As the months have gone on, I’ve grown increasingly uncomfortable with my use of AI. To her credit, my wife, an artist and author, has been against AI from day dot. Go figure, there’s a reason she’s one of the people I admire most.

I’ve said before that I fundamentally disagree with generative AI and stealing the work of artists. I do not ask AI to generate images for me. I also don’t ask AI to write my blog posts for me. Everything posted here is something that started life in my real-life, squishy, jelly of a brain, and I didn’t run it through AI at any point. I also don’t agree with “vibe coding” and offloading creation to AI.

However, it’s such a slippery slope. Occasionally i’ve asked ChatGPT whether an email I want send or a LinkedIn post I want to write comes across in the way I intended, and before I know it I’m making changes I wouldn’t have otherwise. The pressure to get it right is maybe more relevant when I think of my professional identity, and it is genuinely helpful to have a sounding board before I put my raw thoughts out into the world, less I experience the embarrassment of asking another individual to look at them. But recently I’ve been feeling that even the lines between my personal and professional identities should be blurred.

Full disclosure, while I built this website myself, I used ChatGPT to help me build my family history website. To be clear: I didn’t ask AI to build it for me, but I did rely on AI to help me figure out how to build it. The fact that I have to make that distinction clear though shows just how uncomfortable I am with admitting I used AI.

For my family history website, I started out by asking Lovable to create a mock-up of what such a website could look like. There’s so much complexity that I didn’t know where to start. Colleagues have also been experimenting with it at work and it felt like the tool for the job.

I then asked ChatGPT how I could build such a website with Statamic. I’d already built this personal site myself with Statamic, but this felt like a much more complex project. I spent a lot of time asking questions about how the logic could work - genealogy is complicated, with lots of relationships to model - and it felt reassuring to explore this before committing to any one path. There were definitely some insightful ideas there that I appreciated, but I think it was me seeing the potential, rather than ChatGPT giving me the finished article.

When it came to deciding how I would enter information (blueprints) and how it should look for visitors (templates), I created a lot of it myself. However, I did quite frequently ask ChatGPT what fields it thought I needed or what code I could use to produce what I intended. Basically, I’m saying that, while I didn’t ask ChatGPT to build me a website, at times I was “vibe coding” without even realising it(!)

But, regardless, ChatGPT gave me the wrong information. It would double down on features that don’t even exist in Statamic. It would repeatedly produce comments in a format that crashed my (local) website, despite me telling it to stop this. Ultimately, it really struck me, towards the end of active development, just how little I’d actually referred to the official Statamic documentation. This was so different from when I was learning Eleventy, Wordpress, or Unity.

I did not, at any time, ask ChatGPT to write any content for me, and I’m relieved about that. Any time I gave it information about how I wanted my ancestor’s information to be used, I changed all the personal data for dummy information. I was at least aware enough of the risks of letting it have our personal details.

This has come to the forefront of my mind again recently: I’m literally in the middle of a thread I started on a genealogy forum, explaining my discomfort to people who seem happy to paste entire forum threads of others comments about their family history into AI. It's the same attitude of people who feel they're doing you a favour if they "improve" your photos without being asked by running them through AI.

I'm proud of the work i've put into my family history website, but it's also not online yet. In an age where AI can swallow up whatever it wants, i'm seriously reconsidering whether I even want to share my website publicly. No-one decided what AI should have access to, it was just given the world’s knowledge without permission. I don't feel comfortable with the idea that it can just digest my family history without my say so. I don't know. I'm not sure it's really that far removed from giving any megacorporation all this information. I'm still thinking about the best way forwards.

I do think ChatGPT gave me the confidence to attempt this massive project that I wouldn’t otherwise have had, and for a long time I was able to rationalise it as a tool that I was using as a supplement to my own skills and imagination. However, increasingly I’ve been feeling like I lost something.

Anyone who knows me would describe me as “an ideas man”. If someone has a problem, the gears start whirring and I can already picture multiple solutions. Genuinely, it’s not me being bigheaded to say that I can dream up several ideas before most people have had their breakfast. That’s just the way my brain works. If I actually managed to create everything I can picture in my mind.. well there’d be a lot more things in the world(!)

But using AI has denied me that opportunity. I’ve finally realised that. For all the creativity I put into imagining my notebook layout or the cool things I can do with this new OS style interface, using AI at any point robs me of the opportunity to learn things for myself. I’ve been copying into my digital garden things I’ve discovered through my own research before I started using AI. Problems I encountered and the solutions I arrived at. For example, understanding Eleventy-img. I love learning how things work - I could never ask ChatGPT to give me something and be happy with it, I need to understand it - but using AI is such a shortcut.

I remember, when I was at primary school, there was a class competition. I can’t remember the exact brief but it had something to do with Charles de Gaulle (Scotland has strong links with France and we were learning all about them). I remember visiting the library, reading dusty books for hours, and trying to paraphrase what I’d learned into a handwritten project workbook. I even remember that I tried to use our dial-up internet but it crashed so much that the library was actually easier.

I also remember that the classmate who won the competition had chosen the same topic, but they’d photocopied information from the books, glueing whole passages of text into their project submission. In comparison, what I created probably wasn’t amazing, but it was the result of genuine human effort. I remember feeling that the classmate who won the competition had taken an unfair shortcut.

That’s what AI is: it’s a shortcut. I feel like using it denies me the opportunity for critical thinking. It's making me unlearn how to be comfortable sitting with problems for days or weeks on end.

Not only does using AI diminish my own creativity (despite it regularly telling me how insightful, thoughtful and oh so wonderfully creative I am, of course), it’s also disconnecting me from the wider world. I read a fantastic piece this morning by Sightless Scribbles about The Colonization of Confidence. What struck me the most was how it suggested AI erodes our faith in our own abilities.

I’ll readily admit that it’s felt comforting to explore thoughts and problems in private conversation with AI (note: I know it’s never really private), but in doing so I’ve missed out on the opportunity to actually talk these thoughts through with another human being. I have real life friends I can trust with my vulnerability, I feel safe sharing my thoughts on the 32-bit Cafe Forums. Even this blog has always been a space to be my messy, imperfect self.

If I start sharing more of myself with AI, not only will I lose myself, I will lose the opportunity to strengthen the real life relationships I enjoy.

One of the best decisions I made last year was to stop using Instagram. I felt lighter and my experience of the web was slower, more intentional. However, over the last few months I’ve replaced it with ChatGPT without even realising it.

For all my principles - and especially now that I believe so strongly in owning my own content and leaving walled gardens behind - using AI has increasingly felt incredibly at odds with my beliefs and the kind of world I’d like to live in. I’m also not happy with the environmental impact, but have been able to rationalise that I’m somehow using AI more sustainably because I’m not asking it to make an action figure of me.

You spent so long thinking about whether you could, you didn’t stop to think about whether you should - Dr Ian Malcolm, Jurassic Park

Please don’t misunderstand. Me declaring that “I’m over AI” isn’t some grand moral stance. I fully admit that I got it wrong. I thought I was using AI responsibly, that I was somehow master over this tool, but I was fed the pipe dream just like the rest of the planet. I lost control without realising I never had it in the first place. I got it wrong.

Moving forwards, I’m going to do what I told myself I was already doing: learning in public. I’m going to share my raw, unfinished thoughts and how I work through problems, not just here but on my other websites too. For example, I recently wrote a post on my new professional website that I know won’t land well with everyone, but it’s vulnerable and honest. If id asked AI to write it though, it would be a shadow of my true feeling and I’d miss out on the opportunity for growth.

I will undoubtedly make mistakes with this approach, but I’d rather be known as someone who can own up to that, than someone who only shares the glossy, sanitised version of themselves with the world. I read JA Westenberg’s excellent article, Blogging in the ruins, this morning. The idea that blogs are the ideal format for works in progress and an on-going conversation with oneself really resonates with me.

Note to my future self: you just walked away from your ChatGPT account for a reason. If you ever feel the itch to use a similar service again, please see above.

Updated note to my future self: you signed out but you didn’t delete your account. It was all too easy to start using it again. Well you’ve just properly nuked your account and you feel lighter. You don't miss it.

Final(?) update: You're not finding it easy to stay away from AI - it really is everywhere at the moment - but keep taking it one step at a time.