Modern AI, AKA Generative AI, AKA LLMs are here, and it seems like they’re here to stay. That’s just a matter of fact.
I am using LMMs in my personal and professional lives. I don’t deny there’re huge social, economic, environmental and cultural implications to this technology, and that by using it I’m actively contributing to some of them.
Not by a lot, granted - I’m not a heavy user, and am very selective about when and where to use it - but I am undeniably “part of the problem”. Again, that’s just a matter of fact.
Below is the AI policy for materials on this website.
Specificaly for written content - Input-wise
I use LLMs for research in many ways:
- Broadening my inital reach
- Making unknown unknowns into known unknowns
- Hunting down references
- Quick lookups and background summaries
- Quick smoke-test things I vaugely remember
- Misc other uses
Having said that, I ALWAYS:
- Validate references exist, and that they hand-wavingly match what I thought.
- Read the actual source material myself if I decide to use it for more than a reference (or reputable secondary sources about it1 For example, I read parts of Kant, and read books and articles on the rest of his work. For this type of blog, that’s what you get. ).
- Try to understand if the referenced material was itself AI generated or augmented, and note that.
Specifically for written content - Output-wise
I will sometimes use AI assited coding (or even AI driven coding) to generate code examples when relevant. Those will be tested, and usually be incorporated as-is.
Outside of that, all the content I publish will be written by me2 Automated spelling and grammar corrections notwithstanding , a human, for you, a human reader3 If you feed my content to an AI summarizing tool, please go back and go though the original phrasing, and it was deliberately chosen (maybe poorly, but still) .
I never copy-paste AI generated summary or text, or use AI generated writing artifacts wholesale.
As I do use AI for background research and am undoubtedly affected by it, my wording may sometimes incorporate patterns endemic to LLM outputs. That’s mainly because I’m still working on solidifying my own voice and style (so ironically, if I sometimes have LLM-esque phrasings, it’s because I’m only human on the inside).
Other AI uses
I’ve used AI in the technical stack for this site:
- The site’s logo was designed through Claude-Design.
- Some of my code customizations were AI assisted, coding-wise.
If and when I incorporate an AI augmented4 Standard retouches / sharpening tools notwithstanding. or generated image / audio, I’ll always note that clearly.