Using AI to analyze health data has such a huge potential upside, but it has to be done locally.
I use [insert LLM provider here] all the time to ask generic, health-related questions but I’m careful about what I disclose and how I disclose it to the models. I would never connect data from my primary care’s EHR system directly to one of these providers.
That said, it’ll be interesting to see how the general population responds to this and whether they embrace it or have some skepticism.
I’m not confident we’ll have powerful/efficient enough on-device models to build this before people start adopting the SaaS-based AI health solutions.
ChatGPT’s target market is very clearly the average consumer who may not necessarily care what they do with their data.
> supervised by a human who occasionally knew what he was doing.
seems in jest but I could be wrong. If omitted or flagged as actual sarcasm I would feel a lot better about the project overall. As long as you’re auditing the LLM’s outputs and doing a decent code review I think it’s reasonable to trust this tool during incidents.
I’ll admit I did go straight to the end of the readme to look for this exact statement. I appreciate they chose to disclose.
Nobody who was writing code before LLMs existed "needs" an LLM, but they can still be handy. Procfs parsing trivialities are the kind of thing LLMs are good at, although apparently it still takes a human to say "why not using an existing library that solves this, like https://pkg.go.dev/github.com/prometheus/procfs"
Sometimes LLMs will give a "why not..." or just mention something related, that's how I found out about https://recoll.org/ and https://www.ventoy.net/ But people should probably more often explicitly prompt them to suggest alternatives before diving in to produce something new...
Neither do you need and IDE, syntax highlighting or third party libraries, yet you use all of them.
There's nothing wrong for a software engineer about using LLMs as an additional tool in his toolbox. The problem arises when people stops doing software engineering because they believe the LLM is doing the engineering for them.
Every IDE I've used just worked out of the box, be it Visual Studio, Eclipse, or anything using the language server protocol.
Having the ability to have things like method auto-completion, go-to-definition and symbol renaming is a net productivity gain from the minute you start using it and I couldn't imagine this being a controversial take in 2025…
> I don't know what “tarpit” you're talking about.
Really? You don't know software developers that would rather futz around with editor configs and tooling and libraries and etc, etc, all day every day instead of actually shipping the boring code?
Perhaps. But you also need to ask why Paxton is doing this as this case will vaporize as soon as that is accomplished. I would be much more optimistic if California were also signed onto this.
Paxton, however, doesn't give one iota of damn about individual freedom. So, this is either a misdirection, shakedown or revenge.
Unfortunately, we don't have Molly Ivins around anymore to tell us what is really going on here in the Texas Laboratory for Bad Government.
> So, this is either a misdirection, shakedown or revenge
This is about being in the news as much as possible. He is in a close 3 way race for the 2026 Republican spot for US Senate. The other two are current old-school conservative senator John Cornyn, and new comer MAGA Wesley Hunt (but not as MAGA as Paxton). Lots of in-fighting over funding, so Paxton is making sure to get in the news as much as possible.
Throughout the year he has been in the news for things that are useful like this and another suit against a utility company for causing a fire and others for typical maga things like lawsuit to stop harris county (Houston) funding legal services for immigrants facing deportation or immigrant-serving nonprofits or a "tip-line" for bathroom enforcement or lawsuits against doctors...it goes on and on and on. It's a page out of the Trump playblook, its like watching a trump clone. And thats the point.
While I agree (and I agree with the upstream comments, too), there's often deeper reasons why we can short circuit fully evaluating an argument made on its merits: often the "merits", or lack thereof, are derived from the party's values and beliefs, and if we know those values to be corrupt, it's likely that subsequent arguments are going to be similarly corrupt.
There's only so much time in the day, only so much life to live. Could a blog post written by the worst person you know have a good point, even though it's titled something like "An argument in favor of kicking puppies" by Satan himself? I mean, true, I haven't read it, yet. There could be a sound, logical argument buried within.
This is also what "The Boy Who Cried Wolf" teaches, essentially. Trust is hard-won, and easily squandered.
"A lie is around the world before the truth has finished tying its shoes."
"Flood the Zone" is why some of us are so exhausted, though.
In these instances, the argument has to come from someone who is self-aware enough of the short-circuit to say "okay, look, I am going to address that elephant" — but mostly, that's not what happens.
Thankfully in this case, all we need get through is the title.
I don't care about people's values, unless I am evaluating them; that's their own business, and I am not the value police or thought police. Goodness knows there are people (hi, mom!) who are appalled by some of my values.
Roman Polanski and Woody Allen: terrible humans, but they have still made some of the best films that exist.
Everyone is the value police, though, at some level. It is either cowardice or willful ignorance to pretend you don't have judgements about how other people behave, some of which might compel you to act in some way.
It's always important to read the fine print. That would be part of evaluating an argument on its merits. His lawsuit over Tylenol + autism is easily rejected on its merits. That means nothing about this issue.
I work for an R1 university that just launched Workday recently and it has been a total disaster.
Consultants + vendor pitch a nice shiny solution that handles everything & works flawlessly. In actuality it resulted in a net efficiency & productivity loss vs the homegrown systems we came from.
It sure did generate plenty of billables for the consultants though, who mind you, are still contracted over a year later.
This may be a controversial position but I actually enjoy using Firefox. Vertical tabs, better profile management, etc. have all been welcome things I had in other browsers (cough Arc cough) that have made it bearable enough to use Firefox as my daily driver. They may not be as privacy-focused as some of the other derivatives of Firefox but they’re sure not Chromium.
My biggest concern with a toy like this is that my future kid might ask for a water park in our backyard and then Santa would respond with an enthusiastic “That’s a great idea Kyle! I’ll consult with the elves to see how I can make it happen!”
I had an app idea that was effectively this Walmart product but Santa was always in a blizzard and/or hard of hearing, and continuously misconstrued requests. "What?! You want to order bark?" Idea being that kids love nonsense, but also the scenario pointed towards kids not expecting anything to be a real request.
Seriously, I don't understand how this is meant to work, even on the 'happy path'. I've never done any Santa stuff in my family except stockings with small presents + fruit, so the stakes for make-believe are low.
Do the children ask for stuff and then the parent is on the hook to buy it? What if it's too expensive or unavailable? Just a massive disappointment on the day? Does the child expect that it's some kind of binding contract?
Children's imaginations are wonderous, flexible things. As an adult I have sometimes found it a weird experience to play along with my child because my brain keeps trying to delineate between reality and imagination. So who knows how the it's perceived when you're writing a letter.
But if this really does sound realistic, isn't it in danger of leaving the imagination space and setting an expectation?
I use [insert LLM provider here] all the time to ask generic, health-related questions but I’m careful about what I disclose and how I disclose it to the models. I would never connect data from my primary care’s EHR system directly to one of these providers.
That said, it’ll be interesting to see how the general population responds to this and whether they embrace it or have some skepticism.
I’m not confident we’ll have powerful/efficient enough on-device models to build this before people start adopting the SaaS-based AI health solutions.
ChatGPT’s target market is very clearly the average consumer who may not necessarily care what they do with their data.