When refusing AI wastes more

I sometimes wonder whether we are already asking the wrong question about artificial intelligence. Why are we still debating whether AI should have a place in our work? It already does. I see teachers carrying paperwork home, researchers spending hours sorting information, and employees losing precious time to repetitive tasks that software can help handle. I have never believed that making a job unnecessarily difficult makes the work more honorable.

That does not mean we should trust AI with our eyes closed. It makes mistakes, sometimes rather confidently. Students can misuse it. Employees can depend on it too much. Privacy, copyright, bias, misinformation, jobs, and environmental costs deserve serious attention. But caution and rejection are not the same thing. The better response is to learn how to use the tool without letting the tool do our thinking.

Think of a teacher after a full day of classes. There are still quizzes, reports, parent messages, and tomorrow’s lesson. If AI can organize rough notes, prepare practice questions, summarize a long memo, or give a routine letter a decent start, let it help. The teacher still checks everything and decides what is appropriate for her students. But instead of spending two hours on the mechanical part, she finishes in thirty minutes and gets some of her evening back.

I see the same thing in offices. Minutes have to be cleaned up. Survey comments need sorting. Policies must be compared. Reports need first drafts. Data have to be organized. Why should an employee spend half a day manually arranging material that a machine can organize in minutes? Let AI do the first pass. Then let the human being verify the facts, improve the language, talk to people, exercise judgment, and take responsibility for the final output.

That is not intellectual laziness. We did something similar when calculators entered classrooms, spreadsheets entered accounting offices, and search engines changed research. None of them eliminated the need to know what we were doing. They changed where we spent our effort. The OECD-reviewed research tells us two things at once: generative AI can make some work more productive, and depending on it too much can make us think less for ourselves. We should be mature enough to hold both ideas at the same time.

The environmental cost is also real. AI needs power, water, hardware, and data centers. The International Energy Agency puts data-center electricity consumption at about 415 terawatt-hours in 2024, roughly 1.5 percent of global electricity use. And demand is rising. That deserves attention, not dismissal.

But environmental accounting should be fair. AI has a footprint, yes. So does almost everything else we do digitally. Printing has a footprint. Offices have one. Meetings have one. Paper, ink, air-conditioning, computers, storage, transport, and travel do not suddenly become environmentally free simply because AI is not involved. Five hours of repetitive office work also consume electricity and human time. This does not mean an AI query is automatically greener. It simply means that a fair environmental accounting should examine the whole process, not just the newest machine in the room.

AI can also help reduce some forms of waste. It is already being used to manage energy better, improve delivery routes, support farming, streamline transportation, and make manufacturing more efficient. So the question cannot simply be, “Does AI use energy?” Of course it does. The more useful question is whether a particular use is worth the resources it consumes.

You can see the difference easily in education. One student may let AI finish the assignment and submit it without even reading the answer. The work gets done, but not much learning happens. Another may keep asking, “Why?” or “Can you explain that more simply?” The student practices, checks other sources, revises, and eventually makes the final call. AI helped, but the student still did the thinking.

The same should be true at work. AI should not make doctors less careful, engineers less responsible, journalists less questioning, researchers less rigorous, or teachers less present to their students. Let the machine help with the routine stuff. Save more of our time for listening, deciding, creating, mentoring, consoling, and those moments when experience tells us, “Wait, something is not right here.”

That said, not everything needs an AI assist. Some things are worth doing slowly. Students still need to wrestle with problems. Writers sometimes discover what they really think only after struggling with a paragraph. Professionals need enough knowledge to recognize when a machine is confidently wrong. An office that automates a bad process may simply produce bad work faster.

There is no prize for spending three hours on mechanical work that a useful tool can help finish in thirty minutes. I would rather give those saved hours to a struggling student, better work, another problem worth solving, my family, or a little rest. Time saved is valuable only if we do something worthwhile with it.

So yes, use AI when it makes sense. Just do not switch off your own judgment. Check the facts, and keep confidential information safe. Question suspicious answers. Give proper credit. Know enough about your work to catch the machine when it gets things wrong. And never outsource judgment, ethics, or responsibility.

There is nothing intelligent about using AI blindly. But doing everything the hard way just to prove that we still can is not particularly intelligent either.