Hacker Newsnew | past | comments | ask | show | jobs | submit | fwipsy's commentslogin

The suggestion is not to depend on AI for privacy, but to use it to build products like signal which guarantee privacy. Using AI to build a product doesn't mean sharing the data from that product. In addition, right now AI has little vendor lock in and there are multiple competitive alternatives, so becoming dependent on a single company is not so likely.

>The suggestion is not to depend on AI for privacy, but to use it to build products like signal which guarantee privacy.

Uh huh. No. You use their system to do it, they have your prompt, and the output on hand. Even more so, they have the capability to tamper with it. They are essentially in a position to own the entire instance of the work product. It doesn't matter if they don't yet. It matters that they can. Furthermore you lose out on the learning. You lose out on any innovation. You lose out in the eyes of the law on the privacy of the communique you use to drive the black box.

>In addition, right now AI has little vendor lock in and there are multiple competitive alternatives, so becoming dependent on a single company is not so likely.

Yup. Right. Like we don't know how that ends. <gestures to />50 years of market consolidation in the distance, letting the illusion of choice speak for itself>


It's not really meant to be advice. It's the author's own experience, ironically written as if it were advice.

For example:

"You did bleach ten gallons of well water for long-term storage already earlier in the year, right? Good."

This is sarcasm, because the author did not do that.


Ah I see. That didn’t translate well for me. Maybe because the title primed me into thinking that this was meant to be helpful.

That's fine, humor is subjective. I had a similar experience watching the "manchild" music video recently. I knew it wasn't serious, but I was still annoyed until I thought it through and understood the satire.

It's obviously a self-depreciation/joke style.

There is some truth in it [that doesn't translate well over to some other part of the works]. It requires rather poor infrastructure to be present.

There has been snow for over 2 months here, with relatively low lows (-29C) but no issue like lack of electricity or water.


Perhaps it's an AI generated article. A real human would have realized quite quickly that you can put snow into the tank of a toilet when the power is out.

This is about _tankless_ toilets. They only work with electricity-powered flush pumps. That's why the author wrote about having to physically dump water into the toilet to flush it.

For our new home we're making we have two toilets (always practical). One of them is tankless, but we made sure the second one is a traditional cistern toilet with no electrical requirements. Just in case.


Most well pumps are electric powered. The holding tank will give you a very small amount of water that’s in it if it’s up high but after that without electricity it won’t refill.

In the USA most residential toilets are tank type and don’t directly use electricity.


At risk of stating the obvious - computers produce heat. "Keeping them cool" really means dissipating that heat. Insulating them will cause them to get hotter.

Chatbots aren't better than doctors can be. But in the US, doctors are a highly credentialed position, therefore expensive, therefore their time is split into minimal parcels. (Still longer than 5 minutes at least!!) In my experience, chatbots are often better than doctors are in the real world, at least for savvy users.


Technically they don't have incentives either. It's just difficult to talk about something that walks, swims, flies, and quacks without referring to duck terminology.


Psychology can change neurochemistry but only in certain limited ways. Many people are on antidepressants long term because that's the only thing that works for them. Taking antidepressants is already stigmatized enough. People should just do what makes them feel best over the long run. Your rule of thumb does not trump hard-won personal experiences.

We don't really know how SSRIs work, but there's some evidence that it's through desensitizing serotonin receptors, not directly addressing the lack of serotonin. If so, "use it or lose it" doesn't apply; long-term adaptation is the point, and SOMETIMES does persist after quitting.


You mean the real-life save icons?


Iran is the 17th most populous nation in the world, with 93 million people. These protests seem to be occurring across the entire nation. Another comment mentioned over 4,000 separate clashes. Other sources have already corroborated a lower bound in the mid-thousands. I think the burden is on you to refute these numbers by showing that the sources are deliberately misleading or finding a flaw in the methodology. Simply saying that you find them "not credible" and that some people might have a political motive behind sharing them is not an argument.

Note, I'm not saying that they have been confirmed, but I do not think that you have given sufficient cause for rejecting them out of hand.


https://www.en-hrana.org/day-twenty-eight-of-the-protests-ar...

This is the organisation most commonly cited in news reports, they estimate ~5200 protestors confirmed killed (+ a few hundred more for security personnel killed)

They are a group of anti-regime Iranian dissidents based in the US. I don't know why they would seek to provide a deliberately low estimate.


Confirmed != estimated. This source does not make any estimates. They are investigating every death individually. Given the lack of transparency, the true number of deaths is likely higher than the number which can be confirmed at this time.

As of writing this comment, the subtitle says "The number of deaths currently under investigation stands at 17,031." They do not claim that this is the total number of deaths either.

30,000 is not confirmed but cannot be ruled out.


How do you know the casuality isn't reversed? Maybe the social media prioritizing outrage became the biggest. Don't hate the player, change the game.


Aha! The 8x8 bitmap approach is the one I used back in college. I was using a fixed font, so I just converted each character to a 64-bit integer and then used popcnt to compare with an 8x8 tile from the image. I wonder whether this approach results in meaningfully different image results from the original post? e.g. focusing on directionality rather than bitmap match might result in more legible large shapes, but fine noise may not be reproduced as faithfully.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: