I've found this to be the case as well. My typical workflow is:
1. Have the ai come up with an implementation plan based on my requirements
2. Iterate on the implementation plan / tweak as needed, and write it to a markdown file
3. Have it implement the above plan based on the markdown file.
On projects where we split up the task into well defined, smaller tickets, this works pretty well. For larger stuff that is less well defined, I do feel like it's less efficient, but to be fair, I am also less efficient when building this stuff myself. For both humans and robots, smaller, well defined tickets are better for both development and code review.
Back in 2019 I created a Google Spreadsheet titled "family debts" that allows my family (4 siblings and my parents) to record when we owe each other money, and periodically settle up. I later learned that I recreated Splitwise, but having something like this with trusted folks has been hugely useful. We have over thousand entries, and use it constantly for splitting gifts, buying something at the store for someone, etc.
Om Friday after Thanksgiving I spent half a day building a telegram bot that accepts an address and a list of Amazon links, and in turn orders the item (at a discount since it uses my Amazon credit card), and adds it to the above "family debts" spreadsheet.
I really like the idea of programmable, trusted lending like this, and feel like it could be extended to other groups that you implicitly trust.
If it's an all out race between the different AI providers, then it's logical for OpenAI to hire employees that are pre-trained rather than training up more internally.
Switzerland's wealth tax is a lot more reasonable though. UK is 2%, Norway is 1.1%, the Swiss ones are all below 1% (depends on your canton), ex the Zug canton is 0.21%.
In normal cases I mostly agree - you can just charge it while taking a shower or something that you do every day. That said, Garmin's battery life shines when you are using it on a non-normal day.
Overnight camping, and sleeping in a tent for a few nights is a good example. I'm not "taking a shower" and hence don't really have a great time to charge it. With my garmin I just leave it on, and it keeps working for the entire trip.
Same thing with other "adventure" travel, flying overnight, etc.
1. Have the ai come up with an implementation plan based on my requirements
2. Iterate on the implementation plan / tweak as needed, and write it to a markdown file
3. Have it implement the above plan based on the markdown file.
On projects where we split up the task into well defined, smaller tickets, this works pretty well. For larger stuff that is less well defined, I do feel like it's less efficient, but to be fair, I am also less efficient when building this stuff myself. For both humans and robots, smaller, well defined tickets are better for both development and code review.
reply