My Use of AI
I think it’s important to be transparent and intentional about how we use AI. Here is how I do and don’t use AI on this site and in general.
General views
In general, I am not against AI as a concept and as a technology. However, I do have some serious reservations about how it has been deployed in society (flagrant disregard of consent and privacy in many instnaces) and about the motivations of those driving the adoption and funding the sector. I think there is massive potential for harm without the requisite checks, balances, and preventative measures, and there simply has not been enough time to develop, legislate, and implement those; the rate at which AI has been deployed in comparison to these safeguards is, frankly, wreckless. (Perhaps I’ll say more about this in a future blog post.)
No AI on this site
There is absolutely no embedded AI in this site at all. This is a completely statically generated site, with very few dynamic elements (e.g. the contact form and fetching my live Lichess ratings).
There is also absolutely no AI-generated content on this site.I am very against AI-generated content (videos, music, articles, etc), and so my commitment is that none of the content on this site is or will be AI generated. I write everything myself — no drafts by AI — and will continue to do so. (On a related note, I was using em dashes long before ChatGPT was invented, so — while that seems to be a tell-tale sign of AI-generated copy — please don’t let that fool you.)
How I do use AI
The above said, I do use AI in a couple different ways. Mostly I use GitHub Copoilot for inline auto-complete suggestions during development. This is tantamount to fancy auto-complete, which existed before the advent of LLMs. I make extremely minimal use of “agentic” tools; these are generally only for very small or tedious tasks. I try not to use these, as they do require a lot more compute, and thus have a much higher environmental tax.
If you have any questions, please don’t hesitate to reach out.