What we call “AI” today are large statistical models trained on language, images, videos, and other inputs. When I refer to “AI,” I am referring to these statistical models.
Short Answer
- No AI-generated content on this blog.
- Language models for grammar checking. Current tool: Grammarly with generative AI tools turned off.
AI For Writing
I do not use AI to generate text, images, videos or any other consumable content.
I do use a grammar editor that has large language model capabilities. I avoid the generative AI features. The times I have dabbled with them, I find they weaken my writing and make the overall tone generic and uninteresting.
However, as someone with dyslexia, I rely heavily on spelling and grammar corrections.
My current grammar checker is Grammarly, and when my annual subscription runs out, I am going to switch to Language Tool. While Language Tool is slightly less polished, they have a better privacy policy and is far less pushy on generative AI.
On Hallmarks of AI
During my education, I was trained to treat em dashes, semicolons, and tricolons as powerful writing tools. As you can see, I still use tricolon today. This is my own rhetorical style, and I have no plans to change it unless I want to change the way I convey my thoughts.
AI For Research
As far as I know, every search engine uses a language model to infer search intent and return results from the indices, which happens regardless of whether there is an “AI” answer displayed on the page.
Conversational search is becoming the new norm, and LLM usage is difficult to avoid on the web.
Occasionally, I use generative search results to get a high-level answer when the initial search results are hard to parse. I follow this by verifying the answer with a few sources.
AI Ethics
I am deeply unsettled by the current AI obsession and the detrimental impacts it has on the environment and social inequality. Many stronger authors than I have written about the damage that is being done.
Advertising and AI: Adversaries that Degrade Information Finding
I can’t talk about AI without talking about advertising. Over the years, the web has degraded into a strangling network of cookie consents and ads. On any given search result, the top few links bombard you with ads and newsletter signup modals.
AI answers in search results are a noxious antidote because they filter the noise of ads but control whose knowledge is shared.
I do my best to link to high-quality resources, regardless of whether there are ads on the page. Sites like Etymonline are an invaluable resource, and they have chosen ads to sustain their presence, with options for becoming a member to remove ads.