TECHSTORY 01· SEPTEMBER 16, 2026

With AI, 'I told you so' will be too late

Slowing artificial intelligence, meaning reducing how fast new AI systems are built and put to use, is the right decision. That is the argument. It begins by acknowledging that getting there is hard.

NDNewsDialy Desk1 MIN READNVDA

Key takeaways

The article argues that slowing artificial intelligence—reducing how fast new AI systems are built and deployed—is the right decision.
It acknowledges that reducing the speed of AI innovation will be hard both practically and politically, but reaches the pro-slowdown conclusion anyway.
The title's point is that 'I told you so' has value only when something can still be done about being right.
If harm from unchecked AI development proves difficult to reverse, waiting for that harm to become undeniable is itself the mistake.
The argument holds that the time to act is before the evidence becomes impossible to deny.

Slowing artificial intelligence, meaning reducing how fast new AI systems are built and put to use, is the right decision. That is the argument. It begins by acknowledging that getting there is hard.

Reducing the speed of AI innovation will be hard both practically and politically. The argument treats those as real obstacles. It reaches the same conclusion anyway.

The headline makes the sharper point. "I told you so" has value only when something can still be done with being right. If the harm from unchecked AI development proves difficult to reverse, then waiting for that harm to become undeniable is itself the mistake. The time to act, the argument goes, is before the evidence becomes impossible to deny.

More from the desk.

07TECH

Nvidia and Meta chiefs reject push to slow AI development

Calls to coordinate a slowdown in artificial intelligence development, meaning a joint effort by companies to deliberately pace themselves in building new AI systems

tech1 MIN

35TECH

AI at work is turning job-territory disputes into a company-wide problem

Artificial intelligence is making it harder for workers to stay in their professional lanes, and that difficulty is registering as a cultural problem inside organizations. "Toe-treading," the informal term for stepping into a colleague's…

tech2 MIN

47TECH

Google commits $15.1 billion to AI infrastructure in Finland, its largest European investment

AI infrastructure, the physical layer of data centers, servers, and power systems that run artificial intelligence at scale, is set for a record expansion in northern Europe. Google said Wednesday it wo

tech1 MIN

Tomorrow · 06:00

Start tomorrow already ahead.

Five stories, a two-minute read, zero noise — the world before your first coffee.

Get the 6AM brief Back to the desk

Frequently asked

What is the article's main argument?

It argues that slowing down AI—reducing how fast new AI systems are built and put to use—is the right decision, even though achieving that is difficult.

Why does the title say 'I told you so' will be too late?

Because being proven right has value only if action can still be taken; if AI harm is hard to reverse, waiting until it is undeniable means it is already too late to act.

Does the article claim slowing AI would be easy?

No, it explicitly acknowledges that reducing the speed of AI innovation will be hard both practically and politically, treating those as real obstacles.

When does the article say action should be taken?

It argues that action should come before the evidence of harm becomes impossible to deny, rather than waiting for the harm to become undeniable.