I actually had not used TurboTax before. And as a result I could do state and federal free with them this year, so it was cheaper than freetaxusa.

But won’t be using it again next year, it was fine but not particularly impressive compared to usually cheaper alternatives.

Hey LLM. I’m thinking of deducting my Corvette as a business expense for my landscaping business, is that a good idea?

What a creative way to lower your tax burden! This totally makes sense and you can be confident that your decision will be well received.

(Others can take the LLM tone better than me, and I don’t have the patience for LLM verbosity).

I wonder if I counted…

So I did the tax prep using a free offer from TurboTax. Everything seemed traditional.

Then, at the end it generated an AI summary of my return. I didn’t have a choice, it just did it. I have the “unhelpful” feedback because:

• Despite saying it was “explaining” the numbers and why, all it did was just list the numbers from the fairly straightforward table right above the AI response in a more awkward form, not explaining anything.
• Further, despite the seemingly easy task of “Take a table of figures and repeat them in prose”, it still screwed up and messed up and of the figures that all our had to do was repeat verbatim.

So AI was forced into my tax prep and did nothing substantive (thank goodness) and flubbed the cosmetic role it tried to play.

Problem with the theory is that people believe in LLM strongly enough that whatever pressure there is within a market to be vaguely similar evaporates. SQL certainly has dialects, but at least the basics are vaguely similar, as an example.

Working with a vendor that is oddly different from every other vendor in the space and we applied pressure to implement more typical interfaces. Their answer was “just have an LLM translate for you and use our different and frankly much weirder interface”. When we did cave and use it and demonstrated the biggest LLMs failed, they said at least they give you the idea. Zero interest in consistent API with LLM as an excuse.

On the write your code for you, it has to be kept on a short leash and can be a nightmare if not overseen, though it can accelerate some chore work. But I just spent a lot of time last week trying to fix up someone’s vibe coded migration, because it looked right and it passed the test cases, but it was actually a gigantic failure. Another vibe coded thing took 3 minutes to run and it was supposed to be an interactive process. The vibe coded said that’s just how long it takes, if it could be faster, the AI would have done it and none of the AI suggestions are viable in the use case. So I spent a day reworking their code to do exactly the same thing, but do it in under a second.

For the jira ticket scenario, I had already written a command line utility to take care of that for me. Same ease of use instead of using jira GUI and my works torturous workflows, but with a very predictable result.

So LLM codegen a few lines at a time with competent human oversight, ok and useful, depending on context. But we have the similar downside as AI video/image/text creative content: People without something substantial to contribute flood the field with low quality slop, bugs and slow performance and the most painful stuff to try to fix since not even the person that had it generated understood it.

Yes, the key thing is it might have extracted useful info from otherwise confusing data, it might have mixed up info from the data incorrectly or it might have just made it up.

So it can be useful, if you can then validate the info provided in more traditional means, but it’s dubious as a first pass, and sometimes surprisingly bad when it’s a scenario you thought it would work well at.

Eh, I’m not really interested in disc based copies, really the disc is there for ripping and then stored, jellyfin to stream it to watch as I please. Once ripped then I can handle the resultant file nice and easy.

Pro tip to BK: I probably wouldn’t even notice the lack of ‘please’ and ‘thank you’. I would, however, be significantly happier if you stopped making them say “You Rule”. Seems like they have to say it as both greeting and a “your order is finished”. It’s just unpleasantly cringey.

It’s a bit trickier last time I did it to be confident I can rip a Blu-Ray.

I actually don’t want to juggle discs to watch stuff, I like the general concept of streaming, but I don’t like paying eternally for it, for shows to jump between providers and for my access to cut out part way through and/or even if I have the new service, my progress being forgotten so I have to try to look for where I left off.

So I want to rip content. DVDs are always dead simple. As I rip blu-rays, MakeMKV is kind of a hassle, it wants to expire itself all the time, and like right this second the place to update from seems down. Maybe someone will comment with some easy way to rip blu ray that internet search doesn’t make obvious.

If folks sway me, might go buy a 4k friendly Blu Ray drive and hop to it.

Instead of new SSNs, how about we maybe the number less risky in general?

It should never have served as a “secret”. Authenticating someone needs more than some account number. SSN should be more of a “username”, not a password.

Look at the feature list bragged about. It’s really simple stuff. I can absolutely believe they vibe coded that stuff.

The “hardest” one was to feed listener history to an LLM and have it generate a playlist based on the titles. That’s such an absurdly trivial thing to do.

It’s not rocket science. It’s a trivial streaming music player.

And even if you babysit it and carefully tell it all the mistakes, it will learn nothing and suggest the same stupid mistakes next tim. I did actually know a human just like AI and he kept his job for years before quitting to grift another company because management refused to believe he sucked. So I’m not optimistic about AI screwing up discouraging business leaders.

Or even to true believers, they can just vibe code up just as good of an app.

It basically declares point blank that the technology does not matter, it’s their marketing and music rights only that matter.

Well, they are helping out with that one…

Some of the reactions are some in an effective way, and I assume this example is one of them. The problem being evidently they didn’t think any what might be in big base64 blobs in the PDF, and I guess some of these folks somehow had their email encoded as PDF, which seems bonkers…

Certainly his use of LLM was stupidly egregious, but he found that even by those standards the math results underpinning the LLM were way off.

Don’t worry, while they may have worse business results than a lot of other companies, displaying steeper loss of market, whose non car initiative have failed, whose leader squandered his reputation to throw in with a political movement that hates EVs in hopes of political clout that evaporated within a few months…

They still have a market cap bigger than all their competitors combined, because why not…

It’s a bit directly on Microsoft, unless you go out of your way, bitlocker will upload the keys to Microsoft. They assume you want them to help recover your data if your tpm becomes unavailable.

Interesting fun fact, when I tried to swype type bitlocker it really wanted to put bootlicker instead.

As I said. It’s an extrapolation of the rules from once upon a time to a totally different approach. It’s marketing and increasingly subjective. Any number can “make sense” in that context. The number isn’t based on anything you could actually measure for a long time now, it’s already a fiction, so it can go wherever.

To be fair, the industry spent decades measuring a distance, so when they started doing features that had equivalent effects, the easiest way for people to understand was to say something akin to equivalent size.

Of course, then we have things like Intel releasing their "10 nm* process, then after TSMC’s 7nm process was doing well and Intel fab hit some bumps, they declared their 10 to be more like a 7 after all… it’s firmly all marketing number…

Problem being no one is suggesting a more objective measure.

For a while now the “nm” has been a bit of a marketing description aiming for what the size would be if you extrapolated the way things used to be to today. The industry spent so long measuring that when the measurement broke down they just kind of had to fudge it to keep the basis of comparison going, for lack of a better idea . If we had some fully volumetric approach building these things equally up in three dimensions, we’d probably have less than “100 pm” process easily, despite it being absurd.