Piero Bosio Web Site

@jan Mja maybe we (the collective humanity “we”) are simply gpung through a learning curve. Especially in areas where people are inexperience, using a llm is both helpful and problematic. Helpful to get them started but you tend to rely on it too much. I have no idea (for now) if it inhibits real learning or helps it, but using it to contribute too quickly can indeed be harmful.

@phillip @me @serebit @protonprivacy not saying there isn’t plenty management bullshit that is just overhead or attempts to push ppl to lower quality and squeeze more out of them, no doubt. But I have no indications Proton has this issue. Maybe I’m wrong of course, I’m just a customer and have a few minor relations there.

@Hierarchy @ulveon @serebit @protonprivacy well in all fairness to ulveon, if you give them the benefit of the doubt - there are issues with outsourcing & offshoring that go beyond racism… companies do try to cut cost in unethical and exploitative ways. Finding talent wherever it is is great. Firing your European developers because a coding sweatshop in a poorer country can do it cheaper - not cool. But this doesn’t seem what proton does.

@anderslund @Linux if it doesn’t boot it’s not a testing build 🤪🤪

@Kmachel @serebit @protonprivacy well you do r have to use it, it is optional. I appreciate they try new things - their solution is still more privacy sensible than any other hosted ai solution as it has no user data at rest.

Honestly it’s smart to do this - got them attention and means they can deliver something some customers want. Would be quite a sign of incompetence if they hadn’t done it I would say…

@phillip @me @serebit @protonprivacy it’s sounding nonsense. Proton isn’t a for profit, it’s owned by a foundation. That, to have maximum impact to reach their goal of providing privacy, they measure and guide employees with kpis seems wise, what else, letting the vibes dictate if ppl do good work?

@serebit @protonprivacy
> If this prompt is the real deal, then your files and everything else are just dumped into the model’s context. That means your request and your previously end-to-end-encrypted files will now exist unencrypted at rest on Proton’s servers.

What is the logic going from prompt to “they store data”? They have been 100% transparent - the llm processes data when you do a request, and doesn’t store data….

This article is honestly bull.

@serebit @ulveon @protonprivacy I dont think it is true. They have a bunch of offices including Spain etc and sure salaries will differ but why is that an issue, being willing to hire outside of Switzerland?!?

@troed @serebit @protonprivacy yes, and Germany, which I thought they were moving to, has had better privacy regulations than Switzerland for a loooong time…