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

That's how the new 1927 starts.

It will be in the history books.


First, let's acknowledge that when the actions of some people lead to endangering a human life, it's natural to be angry about it.

Personally, when anger is justified, I feel more comfortable with people who clearly and openly express it.

I find them more reliable, more honest, and usually better people to be around.

By my standards, I'd even consider the style very mild, considering a life is on the line.

I would also consider it normal to be on the receiving end of such language if my own work resulted in such a situation. That would cause me to pause and reflect.

The intensity of feedback is information. If everything is bland, it's harder to know how important something is or not. Logic has a limit because you don't have all the parameters from the other side.

It works the other way around. The overuse of superlatives and day-to-day outrage is equally unproductive.

This is not the case here, IMO.


> First, let's acknowledge that when the actions of some people lead to endangering a human life, it's natural to be angry about it.

The thing that's really bugging me is that this is not what happened at all!

"I'm not at home, please send the new pump to my hotel."

That's it! That solves the entire problem in 5 seconds plus the time to tell the customer service rep the hotel's address and verify it's correct. Yeah, there were some other structural problems that led to other dumb things, but none of that would have even come up if she'd done the blindingly obvious thing and had the life-saving, 24/7-required device shipped to her actual location. If anyone endangered her life, it was her, through her own actions.

She really crossed the line for me by wishing harm on the people who designed and built her insulin pump. What the hell? They've done nothing wrong. Yes, her pump was beginning to fail, but 1) it was still limping along with a workaround and she wasn't in immediate danger (and if it did fully fail, she said she was within an hour's drive of an ER at all times, which she felt was fine), and 2) she had never had a pump problem in twenty-five years, which is an amazing testament to the reliability of these devices. The people who work on these things deserve a freakin' medal, not bullshit like "the people who design, sell, and service these machines are both keeping me alive and also my mortal enemies".


It's more than alright to be outraged, that's very different than it being alright to wish harm. That shouldn't negate anything else said in the conversation, but it's also just as much of the conversation to call it out.


Until they learn to do that. So cat and mouse. So nothing new.


They already have. Claude and OpenAI are not trying to write captcha-defying AI agents. These tests wouldn't hold up as well against proper bot operators who mimic user behaviour. However, the signals are still valid as part of a larger toolset.


think the point is that they can't just "learn to do that", because to do so would mean solving human mind (that famously hasn't been going well)


Well no, the idea is a tradeoff between interfaces and telemetry.

OK, the agents don't click in the same way as humans. You learn that, what about mouse hovering telemetry, time spent, etc. And one of the most extreme is to force biometrics - a lot of telemetry, breaks the interface a lot - but hey, you have assurance.

And none of these tradeoffs require understanding the deep processes of the human mind. Just, map is not the territory, how you do game the map harder and harder and how do the mapmakers respond to that?


did you look at the paper? they specifically look at mini tasks with cognitive processes (Eg what dictates the strategy of how people solve tasks)


LLMs can solve original math problems at the IMO level and beyond, and you might be talking to one now. I don't think they are going to have problems with any CAPTCHA short of separate device attestation.

Whatever mechanism the paper proposes, rest assured it can be trained on.


until Google trains an AI model off that data, too


I mean, their CAPTCHAs presumably have tons of data collected over the years, and they can't detect a pretty clear AI agent here: https://www.youtube.com/watch?v=UeTpCdUc4Ls


If you know happy rich people that don't have a job, you got your counter example, and one is enough.

I do.

People usually need to have a purpose, but it doesn't need to be a job.


I agree "people need a purpose to be happy" is much more digestible than "people need a job to be happy". However, it has to be qualified with "some people need a purpose to be happy". Defining, or worse dictating, happiness for everyone is a fool's errand and, ironically, usually leads to large scale mass murder or starvation.


You may not need to have a job to be happy, it varies person to person. However, the idea that the billionaires will save us and our leverage is not needed is ridiculous. It is much more likely we would see poverty like is seen in much of the rest of the world.


I still haven't seen anybody demonstrating we have to do such thing.

We do have to spent a bit of each for any new tech advancement, but the alarmist, disproportionate claim you make is really not helping.

ArcelorMittal Dunkirk rolling & steel complex alone is ~450–550 hectares (more than colussus) and consume 2.23 TWh/year (colussus is ~2.6 TWh/year at 300 MW continuous load) and of course, water consumption for metal working is gigantic.

That's just ONE single facility in France.

I don't think anybody who understands the basics of civilization would want to go back before the Industrial Revolution.

Tech has a cost, and you usually pay a lot more at the begining of creating it.

Does it cause problems? Sure. Should we take it very seriously? Definitely.

But just repeating internet outrage is not a way to make good decisions.


The question is not about the cost the question is who pays that cost.

Tech companies clearly want to coast on the infrastructure built out by the government. Microsoft doesn't really want to pay 10 billion euro to upgrade the Dutch power supply.

What happens is that they will go to countries that let them pollute- a tale as old as the industrial revolution.


That's a different issue than OP's superlative claims and is something we as a society need to be working on.

But it's not AI related, it's true every time the eich and powerful privatize benefits while making the costs and risks to society.

It's a more global and important problem.


7 litters...


...Not that much... 3ish liters spread out over the day is good. Too much water will strip your body of electrolytes and salt.

I recently upped my water intake and can report that my skin and mood feel better. I caffeinate with a strong coffee and 2 cups of tea in the morning, and around 12 I switch to pint glasses of water every 2 hours or so until 6-7 o'clock.


Yea 2.5-3l seems to be the proper amount. Only drawback is having to pee every 30 minutes.


That's an advantage - it means you're not sitting e.g. at a computer monitor and keyboard for too long. Getting up twice an hour (to pee or not) greatly reduces the potential for ergonomic harm that desk jobs - especially computer use -are known for.


Fair :D I walk everywhere anyway so I think I'm good in that regard


Bug == Feature


If you're peeing that much then evidently your body is not using all that water you're consuming. Either it's more than you need, or you're lacking in electrolytes and should increase electrolyte intake alongside the water.


Your body doesn’t use up water turning it in to nothing. You are meant to pee out what isn’t lost to sweat.


Parent is correct. NHS sources say upto 7 times during the day is normal and "Every 3 to 4 hours is considered healthy. Going every 1 to 2 hours usually means you either have an overactive bladder or are drinking too much at once."

Saying every 30 minutes is normal is just the same as advocating drinking too much water


By that logic, if I drank the right amount of water with 100% accuracy for multiple days, I'd only have to pee every few days.


I’m not arguing that 100% of water consumed can be retained, just that peeing every 30 minutes means you have an unnecessarily large surplus. This is simply a truth.


It's actually fantastic for job interviews. Use that trick with my clients when asked to review candidates.


Note that 3.15 is not released yet. It will come out in 4 months


I think the opposite. Automation and scale made it possible to cheaply influence way more people. Information bubbles mean people are more convinced than ever.

Before you had one big influencer, usually the state, manipulating the news of your country. The alternative players were more trustworthy, because they had to if they wanted to keep credibility.

Now, any big company can buy a narrative when they need one. No reputation matters. No memory. No consequences for the long run.

It's literally pay to play.


> Information bubbles mean people are more convinced than ever.

And therefore critical thinking is more important than ever. That isn't intrinsically a problem, but it is a problem for people that lack it.


> Before you had one big influencer, usually the state, manipulating the news of your country. The alternative players were more trustworthy, because they had to if they wanted to keep credibility.

There isn't much in the way of evidence that is true. What we discovered when the internet came around is that the major players in the public sphere are generally low-credibility and just keep repeating that they are credible over and over again. As more diverse alternatives appear with the internet that narrative has been struggling. The new replacements have proven to be low-quality and a lot better than what they are overtaking.

The US discourse is a fascinating example of this where the traditional narrative setters may have gone legitimately crazy trying to rationalise how they're part of the reasonable and credible crowd and they just got rolled by Trump screaming that they were liars and that Mexico was going to pay for a border wall. That's what US voters thought looked like a better bet than these traditionally credible alternatives.


Maybe as often, the laws is just a pretext for this little nuget. They innovate for once, not using terrorists or pedophiles, nobody checked.


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

Search: