Artificial intelligence is not evil.
Neither is money. Neither is fire. Neither is water.
In the right amount and used for the right purpose, water is essential to life. Too much of it becomes a flood. Fire can warm a home or destroy one. Money can help organize an economy or distort one.
AI is no different.
The question is not whether AI is good or bad.
The question is what we choose to do with it.
If AI can help us develop a material that performs better than plastic without leaving a trail of pollution behind, I am all for it.
If it can help us actually recycle the materials we already produce instead of merely placing a recycling symbol on them, good.
If AI can help us understand climate change, improve energy systems, prevent pandemics, send robots into burning buildings, contain a nuclear accident, fight a forest fire, or perform work that is simply too dangerous for a human being, those are extraordinary uses of technology.
Those are examples of technology doing what technology should do: extending human capability and solving problems we have struggled to solve ourselves.
We are also racing to answer a very different question:
And we are pursuing that question before we have answered the one that should come first:
What happens when millions of people no longer receive the income they once earned?
What happens to housing when fewer people can afford mortgages or rent?
What happens to healthcare, retirement savings, pensions, college tuition, tax revenues and consumer spending?
What happens to an economy built around the idea that people earn income by working when increasingly sophisticated machines can perform both cognitive and physical work?
These are not arguments against technology.
They are arguments for thinking before acting.
Protect the human.
Send the machine into the burning building, the unstable reactor, the wildfire, or the environment where a human life should not be placed at risk.
Ask what comes next.
Replacing human income across an economy is not merely a productivity decision. It changes the system that creates purchasing power and demand.
There is an important ethical distinction between automating danger and automating livelihood.
Replacing a firefighter inside a collapsing building with a machine can save a life.
Replacing an accountant, programmer, designer, attorney or eventually a plumber because a machine can perform the job more cheaply may improve a company's margins.
But when that happens across an entire economy, the consequences do not stop at the company's income statement.
The worker is also a customer.
The worker pays a mortgage.
The worker buys food, cars, clothing and services.
The worker contributes taxes and retirement savings.
The worker is part of the economic system that makes the company's profits possible in the first place.
Technology has displaced labor throughout history. The printing press, sewing machine, assembly line, tractor and computer all changed the nature of work.
But previous transitions generally unfolded slowly enough that society had some chance to adapt. Even highly disruptive machines often still required humans to operate them.
AI potentially changes both sides of that equation.
It can replace cognitive work directly through software, requiring no physical machine to arrive at the office at all.
And as robotics improves, increasingly sophisticated physical work will follow.
We appear determined to discover how much work can be automated before we have seriously decided what economic role humans should occupy afterward.
That seems backward.
If artificial intelligence is as transformative as its advocates say it is, then the responsibility surrounding its deployment should be equally transformative.
The fact that we can automate something does not automatically mean that we should automate it immediately.