The Space Monkey and the Copilot: How I Use AI in My Research

My life has changed in ways I would not have predicted even a few years ago.

Artificial intelligence entered my life around 2023. At first, it was mostly an object of curiosity. Interesting, occasionally useful, sometimes impressive, but not something I saw as fundamentally changing what I was capable of doing.

By 2025, that had changed.

AI had become part of how I research, write, organize information, test arguments, learn technical skills, and turn ideas that once lived almost entirely inside my head into something another person could actually read.

I have come to believe that artificial intelligence is not simply another useful technology. It may be one of the most consequential tools Homo sapiens has ever developed.

And, somewhat jokingly, I think it has created a new creature:

the Space Monkey.

The Space Monkey

I have called myself the Space Monkey for a while.

Part of it is self-deprecating. I am not an academic philosopher, a professional journalist, a lawyer, or the head of some well-funded research institute.

Sometimes I really do feel like a monkey hitting buttons until something interesting happens.

But there is a more serious idea underneath the joke.

For most of human history, doing serious public research required significant resources.

You needed access to information. You needed time. You often needed institutional support. If you wanted to publish something credible, it helped enormously if you could write extremely well, understand research methods, organize large quantities of material, interpret technical language, navigate bureaucracy, and present your argument clearly enough that other people would take it seriously.

Those barriers excluded a lot of people.

I know because, for much of my life, I would have been one of them.

I struggled in school. Writing did not always come naturally to me. I could become deeply interested in questions, but translating what was happening in my head into a polished argument was much harder.

Later in life, I began thinking seriously about questions such as determinism, punishment, responsibility, human dignity, and how society treats people who cause harm.

Those ideas were there.

What I often lacked was a practical way to articulate them.

AI changed that.

The copilot

I sometimes describe AI as my copilot.

That does not mean I type "write an article about prisons" and publish whatever comes back.

The process is much more interactive than that.

I usually begin with the question.

I decide what bothers me.

I decide which argument I want to explore.

I bring documents, examples, news stories, statistics, court decisions, academic studies, government policies, access-to-information releases, and sometimes an idea I have been thinking about since I was eighteen.

Then the conversation begins.

I challenge wording.

I reject arguments I do not agree with.

I ask for sources.

I suggest sources myself.

I ask whether a claim is actually supported.

I ask for stronger counterarguments.

I change the structure.

I remove sentences that sound too certain.

I add caveats.

I sometimes spend far too long arguing with the machine over a single word.

When I wrote about determinism, for example, I wanted the article to engage with the work of Robert Sapolsky and Sabine Hossenfelder. I wanted to discuss the famous study of parole decisions that Sapolsky frequently references, while also acknowledging methodological criticism of that study rather than presenting it as unquestionable proof.

Those were editorial decisions.

AI helped me execute them.

That distinction matters to me.

"But AI wrote it"

People who know me sometimes immediately recognize that I do not naturally write in the polished style found on parts of this website.

They are correct.

I have never tried to hide the fact that I use AI extensively.

But I also think the statement "AI wrote it" can become a way of avoiding a much more interesting question.

Is the argument correct?

If I write an article asking whether determinism should affect how criminologists think about punishment, saying that AI helped construct the prose does not answer the argument.

If I cite a study incorrectly, that matters.

If I misrepresent a source, that matters.

If my reasoning is poor, that matters.

If the evidence contradicts my conclusion, that matters.

But the fact that an AI helped me express the argument is not, by itself, a rebuttal.

The question remains.

What happens when a Space Monkey discovers access to information law?

The same applies to my research into Canadian corrections.

A large part of this project involves access-to-information requests.

Some of those requests are deliberately narrow and technical. They ask for engineering records, safety evaluations, policy discussions, procurement documents, internal correspondence, or records explaining how a particular institutional decision was made.

AI helps me formulate those requests.

It helps me identify ambiguity.

It helps me separate one large research question into several smaller requests.

It helps me think about what records might exist and how a government institution might describe them internally.

In another era, work like this might have been undertaken by a journalist, an academic researcher, an advocacy organization, or a small research team.

Now one curious person with enough persistence can do much more of it independently.

The institution receiving the request cannot simply answer:

"A chatbot helped you write this."

The records either exist or they do not.

The request is either valid or it is not.

The applicable access-to-information law still applies.

That is what I find so interesting about this moment.

Government institutions, universities, corporations, journalists, researchers and advocacy organizations are beginning to encounter people who have dramatically more research capacity than one ordinary individual would previously have possessed.

The Space Monkey has entered the bureaucracy.

This does not make me an expert

There is an important limitation here.

AI does not magically turn me into a criminologist, engineer, lawyer or philosopher.

It can make mistakes.

It can misunderstand a source.

It can produce confident nonsense.

It can offer an elegant explanation that turns out to be wrong.

And someone using AI without skepticism can produce misinformation much faster than they could before.

That is why I try to distinguish between what I know, what a source actually says, what I infer, and what remains uncertain.

Primary documents matter.

Sources matter.

Expertise matters.

Being willing to change your mind matters.

AI increases capability.

It does not eliminate the need for judgment.

The questions are not necessarily new

There is another thing worth admitting.

Most of the questions I am interested in are not original.

Determinism has been debated for centuries.

Questions about punishment, rehabilitation, human dignity and state power are not new.

Criminologists have spent generations studying the biological, psychological and social factors associated with harmful behaviour.

Human-rights advocates have questioned degrading prison conditions long before I arrived.

Journalists have filed access-to-information requests far more sophisticated than mine.

I am not claiming to have discovered these ideas.

What AI has changed is my ability to participate in the conversation.

I can take an idea such as determinism and ask:

If people do not ultimately choose the brains, environments, histories and capacities from which their decisions emerge, what does that mean for punishment?

What does it mean for retribution?

What does it mean for how we treat people once they are already securely confined?

Those questions existed before me.

I simply have a much better tool for pursuing them now.

Lucy, Power BI and the thinking ape

One of my favourite conversations with AI happened while I was struggling with Power BI.

After several rounds of debugging, I joked that AI was doing most of the learning while I was violently hitting the keyboard, eating bananas and making monkey noises.

The conversation eventually escalated into an imaginary technological timeline:

Tomorrow: calculated tables.

Friday: fire.

2027: nuclear fusion.

2029: monkeys and machines merge.

2030: Dyson sphere.

Eventually we ended up imagining travelling roughly 3.2 million years into the past and trying to explain modern civilization to Lucy, the famous Australopithecus afarensis specimen.

Imagine trying to explain it.

First, her descendants get better at hitting rocks together.

Then fire.

Agriculture.

Cities.

Writing.

Mathematics.

Steam engines.

Electricity.

Computers.

Then we take ordinary matter, manipulate it at microscopic scales, and create machines capable of processing language and helping us reason.

And eventually one of Lucy's distant descendants sits in an office in Canada, using the thinking machine as a cognitive exoskeleton while becoming increasingly angry because Microsoft will not let him format the word TOTAL properly.

Somewhere between Australopithecus afarensis and the Dyson sphere, that happened.

It is absurd.

It is also kind of extraordinary.

Tools have always changed what humans can be

Humans have always extended ourselves through tools.

Stone tools extended the hand.

Writing extended memory.

Printing extended the transmission of knowledge.

Computers extended calculation.

The internet extended access to information.

AI is beginning to extend something closer to cognition itself.

It does not replace the person using it.

It changes what that person can attempt.

That is what makes the Space Monkey interesting to me.

The Space Monkey is not literally a new biological species.

It is a joke about a genuinely new type of capability.

A person who may not write like an academic, possess institutional backing, employ research assistants, or have decades of specialized training can nevertheless investigate questions, organize evidence, communicate ideas, and challenge institutions at a level that would have been extraordinarily difficult for that same person only a few years ago.

That will have consequences.

Some will be bad.

People will use these tools to create misinformation, spam, propaganda and mountains of low-quality content.

But there will also be people who previously had very little ability to make themselves heard who suddenly discover that they can ask serious questions and insist on serious answers.

I think that matters.

The monkey still chooses where to go

I do not know where all of this leads.

AI will improve.

The boundaries between human reasoning and machine assistance will probably become increasingly difficult to describe.

Our norms around authorship, expertise and intellectual work will change with it.

But, at least for now, the relationship feels fairly simple to me.

I bring the curiosity.

I bring the questions.

I bring the values.

I decide what matters enough to pursue.

I decide what gets published.

The copilot helps me get somewhere I could not have reached nearly as easily on my own.

And I am completely comfortable acknowledging that.

Because ultimately, whether an argument was typed entirely by one human being, refined by an editor, developed by a research team, or constructed through hours of conversation between a Space Monkey and an artificial intelligence, the same question remains:

Does it hold up?

That is the question I want readers to ask of everything I publish here.

Not whether I had help.

Whether the evidence, reasoning and conclusions deserve to be taken seriously.

And if they do, then the Space Monkey has done its job.

🐵🚀