intention and judgment multiplied by an increasingly capable tool.

The magic works better when you keep your brain switched on

AI can help you do remarkable things, but its suggestions still need your judgment. It can recommend a route that makes little sense, or overlook a much simpler way to reach your objective. You remain in charge of what you do.

Tim encountered this while trying to display expanded folder trees in Windows, with all the files in a set of folders visible together. He recalls spending several hours working through GPT’s Windows instructions, then downloading and testing recommended applications. None worked as he needed.

Eventually, he asked a different question: “Could you build an application that does what I want?” GPT said yes. In Tim’s account, it produced the first version in less than four minutes. The better approach had been available, but neither the original question nor AI’s recommendations had brought it into view.

The lesson goes beyond checking whether an answer is correct. Ask whether the proposed process is worth following at all: “Is there a simpler or more efficient way to achieve this? Could we create the tool instead?” A useful answer can still miss the best available approach.

And when AI replies, “You’re right,” that agreement is not proof. Ask what evidence supports the recommendation, compare alternatives, and test the result. A persuasive explanation should help you evaluate a decision; it should not replace that evaluation.

AI can expand what you can do. You still have to decide what makes sense.

AI feels like wandering into Aladdin’s cave

For Tim, the magic is not merely a metaphor for technical capability. It describes the emotional experience of discovering that possibilities which once felt inaccessible are suddenly open for exploration.

As a child, he built jet aircraft from Airfix kits and played with Lego and Meccano. Sometimes he followed the intended design; often he made something chaotic and completely his own. He still gets the greatest pleasure from woodworking when he creates something unconventional that previously existed only in his imagination.

AI expands that kind of freeform play across writing, images, research, software, design and business creation. It does not grant every wish or remove the need for effort and judgment. But it removes so many old technical barriers that imagination can become the starting point instead of the distant reward for years of specialist training.

AI is the most versatile creative toy ever invented.

The same tool can feel like a threat

Other people encounter AI and feel irritation, anxiety or resistance. Some reasonably worry about employment, accuracy, power or social consequences. Others mainly see an unwelcome burden: another unfamiliar system to learn when they would prefer to continue doing a familiar job in a familiar way.

These are not two rigid kinds of person. Everyone moves along the spectrum depending on the situation. But the contrasting questions reveal two very different emotional orientations:

The creator asks, “What can I make with this?”

The resistant user asks, “What will this change, take away, or force me to learn?”

For one person, change is possibility. For the other, change is threat.

Resistance is still a choice—with consequences

Tim and Sonja encountered an earlier version of this divide when personal computers and desktop software emerged in Britain during the 1980s. They embraced the unfamiliar technology, taught themselves how it worked and converted that learning into new careers, much higher earnings and opportunities that had not previously been available to them.

Tim also knew people who resisted using PCs. Avoiding the technology preserved their short-term comfort, but it did not prevent computers from changing the workplace around them. It reduced their ability to benefit from that change.

The fuller story appears in Optimism creates energy. Its relevance here is direct:

You do not avoid the consequences of technological change by refusing to learn the technology. You surrender more of the opportunity to people who do.

Rosanne crossed the capability boundary

Tim’s daughter Rosanne is intelligent and highly social, but she is not particularly computer-oriented. She rarely used her laptop and knew nothing about creating websites.

With a small amount of help from Tim, she used GPT to create and publish Luxury Appliance IQ, a website offering practical guidance about choosing, using and caring for kitchen appliances. She became excited about continuing to develop it.

The site is still being developed, so this is not a claim of commercial success. The achievement is more fundamental: a complete novice crossed from “I don’t know how to build a website” to creating, publishing and improving one herself.

The box will not remain confined to the screen

Today the effect is clearest in writing, images, software, research, analysis, websites, and business processes. But AI is already connecting with design systems, robotics, logistics, and automated manufacturing.

It is not hard to foresee a person describing a physical object, having AI design and test it, routing the design to an automated factory, and arranging delivery with little or no routine human involvement. An intention expressed in ordinary language could eventually emerge from the other side of the system as an object delivered to the front door.

It is just engineering

This is not magic waiting upon undiscovered science. The necessary capabilities already exist in emerging forms. Joining them into dependable and economical systems may be a major engineering undertaking, but that puts it in the same category as building a bridge over a vast distance: difficult, certainly, but neither mysterious nor dependent upon rewriting the laws of nature.

Humanity repeatedly completes difficult engineering projects once knowledge, incentive, and tools converge. If an AI-directed design-to-delivery system can be built, competitive pressure and human inventiveness make its arrival not merely possible but, in Tim’s view, inevitable.

The magic is real as an experience. The box is still being built.

The practical magic

If you have barely used AI, begin with curiosity. Ask about an everyday problem or something you have always wanted to understand. Request another explanation, an example or a different perspective. Learning becomes a responsive conversation rather than a search for the right book and the right chapter.

I’ve been wondering about ________. Help me understand it by first asking what I already know, then explain it in a way that makes sense to me.

If you already use AI for answers, move from asking to making. Choose one small thing to create. If you do not know what that should be, ask AI to help uncover it.

Based on what you know about me so far, suggest five things I might enjoy creating with your help. Explain why each one might suit me.

Or begin with no idea at all:

Ask me one question at a time to uncover something I would enjoy creating. When we find a promising idea, help me make the simplest first version.

The door is open. Whether it reveals possibility or inconvenience depends partly upon the question you bring to it.