A.I. policy

Introduction

The over-hyped umbrella term ‘A.I.’ is used to describe all manner of different technologies that are certainly artificial, but are definitely not intelligent. Some of these technologies have the potential to make the world a better place; the vast majority of them won’t—and, I believe, will do more harm than good. In general, I aim to avoid using almost all so-called ‘A.I.’ technologies, and I encourage everyone else to do likewise.

Use of so-called ‘A.I.’ in my writing

I do not—and will not—use so-called ‘A.I.’ to generate or amend writing for me. This applies to both my published and private writing.

I write to work out and express what I think, or what I’ve learnt—hopefully for the enjoyment of others. When people read my writing, I want them to know they’re reading my own words, not the algorithmically plagiarised work of other writers.

Use of so-called ‘A.I.’ in my research

I will not use so-called ‘A.I.’ to generate summaries or any other content in the extensive personal note-making system that underpins much of my writing.

Although I don’t envisage using so-called ‘A.I.’ as a matter of routine for search purposes, I don’t rule it out entirely. A couple of examples where I think use of so-called ‘A.I.’ searches is acceptable:

Example 1: When I was revising a chapter for my book Through Darwin’s Eyes, I realised I’d somehow lost the reference to a scientific paper I’d summarised in my own words in the chapter. I was unable to re-find the paper in question using standard web searches, so I fed my summary into a so-called ‘generative A.I.’ chatbot and asked it to identify the illusive paper. The chatbot found the paper straight away.

Example 2: Certain apps on my phone use so-called ‘A.I.’ to allow me to identify animal or plant species in the field using the phone’s camera, or identify the calls of bird species using the microphone. I think such functionality is useful and valid.

Use of so-called ‘A.I.’ in my photography

As a keen photographer, I make extensive use of Adobe Lightroom and occasional use of Adobe Photoshop. Both of these product now have extensive so-called ‘A.I.’ functionality bundled into them. I will avoid using such functionality, except in the following cases:

• I routinely remove minor blemishes from photographs and occasionally remove non-prominent features from landscape photographs (e.g. litter on the ground, or people in the background spoiling the view). It is possible to make such corrections without using so-called ‘A.I.’—which is what I still usually do—but so-called ‘A.I.’ occasionally makes this process easier, and I see no harm in using it in this way.
• Similarly, so-called ‘A.I.’ makes it easier to remove or reduce unwanted grain in photographs, and I’m happy to use it to do this.
• I will almost never use so-called ‘generative A.I.’ to add features into photographs that weren’t really there. On the rare occasions I do this, I will make it abundantly clear what I’ve done and why I’ve done it. (For example, when trying to convince my brother-in-law to grow a beard, I used so-called ‘A.I.’ to add a beard to his chin in an existing photograph. I made it clear to everyone who saw the photo that the beard had been Photoshopped on.)

In terms of taking photographs, my DSLR camera has the ability to identify certain subjects (e.g. people, vehicles or animals) using so-called ‘A.I.’ and automatically focus on such subjects. I find this feature incredibly useful—especially when photographing fast-moving birds—and have no qualms about using it. I see it simply as a better form of traditional autofocus.

Use of so-called ‘A.I.’ elsewhere

So-called ‘A.I.’ is bloody everywhere these days. Hopefully the bubble will burst soon. Whenever possible, I disable so-called ‘A.I.’ features in all the apps and services I use. In some cases, I’ve even changed to different apps to avoid such features.

I will continue to resist using so-called ‘A.I.’ except in the cases outlined above.