How I Use LLMs, June 2026

the above to:

Nano Banana;
Gemini recognized and remembered Dusty's name, but didn't remember Cody's. 

Nothing below really takes full or deep advantage of what LLMs can do. Here is a new bit from Andy Masley on working with Anthropic's Claude Fable. That piece really gets to some of what can be done at the (current) frontier. 

Here is a good Reddit thread: What is the most underrated use of ChatGPT in daily life?

I use ChatGPT Business (free via a credit card) and Gemini Pro (free because I bought a Google Pixel phone). I don't inherently trust an LLM. Even today - June 2026 - hallucinations are quite prevalent, at least for me. Their memory pretty much sucks, and they insert random things. Knowing I'm vegan leads them both to assume I'm hyper-green / love the Earth / care about eating organic, etc. Ugh.

Anything important I double check. I often ask both ChatGPT and Gemini the same question and play them off against each other. They frequently disagree. 

I've done several big projects with AIs:

1. An animated vertical video for One Step for Animals. This took days.
   
   
2. Designing, promoting, and analyzing a survey for One Step's Spanish-language campaigns. This was done for the donor who currently underwrites that campaign (along with the donation match). The results were very similar to the English survey.
   
   
3. A big research project last year attempting to compare the relative efficacy of nonprofits in a space unfamiliar to me (not directly animal related, and coming out of an ongoing conversation with a long-time friend).
   
   
4. Ongoing discussion of our personal asset allocation of retirement savings, given my general distrust of bonds, as well as our need to keep our income well within the ACA subsidies cliff and keep One Step alive as long as possible.

Other projects / ongoing threads in ChatGPT include: 

• Four upcoming or hopeful vacations over the next four years (I also used it extensively to plan and prepare for our 2026 trip to Japan).
  
  
• Catch-all Sleep / Health. Dozens of threads here, including:
• A specific ongoing conversation regarding a bizarre, major medical screw up last year.
  
  
• Scotch (and other whiskies, including when I was in Japan). 
  
  
• Car issues (e.g. do I need new tires / which tires to buy from where, when to replace our current vehicle, what's coming down the pike and what current cars have that our old one doesn't).
  
  
• How to replace a light fixture in our garage, and many other house threads (roofing and carpet cleaning companies comparisons, what to check to make sure our fridge is optimally placed, installing a camera doorbell, etc.). 

But on a day-to-day basis - and I use it every day - I turn to LLMs for many things; a random collection from my ChatGPT sidebar:

• Possible purchases (comparisons, alternatives) 
• Eye twitching causes
• Substack and profanity
• Bird or plant identification
• Plant diagnosis and care
• What credit card to get next 
• What credit card points to hoard for a specific "unicorn" future flight 
• Sokal Affair explanation (I couldn't remember the name - just the general event - but it was a friend who told me about Sokal Squared) 
• CO2 impact on crop yields (for this)
• Itchy arm diagnosis help
• Recovery from (and writing about) back injury
• Writing feedback (don't generally like that feedback - LLMs would hate this post, I think)
• Checking server configuration and Mailchimp settings to improve email delivery
• Drafting messages to doctors, insurance companies, and customer service
• Summarizing long pieces that aren't important / worth my time
• Bernoulli Principle clarification (super-interesting - not what I learned in my undergraduate Aerospace program; research prompted by a pilot friend)
• Best local movie theater for seeing Project Hail Mary (we hadn't gone out to see a new movie since the pandemic - I had no idea some theaters had gotten so fancy!)
• Where to get something (OJ, wine, vegan take-out) near our hotel or Airbnb
• Troubleshooting computer problems (e.g., replacing a failing harddrive)
• Vitamin C and other supplement questions
• Cat photo manipulations

Much (but not all) is basically using them as a meta-search engine. LLMs are good at compiling and summarizing information from diverse sources, saving me a lot of time.