Ant wrote on Sep 29, 2024, 19:35:
Blue, are you happy with your job? We're happy with your work.

+1

Blue, are you happy with your job? We're happy with your work.

WaltSee wrote on Sep 26, 2024, 17:24:
Yes, indeed... In life, there's no substitute for a happy wife...;)

Good luck to Mrs. Blue in her new role!

And I hope everyone is safe - that hurricane is clearly a nasty piece of work.

Thanks BoP. All okay here. Much ado about nothing for us, We were outside the hurricane's major influence, and also dodged most of the worst bands for the most part.

TFP, and any other members of Blue's that are in the path of the hurricane, I hope you and your family remain safe and whole.

Zerg wrote on Sep 26, 2024, 16:47:
Last year, after five years, I switched jobs to join my wife at another company. We work in the same team now and it's always funny to hear "I had no idea that was allowed!" from anyone who doesn't know us.

It's made our life very interesting as we grow into people that can work together in both professional and personal environments. Hope your wife will enjoy her new job.

I would sooner live under a bridge and eat nothing but sawdust for a year before I went to work with anyone I had close familial ties to. I am cringing even at the thought of it.

Yes, indeed... In life, there's no substitute for a happy wife...;)

Last year, after five years, I switched jobs to join my wife at another company. We work in the same team now and it's always funny to hear "I had no idea that was allowed!" from anyone who doesn't know us.

It's made our life very interesting as we grow into people that can work together in both professional and personal environments. Hope your wife will enjoy her new job.

If there is no PII attached to the data, and OpenAI is not storing the data you send them, then this might be feasible. The problem I see is that OpenAI (or any LLM) will need training data to be of any use. If this is novel usage, then it probably doesn't have any, so it's not going to do what you want it to.

Riahderymnmaddog wrote on Sep 26, 2024, 14:22:

The Flying Penguin wrote on Sep 26, 2024, 12:59:

Jaysen wrote on Sep 26, 2024, 12:29:
Happiness at work is so important. Happy for you both.

I can't see ANY way this could be made HIPAA complaint. You'd need, at the very least, for OpenAI to somehow certify data privacy in whatever model they use, and take responsibility for any data breach.

It's possible. you could look into Regard

I'll pass that on, for what it's worth. All the charting companies ARE working on incorporating "AI" into their EHR products, but like everyone else jumping on the hype bandwagon, I have no illusion they will put safety and privacy on the front burner.

I'm getting the feeling my friend wants to do this as a business model, not just for himself, but it sounds like he's under the illusion he can use a public LLM. Yeah he can try to create a personal one, assuming he can figure out where to get the training data, but it feels like a legal nightmare.

He's a great guy, and a top notch pulmonary doctor, but he knows nothing about the underlying tech.

This reminds me of a client who tried to convince me to write a "stock investment program" 20 years ago. It would be based on his investing knowledge. I would have the "easy part" of just writing a program to do that.

The Flying Penguin wrote on Sep 26, 2024, 12:59:

Jaysen wrote on Sep 26, 2024, 12:29:
Happiness at work is so important. Happy for you both.

I can't see ANY way this could be made HIPAA complaint. You'd need, at the very least, for OpenAI to somehow certify data privacy in whatever model they use, and take responsibility for any data breach.

It's possible. you could look into Regard

.

This comment was edited on Sep 26, 2024, 15:47.

Well I hope it works out for her. It's not like finding new gigs - good ones at any rate - gets any easier as you get older unless you have some seriously in demand skillz. Ageism is a brutal reality out there.

Check out that Space Elevator site. Really cool info and well done. It always amazes me when people come up with stuff like that.

While I love the northern lights, I also love reggae nights. Gotta' love Jimmy Cliff!

Those sea robins are creepy, dude.

Fucking people and their magic pills. Semi-glutides are bullshit. They only work in conjunction with diet and exercise. And if you're eating a healthy diet and exercising you don't fucking need them at all. I'd love to see the average supermarket trip for people who take this crap. Dollars to donuts probably 90% of their shopping cart will be filled with ultra-processed junk food. I see morbidly obese people all the time with carts full of frozen dinners and other crap. Not a speck of fresh food - fruit and veg - in them. It's pretty goddamned sad. And now we're starting to see all the negative side-effects of those drugs come to light as it's popularity has soared and it's finally got a real global test base. Hard fucking pass, thanks. Beyond that, the American health care system is such a shitshow. "A Wegovy prescription costs about $1,349 per month in the U.S., while it costs just $140 in Germany and $92 in the U.K. Novo Nordisk’s Ozempic carries similar markups." 1300% more expensive in the US for the same goddamned drug. Fucking unreal!

500k opening bid for a fucking baseball? Rich people seriously need to start paying tax.

Savage Chickens = spot on.

Jaysen wrote on Sep 26, 2024, 12:29:
Happiness at work is so important. Happy for you both.

I kid you not: Former client of mine, and a good friend, who is a doctor, emails me today, asking if I can work with him on adapting ChatGPT for EHR medical charting. He has been talking with reps at OpenAI who are enthusiastic (of course they are). He then brings up the sticking point as to why he wants my help: HIPAA compliance.

I can't see ANY way this could be made HIPAA complaint. You'd need, at the very least, for OpenAI to somehow certify data privacy in whatever model they use, and take responsibility for any data breach.

There's a use case here, I agree. Doctors are burning out due to the workload of updating EHR records, but not only are there privacy issues, I'm not sure this is even safe, not without a hell of a lot of testing. I realize (or at least assume) it would be a closed model, with medical specific training, but I'm not sure this shit is ready for prime time.

In his email he is implying he knows of doctors already using regular ChatGPT to assist with EHR records. Frightening thought.

Sigh. I have a long letter to write later.

Happiness at work is so important. Happy for you both.