This post is about my thoughts on AI rather than a practical guide to its use.

I put off writing this for a while. If you’re a software engineer you’ll be well aware of the coding assistants taking over our poor IDEs. If you live under a rock you might not be aware of the “AI revolution”. At work I have been pressured into taking up using AI for everything. This is keeping up with competition. It has significantly changed how I work. Rather than routing through a codebase and working out which metaphorical nails to hit (where I need to write my code) I am now prompting agents and working with them to complete my tasks. It has increased my productivity significantly.

However, given the state of AI in its infancy as it’s being used today, I can’t help but shudder.

I am not happy with the state of AI in the world. My main complaint is attribution and flagrant abuse of copyright. Anything else scraping and plagiarising all the world’s information would be illegal in any sense. My second complaint is pushing unwanted features to users. The list goes on but is not limited to: overall usefulness, dilution of output, rotting our already ill culture. When there’s so much slop out there it just makes everything bland or worse and useless when used as content. I suppose that’s a win for real creatives as their work now shines brighter because of it.

AI will change the world. It already is. I imagine easier interactions with our current systems. Our daily drivers with software are awful. The interactions I have to make remind me why I do my job. UIs are terrible, UX is an afterthought. I shudder when I have to register for a new system to pay my bills or manage something. I just know the system I have just signed up to is going to be a joke to use and I’d be better off filling in a paper form and handing it to a human. AI will circumvent this and let us interact with our systems with greater fluency.

AI will streamline everything we can imagine. But it needs to do it tastefully. In the short term sacking droves of support staff may seem great for the bottom line but the technology is not ready. At the very least the infrastructure to support these agents isn’t understood. There is a human cost to this. Someone working in a call centre can transition into supporting these models and working with AI, but when I see the headlines about reducing headcount for quick wins I just imagine an economic bubble bursting to the detriment of us all (apart from the financially resistant class, but that’s another essay…)

There are problems but they’re not the problems of AI. When adopting AI we need to emphasise care and ethics. Sadly the progression of technology has not got a good track record for this. Despite all of this I remain hopeful.

My disclaimer:

• I do not use AI for writing on this website
• I do not use AI to code this website
• I do use AI to fix bugs and add features on my personal projects
• I do use AI extensively at work for everything

I may write up my practical uses for AI (there are many). I should better explain my AI usage for this website (although currently close to nil, I do use AI for searching and exploring sources). I will (at some point) explain my rationale for using AI in other personal projects.