About Reflections from a Nerd

Reflections from a Nerd is an independent publication about the widening gap between fast-moving technology and slow-moving institutions.

I write about artificial intelligence, bureaucracy, democracy, education, work, geopolitics, truth and ethics. And, when reality becomes too predictable, there is Quantum Spielerei: attempts to make the deeply strange world of quantum physics understandable without pretending that it has suddenly become less strange.

There is, however, something else you should know about this publication.

A substantial part of it is made with AI.

Not just spelling corrections, summaries or the occasional suggestion for a better headline. AI is deeply involved in the research, thinking, writing and illustration process.

And that is intentional.

I am the AI director

I’m Dirk, a retired IT professional who spent several decades watching computers become dramatically more capable while the organizations surrounding them often changed at a rather more leisurely pace.

That probably explains much of what you find here.

It also explains how I work.

I choose the subjects. I decide which questions seem worth asking. I determine the initial perspective or hypothesis I want to investigate. Sometimes it begins with an irritation, sometimes with something I have read, and sometimes with an observation that doesn’t quite fit the conventional explanation.

Then AI goes to work—not to prove that I am right, but to find out whether I am.

I ask it to search for background material, studies, statistics, historical examples and opposing arguments. I ask it to compare explanations, find counterexamples, expose holes in my reasoning and actively look for evidence that could disprove my starting hypothesis. It helps structure the argument and, increasingly, does much of the actual writing.

So no, I am not sitting behind my Mac painstakingly composing every sentence you read here.

I don’t think pretending otherwise would make the ideas any more valuable.

My role has gradually become something different: AI director rather than traditional author.

I decide where we are going. AI helps explore the territory.

But who is actually thinking?

That question is more interesting.

Suppose I start with the idea that universities have become too bureaucratic. It would be very easy to ask AI for twenty arguments proving that I am right.

That isn’t particularly useful.

Instead, I ask whether the proposition is actually supported by evidence. Find researchers who disagree. Look for counterexamples. Tell me where the reasoning is weak, where I am generalizing too far and what evidence would undermine the conclusion.

Quite often the original idea changes along the way.

Sometimes considerably.

That is one of the things I find most interesting about working this way. AI isn’t simply a very fast ghostwriter. It is a research assistant, sparring partner, critic and writer. It can challenge my assumptions, improve an argument and occasionally persuade me to abandon one.

It can influence my thinking.

It doesn’t get the final vote.

I decide what survives the process: what gets published, what gets deleted, what needs more research and what still doesn’t convince me. I change emphasis, reject arguments, rewrite passages and sometimes send an entire draft back because, despite being perfectly polished, it is painfully boring.

The final perspective is mine because the final judgment is mine.

One curious person, considerably amplified

There is another reason I work this way.

The subjects that interest me are far too broad for any individual to master.

One week I may be looking at European regulation, the next at university research, autonomous vehicles, political institutions or the peculiar behaviour of entangled particles.

I am obviously not an expert in all of them.

AI doesn’t magically make me one.

What it provides is intellectual leverage.

A single curious person can now explore a question with something resembling a small research team: searching widely, following references, comparing competing explanations, attacking assumptions and producing several versions of an argument before deciding which one is worth pursuing.

That changes what an individual publication can be.

Reflections from a Nerd is partly an experiment in discovering how far that leverage can be pushed.

What you’ll find here

Most essays start with something that bothers me because the conventional explanation seems incomplete.

Why do institutions accumulate rules much more easily than they remove them? Why can technological systems improve at extraordinary speed while political and administrative systems become increasingly difficult to change? What happens to education when access to knowledge stops being scarce? What happens to expertise when AI can suddenly wander across disciplines?

And occasionally: what on earth is an electron actually doing?

I don’t expect every reader to agree with my answers.

I’m not entirely sure I agree with all my earlier answers either.

The point is to examine systems, challenge comfortable assumptions and occasionally push an argument far enough that its weaknesses become visible.

And who is responsible when it is wrong?

Me.

AI systems can invent facts, misunderstand sources, miss context and produce wonderfully confident nonsense.

Humans have demonstrated similar abilities for quite some time.

That is why AI summaries are not sources.

When an argument materially depends on data, research, law or historical fact, I check the important claims against the underlying research, data or original sources rather than simply trusting an AI-generated account of them.

That does not make errors impossible. When something important is wrong, I correct it.

AI is also used to create many of the original illustrations accompanying these essays. Unless a post says otherwise, readers should assume that AI may have played a role in its research, drafting, editing or illustration.

But responsibility cannot be delegated to a machine. I remain responsible for the selection of subjects and arguments, the judgment applied to them, their factual accuracy and the decision to publish.

Ultimately, the distinction is simple.

AI can research the article.

AI can challenge the argument.

AI can write the sentences.

I decide to publish them.

So if something on Reflections from a Nerd is brilliant, provocative, misguided or spectacularly stupid, don’t blame the machine.

The director approved the final cut.

Welcome to Reflections from a Nerd.