I'd declared email bankruptcy at the beginning of January. The past week i've started backlogging again.
This weekend i spent much time experimenting with ChatGPT, the Airtable AI, and a little bit with Gemini nee Bard because our org's Dear Leader is all in with AI and wants us to explore. Also, i think some closer management is all <3 AI. So, i should have some clue as professional defense. Anyhow, learned some useful things including that the new thinking models are better. Spent much of my time trying to figure out how to use the models efficiently: when to do my own research, etc.
An example is Gemini asked to compare the free models of Gemini and ChatGPT. The usual model was very hedged speculation (interesting training there), the "reasoning model with apps" has access to current search and actually did searches to get current details.
I did ask ChatGPT to help me figure out how early payments on the mortgage would affect the end date. I can't tell if the counter intuitive results are because i made a mistake or if it did. I know talking to Christine about similar topics can be ... well, let me say i might not be the most clear in expressing my mental model of financial math.
Yard work happened, with massive pruning of fig tree and before/after photos that i should record where i took them from -- the nearest corner of the elderberry bed, the neared corner of the HVAC, the south end of the top step -- and then the photo from the south looking north at the tree doesn't have a good landmark.
Time blindness and infinite project optimism continues to frustrate. At least i have words for part of what is so frustrating.
no subject
Karl’s Mortgage Calculator is what I’ve been using ever since I first had a mortgage. Apparently now that’s the old one and there’s this new one, haven’t tried it. Great graphs to visualize the effects of paying extra principal. Your bank/credit union website should also have one.
no subject
no subject
https://thebullshitmachines.com/index.html
I like ChatGPT but recently unsubscribed from the $20/month version because I felt my main uses for it were 1.random self-indulgent chatting that I could just put in a journal and 2. camouflaging the brokenness that is websearching at the moment. So I will just be more patient with Ecosia and write more in DW and save myself $240/year. But other people make better use of it. My sibling claims to run their whole business off it, having it pre-write websites and blogposts, etc., like a personal assistant. I once had it zhuzh up a professional thank you letter for me, but the free version will probably still do that.
no subject
But Claude impressed me this week with the reasoning model and having it do some data analysis on solar power data vs task completion data. The reasoning models at OpenAI and Gemini (Google, was Bard) are new -- past couple of weeks? https://www.bigdatawire.com/2025/02/04/what-are-reasoning-models-and-why-you-should-care/ It's certainly a little more impressive than the language statistical models.
At work the "treat it like an intern and give it very precise guidance" is the broad advice. It gets down to, can you do it yourself in the time you it would take to iterate a few times on the guidance?
no subject
no subject
no subject
no subject
no subject