I asked an LLM (Gemini) about a calculated field in my taxes that was wrong but I couldn’t figure out why and every time it tried to tell me something like “It’s a common glitch for tax software to calculate like this.”
When I did figure out what was wrong and asked if that made sense, it told me I was absolutely right though.
I think people are lucky the IRS fired all their employees this day and age so this work isn’t getting checked as much.
You're being deprived of government services and a sustainable economy because the wealthy are allowed to game the system they've rigged for themselves. I wouldn't call that luck.
Gemini is a pretty bad choice for an LLM. Most people using it are doing so because Google bundled it for free with a couple things, not for its quality.
Serious question: to figure out the answer why did you not go to the IRS website for that field? This is how I ultimately answer any tax questions I have.
Once every single thing connected to the internet (iPhone + 10 years?). Tech products, software, video games, etc. were new and exciting ways to solve problems. Now every single thing is a way to monetize, steal your data, or lock you into a platform to do those things later.
To fill the gap left by all the sites being blocked, the company intends to offer access to a library of religious content, including AI-generated Bible videos.
Truly a serious and spiritual company. Maybe you can chat with AI Jesus instead of going to PornHub.
Since last year Amazon doesn’t let you download your books as files from the web anymore. You have to sync purchases to a (presumably supported) Kindle device. I have a jailbroken Kindle and was shocked at the added hoops to read a legally purchased book.
Obviously the solution is to stop supporting Amazon. Some authors unfortunately only sell ebooks there.
Yep. I moved off Kindle onto kobo several years ago. I would still buy books on Amazon if they were only available there, and used dedrm to move them to the kobo. That doesn't seem possible anymore, so I guess I just won't buy anything from Amazon now.
This could just be a skill or wrong use case thing, but do you only use spreadsheets for pure number-crunching? I've played terminal spreadsheets, mostly sc-im, but I often have some longform text field (like 'Notes') that becomes more fiddly to deal with than a GUI.
Visidata is the only terminal program I've found that handles large text fields in tabular data nicely the way you can drill down into a table row, then Ctrl+O to edit a field in your editor, but it's not a spreadsheet.
I have AI automation metrics that are tracked. We have automation that automates the AI usage whether you want it or not so our metrics are 100% for everyone. Presumably this makes the stock go up somehow.
I know this is nothing new, but it's insane that we need policies like "When talking to us you have to use human words, not copy pasted LLM output" and "You must understand the code you're committing."
When I was young, I used to think I'd be open minded to changing times and never be curmudgeonly, but I get into one "conversation" where someone responds with ChatGPT, and I am officially a curmudgeon.
Brazen usage of LLM output is a disrespect to the target audience to begin with. If I'm being expected to employ the mental capital needed to understand the context and content of your writings, I at the very least expect that you did the same when actually authoring it.
It also feels like using one of those cereal encoder wheels, to some degree. If someone sends me 10 paragraphs of output from chatGPT, and they only wrote a sentence to prompt it, then the output is really just a re-encoding of the information in the original prompt.
Quite literally - if they sent me the text of the prompt I could obtain the same output, so the output is just a more verbose way of stating the prompt.
I find it really disrespectful to talk to people through an LLM like that.
Generally speaking, a person can write a long rambling email without much effort. It takes some work to distill it down to keep the meaning without the verbosity.
If anything, AI should be used to take the long rambling email and send off the shorter distilled version.
I hope it becomes as accepted as it is to stick cameras in random people's faces: generally seen as rude, and bad actors who do it anyway are desperate and considered as such.
I am capable of copying and pasting shit into an LLM, do not give me its output and don't insult me by pretending the output is your own work.
The root of the issue is "ChatGPT people" are using artificial intelligence to replace... intelligence. Nobody, not even ChatGPT people, wants to actually read that drivel.
Alternatively, join my meetings on time. You click End Call, then Join. It takes 3 seconds.
You get Outlook reminders 15 minutes in advance. Webex/Teams notifications 5 minutes in advance. I’m sure you can make your watch vibrate or something.
People at my office join every meeting 5 minutes late because no one expects meetings to start on time anymore. So I guess we’re following this advice in all but the nominally scheduled time. Drives me nuts.
I'm absolutely baffled by colleague who somehow manage to be five minutes late to an online meeting while working from home. Because you're right, you get a reminder 10 - 15 minutes in advance, you just need to click the join meeting button, you're already at the computer. We have, for remote meetings, a five minute buffer at the start of every meeting, for people to "settle in" makes no sense, just start the meeting.
In general a lot of people just aren't being serious about meetings, which I guess is also why many hate them. So key indicators of a bad meeting is: runs more than 60 minutes, no meeting plan, documents or talking points provided in advance, more than five people (unless the meeting is more of a briefing).
> So key indicators of a bad meeting is: runs more than 60 minutes, no meeting plan, documents or talking points provided in advance, more than five people (unless the meeting is more of a briefing).
Seems about right, but wouldn't you agree that the majority of those meetings either could have been an email, or could have been handled in 20% of the time, if they had been planed?
Maybe, I think most people just don't give shit about any of this and for them wasting an hour feels "productive", like they've achieved something. After all, nobody can hear your beautiful voice over email.
TBH unless the meeting has a clear agenda and not just a vague title, I only join it when someone mention me. This allows me to be able to actually work and/or take breaks.
Last I used Visidata, it didn't play nice with fields like %sort (they'd disappear if you re-saved the file) and if you had two fields with the same name in one record they'd get combined into a single field like "Name[2]:" when you re-saved. It might've also killed comments? I'm certainly not surprised it only has basic recfile support, because who use recfiles, but I'd be careful using VD with them for anything but viewing.
I haven't put the work into supporting full round-tripping, so yes, at the moment it's mostly useful for reading/viewing. If someone files an issue that would likely go a long way towards getting better support!
When I did figure out what was wrong and asked if that made sense, it told me I was absolutely right though.
I think people are lucky the IRS fired all their employees this day and age so this work isn’t getting checked as much.
reply