Privacy.
That word kept appearing in the comments after Part 1. Readers wanted to know where their information goes, whether someone could impersonate them, and how governments might use AI to monitor or control people.
Those questions deserve answers. They also bring me back to another question: who benefits when we fear AI?
Companies may gain investment. Publishers may gain attention. Governments may gain support for their priorities. The benefit depends on which fear is being presented and what we are asked to do about it.
One principle matters throughout: someone benefiting from a warning does not make the warning false. We still need to examine both the evidence and the interests surrounding it.
Companies: selling the power inside the cage
Imagine two pitches to an investor.
The first promises software that helps people complete everyday tasks. The second describes technology so powerful that controlling it could confer an extraordinary advantage.
Even when the second pitch contains a warning, it can sound like an invitation to own something valuable.
This is what I mean by the “devil in a cage” idea. The danger may frighten the public while making ownership of the cage attractive to someone else.
That is my interpretation of a possible incentive, not proof that a company invented a safety concern to raise money. Fear can also drive customers and investors away.
But when a developer describes a dangerous capability, I want to examine the solution it proposes. Does it invite independent testing and meaningful limits on its own behaviour? Or does the argument leave us increasingly dependent on that developer?
Claims of extraordinary power should come with extraordinary scrutiny.
That standard applies to reassurance, too. In the comments, my advice about switching off model training was too broad. Defaults vary, and opting out of training does not mean conversations are deleted or all access is prevented. OpenAI's data controls guidance explains that distinction. I owe readers the same precision I expect from providers.
Media: earning our attention
A frightening headline can win a publisher or creator our next click.
A 2023 study in Nature Human Behaviour found that negative words increased clicks in the headline experiments it analysed. That result concerns a particular dataset, rather than proving a universal rule about news or AI coverage. Read the study.
It gives us a reason to examine how a story is packaged. Has the headline turned a possibility into a certainty? Does the article distinguish something that happened from something a researcher predicts?
Responsible reporting makes those distinctions visible. Readers should not have to dig through a dramatic story to discover how uncertain its central claim is.
Governments: winning support for the race
Fear of AI is only one part of this conversation. There is also fear of another country getting ahead.
The US government's AI strategy explicitly links leadership to global standards, economic benefits and national security. Competition is central to its public case. US AI Action Plan.
My concern is what that framing can make easier to accept. If falling behind is presented as a national threat, companies seeking infrastructure support and governments pushing rapid development have a persuasive argument for urgency.
We should still ask who receives the benefits, who bears the costs and which safeguards remain in place. The pressure to win should not make those questions disappear.
For readers here in the Caribbean, the debate also concerns our choices. Whose tools will our businesses depend on? What protections will we expect from the companies supplying them?
What should we ask?
When a warning reaches us, three questions can help:
- What evidence supports it?
- Who gains money, attention or authority from the proposed response?
- Will that response actually protect us?
I teach AI, so those questions apply to me as well. My work gives me a reason to encourage participation. It also gives you a reason to examine my advice.
A useful warning should help us make a better decision. Fear alone is not enough.
Join our free School of AI for Business community, where business owners are learning, asking questions and navigating this new era together.
Next: Who Benefits When We Ignore AI's Risks?
Read Part 1: I Teach AI. I Still Question the People Building It.