Web dev at the end of the world, from Hveragerði, Iceland

My writing on AI; the story so far


I’ve cleaned up the notes and essays I’ve been posting on AI over the past few months while doing my research for The Intelligence Illusion and figure it might be useful to have them all in one thread, for reference.

In this one from February, I pointed out that, since the tech industry has blundered every major ‘disruption’ it promised us over the past decade, AI is its last chance at keeping the party going.

Generative AI is the tech industry’s Hail Mary pass

We have these two on AI-generated works and copyright

EU and copyright protections for AI-generated works

Copyright, Situating Search, and other links & notes

Then I wrote this one in March.

If you are claiming double-digit percentage improvements in productivity, it won’t take long for the rest of us to see that you’re bullshitting

When you promise an AI revolution, eventually you will have to deliver

Another from March. On advocacy research. You should ignore most AI studies you see because they tend to be fluffy marketing bullshit.

Why you should ignore most AI research you hear about on social media

Here I point out that if tech companies are basing all of their new, revolutionary AI features on AI summarisation, those better work properly.

And they don’t.

AI summaries are almost certainly unreliable

This one goes over some of the issues with using AI in healthcare. Relevant since Epic is trying AI again, this time in collaboration with Microsoft.

AI in healthcare

This one is a thought exercise, where I’m thinking out loud about regulating AI. Unlike others, I don’t think AGI is an issue and my worry is primarily about what big corporations do with it.

How to regulate AI

Here I point out that all of these AIs are fundamentally American and that is a pretty substantial problem in out multi-cultural world. Because most of us aren’t American.

The AI is an American

Here I show you, in detail, how I read through a study on the productivity benefits of AI.

Critically reading academic papers is a skill and takes practice.

If you can’t be arsed, you can just leave it to me and buy my book instead 😎

The hard truth about productivity research

In this one I riff of a discussion that points out that algorithmically-generated art (such as AI art) is anti-poetry, which I still find to be a compelling point.

AI, Algogen, and Anti-Poetry

Here I go over how AI chatbots trigger anthropomorphism and why that matters.

Three factors of AI anthropomorphism

This one, from earlier this week, is an extract from the book.

Artificial General Intelligence and the bird brains of Silicon Valley

Another extract from the book. This time a warning about the AI industry and AI research. It has a long history of pseudoscience and unfulfilled promises.

Beware of AI pseudoscience and snake oil

Finally, the last extract and the latest essay, where I point out the fundamental friction that exists between programming culture, software development, and language models.

AI code copilots are backwards-facing tools in a novelty-seeking industry

All of these posts and essays are a small part of the research that went into my book, The Intelligence Illusion: a practical guide to the business risks of Generative AI, or are essays extracted from it

8 May 2023

About the worrying polarisation that’s taking place in AI research, where it looks like the more pragmatic and sensible voices are getting ostracised by the rest.

The polarisation of AI discourse serves nobody except power

I also wrote about how ChatGPT and OpenAI is overpriced because we have underestimated its defects and security flaws.

The poisoning of ChatGPT

Turns out that language models can also be poisoned during fine-tuning.

For more of my writing on AI, check out my book The Intelligence Illusion: a practical guide to the business risks of Generative AI.

You can also find me on Mastodon and Bluesky