Issue 157  November 6th 2026

The Apocaloptimist's Dilemma

Hope, fear, and the choices we're making while the future is still uncertain.

On a recent flight, I watched The AI Doc: Or How I Became an Apocaloptimist. It explores a tension that I suspect many of us feel: AI could unlock extraordinary possibilities for humanity, but it could also introduce risks that we're not prepared to manage.

The term apocaloptimist captures that uncomfortable combination of hope and concern.

A few weeks later, I found myself in a very different AI conversation, this time with people from nonprofit organizations through SPARK. We weren't debating whether AI would save or destroy humanity. We were exploring how these organizations could use it responsibly, what they could delegate, and what they needed to retain human oversight over.

The contrast stayed with me.

One conversation was about the future of humanity. The other was about writing a grant proposal.

Yet both were fundamentally about the same question: How do we make the most of what AI makes possible without losing sight of what matters?

I've been thinking about that question a lot.

THE TELL: We're debating the future while making it in the present

Depending on whom you listen to, AI is going to usher in an era of unprecedented abundance, make much of human work obsolete, concentrate extraordinary power in the hands of a few, or eventually escape our control.

Some of these predictions are more credible than others. They also operate on very different timelines and involve very different risks.

In Are We Too Worried About AI — or Not Enough?, Harvard professor Jonathan Zittrain explores three broad perspectives: AI idealists, alarmists, and skeptics. It's a useful way to understand why thoughtful people looking at the same technology can arrive at radically different conclusions.

Meanwhile, Stephen Messer's The Death of Humanity in Fortune challenges the increasingly common narrative that AI is primarily an existential threat. He argues that we're at risk of overlooking the extraordinary benefits AI could bring.

I don't think we need to choose a camp.

Being optimistic doesn't mean dismissing risks. Being concerned doesn't mean rejecting opportunities. And skepticism shouldn't become an excuse for inaction.

There's a more immediate problem with spending all our time debating which version of the future will materialize.

The future isn't simply something that will happen to us. We're making decisions that shape it every day.

We're deciding what to automate, what to delegate, what to optimize, and what to value. Organizations are making choices about whose interests AI serves, who benefits, and who bears the costs.

We don't have to agree on AI's future to take responsibility for its present.

THE HUMAN DIMENSION: What are we gaining, and what might we be giving up?

During the SPARK discussion, we explored a fairly ordinary example: using AI to help a nonprofit write a grant proposal.

On the surface, this seems like an obvious win. An organization with limited resources can produce a more polished proposal, potentially spend less time on administrative work, and devote more energy to its mission.

But there's another side to the equation.

Who verifies the claims, figures, and supporting evidence? Does the organization have the knowledge to recognize when AI has invented something or misrepresented its work? How much human effort is required to check and correct the output?

And if the organization becomes dependent on AI to communicate its impact, does it retain the ability to articulate that impact itself?

These aren't arguments against using AI. They're questions about what responsible use actually looks like.

Someone in the discussion also raised the environmental cost of AI. It was a useful reminder that the cost of a technology isn't always borne by the person or organization benefiting from it. When we talk about making things faster, cheaper, or more efficient, we should also ask: cheaper for whom, and at whose expense?

This connects to something I've been exploring through the idea of synthetic competence: the possibility that AI-assisted output can look increasingly impressive while the human capability behind it quietly erodes.

We can produce excellent work without necessarily developing the understanding, judgment, or skills needed to evaluate it.

And that's a concern beyond the workplace. It affects how we learn, create, solve problems, and make decisions.

Christopher Noessel explores two related risks, deskilling and overreliance, in his Deskilling and Overreliance: Two Long-Term Risks of AI Assistance, and What Designers Can Do About Them. One of his proposed interventions is particularly interesting: Human Goes First.

Rather than immediately showing someone an AI recommendation, let the person form their own judgment first. Then reveal the AI's perspective and compare the two.

It's a relatively simple design idea, but it reflects a much bigger principle: we should design our interactions with AI to develop human capability, not just accelerate output.

Of course, individuals aren't solely responsible for these outcomes. Organizations, technology providers, educators, and policymakers all have roles to play. Some consequences require collective decisions and accountability, not just better personal habits.

But we do have choices within our own sphere of influence.

THE PRACTICE: Become an active apocaloptimist

I like the term apocaloptimist, but I don't think simply acknowledging both possibilities is enough.

I'd rather think of it as a way of approaching AI: remaining open to its potential, being honest about its costs, and deliberately shaping how we use it.

The next time you're considering an AI application, try asking three questions:

1. What becomes possible?

What could we accomplish that wasn't previously practical? Could we improve an experience, make expertise more accessible, solve a problem, or free people to focus on more meaningful work?

Don't limit the conversation to doing the same things faster.

2. What could we lose?

Consider more than accuracy or privacy. Could we lose a skill through lack of practice? Reduce human connection? Transfer accountability? Create dependencies? Externalize costs to people who don't benefit?

3. What will we deliberately protect or shape?

What needs human judgment? What should remain a human responsibility? How might we design the interaction so that people become more capable rather than merely more productive?

These questions don't produce a universal answer. They help make the tradeoffs visible before we've committed to a direction.

ONE THING TO TRY: PAIR before you prompt

Christopher Noessel's Human Goes First approach offers a simple idea: let people form their own judgment before showing them what AI recommends.

I like this because it creates a little space for independent thinking before AI enters the picture.

It's also a good starting point for something I've been sharing with teams: PAIR - Prime, Assess, Interpret, Retain.

It's a way to keep yourself in the driver's seat when working with AI.

• Prime: Start with your own thinking. Define the problem, form an initial opinion, and establish what you're trying to achieve before prompting AI.
• Assess: Interrogate what AI gives you. Check its assumptions, evidence, logic, gaps, and potential consequences. Don't confuse a polished answer with a good one.
• Interpret: Make the judgment yourself. Decide what to accept, reject, or change, and be able to explain why.
• Retain: Keep the learning. What did you discover? What did you almost outsource? What capability do you want to continue developing rather than gradually lose?

Try this the next time you're using AI for something that involves judgment or problem-solving.

Before opening AI, write down your initial thinking. Then use AI to challenge or expand it. Compare the two, make your decision, and reflect on what you learned.

The point isn't to prove you're smarter than AI. It's to make sure you're developing your own capabilities while benefiting from its contributions.

Use AI to develop your judgment, not replace it.

SPARRING PROMPT

If AI made your organization twice as productive tomorrow, what would you refuse to sacrifice to get there? Would it be the development of people's skills? Human relationships? Accountability? Independent thinking? Something else? And who gets to make that decision?

I suspect the answers will reveal more about our vision of progress than our predictions about AI ever could.

Being an apocaloptimist shouldn't mean sitting back and hoping everything turns out well.

It should mean recognizing that the future is still being shaped, and taking responsibility for the parts we can influence.