The Apocalypse Has a Headline Now

Several mornings ago, I came across a headline that was almost designed to make 

one put down one’s coffee: “How AI would ACTUALLY kill all humans has been

 revealed.” It came with a smoky, ruined cityscape and the familiar silhouette of a mysterious figure at glowing computer screens. Well. That certainly gets one’s 

attention.

At my age, I have lived through enough predictions of doom to know that panic 

is rarely a reliable guide to wisdom. We were supposed to be destroyed by nuclear

 war, overpopulation, acid rain, computers that would crash at midnight in the year

 2000, and depending on which decade one remembers, the next epidemic, 

economic collapse, or foreign menace. Some dangers were real. Some remain real. 

But sensational language has always had a way of making fear feel like knowledge.

Artificial intelligence deserves serious attention. It is already changing how we write, learn, work, communicate, diagnose illness, make art, spread rumors, and conduct 

war. But “AI will kill us all” is not a fact that has been revealed. It is a frightening speculation about possible futures and there is an important difference.

What people are really worried about

When experts talk about an AI catastrophe, they generally do not mean that 

today’s chatbot will wake up tomorrow morning with an evil plan. The more serious concern is that future systems could become much more capable, more 

autonomous, and more deeply woven into the machinery of modern life: banking, transportation, military systems, scientific laboratories, energy grids, health care, communications, and government. The fear is less like a Hollywood robot rebellion

 than a slow and dangerous surrender of human judgment.

There are several broad anxieties:

• Powerful AI could be used by people with bad intentions, to produce convincing propaganda, automate cyberattacks, design dangerous biological agents, 

  or manipulate elections and public opinion.

  
  

• Governments and corporations may race ahead because they fear being left

   behind, even when safety rules have not caught up.

  
  

• People may become so dependent on systems they do not understand that they

   lose the ability to question, correct, or shut them down.

  
  

• A sufficiently advanced system, if poorly designed or poorly controlled, might

  pursue goals that are not aligned with human well-being.

  
  

Those are not trivial concerns. The 2026 International AI Safety Report describes a possible “loss of control” scenario in which highly capable AI systems might act 

outside effective human control. Yet the report also emphasizes that the likelihood, timing, and exact nature of such outcomes remain deeply uncertain. Current AI

 systems are not known to possess the kind of independent power imagined in apocalypse headlines.

The danger may be us

I am less worried about a machine suddenly developing a melodramatic hatred of humanity than I am about human beings using powerful tools carelessly, greedily, or irresponsibly. That is the older story, isn’t it?

The atom did not invent Hiroshima. Social media did not invent cruelty or loneliness

The printing press did not invent lies. Every great technology enlarges human

 capacity, for healing and learning, certainly, but also for exploitation, surveillance, deception, and violence.

AI can help a physician notice a pattern in a scan. It can help a lonely older person

 find words for a letter to a grandchild. It can translate across languages, assist 

disabled people, organize mountains of information, and perhaps accelerate 

discoveries in medicine and climate science.

But the same ability to generate words, images, code, and scientific ideas can 

be used to flood the world with falsehoods, impersonate our voices, invade privacy,

 and put sophisticated tools into reckless hands. The immediate question is not 

whether a robot will hunt us down. The immediate question is whether we will

 preserve human responsibility while allowing machines to do more and more for us.

That is a much less cinematic problem. It is also more urgent.

Fear is not a policy

Some AI researchers have offered alarming personal estimates of the chance that advanced AI could pose an existential threat within the next decade. Others argue 

that extinction is not a realistic near-term outcome, even while agreeing that 

present-day harms—fraud, disinformation, cybercrime, unfairness, job disruption,

 and concentration of power—are very real. There is no scientific consensus that

 AI will destroy humanity, and no honest person can claim to know the odds with

 certainty. This uncertainty should not lead us to shrug. Neither should it lead us to believe every terrifying headline.

We need rules before, not after something goes wrong. We need independent

 testing of powerful systems, meaningful oversight, clear accountability when 

companies cause harm, and international cooperation in areas such as 

cyber warfare, autonomous weapons, and biological risk. We need to resist the I

dea that speed is the highest virtue, especially when the technology in question

 may affect billions of lives.

Most of all, we need ordinary citizens to pay attention.

For too long, technology has been presented as though it arrives from the heavens, irresistible and beyond democratic discussion. A company releases a new tool;

 we are told it is inevitable; then we are left to adapt. But technology is made 

by people, funded by people, sold to people, regulated or not regulated by people.

There is nothing inevitable about handing every important decision to systems that cannot love, grieve, remember, forgive, or accept moral responsibility.

Keeping the human in the picture

As someone who writes, reflects, and tries to make sense of a changing world, 

I find AI both astonishing and unsettling. It can be useful. It can even be 

companionable in a limited way. But it has no childhood, no grandmother’s stories,

 no private grief, no memory of holding a baby, no fear of losing a friend, 

no conscience awakened in the middle of the night.

It can imitate the language of feeling. That is not the same as feeling.

Perhaps that is the point we must remember as the machines become more

 fluent and more impressive. Intelligence is not wisdom. Speed is not judgment.

 Information is not understanding. And sounding human is not the same as being 

human. The future will not be decided by whether we can build ever more powerful

 AI. We almost certainly can. It will be decided by whether we are wise enough 

to insist that human dignity, human agency, and human responsibility remain 

at the center of the story.

So when you see a headline announcing that AI will “actually kill all humans,” 

do not dismiss the underlying issues. But do not surrender your common

sense either. The future is not yet a ruined cityscape. It is still a human choice.

My Photo of the Day: