Comment: Open-source developers are working themselves sick on AI bugs

[-]

Valkhir@reddit

What a shitty inaccurate headline.

The article makes it clear that many of the bug reports are actually good quality, and the problem is that while AI enables more bugs to be found more quickly, funding for more developers to handle the is not adequate.

But I guess that doesn't generate clicks.

[-]

FryBoyter@reddit (OP)

The article makes it clear that many of the bug reports are actually of high quality, and the problem is that while AI enables more bugs to be found more quickly, there isn’t enough funding to hire more developers to handle them.

And that is exactly why developers like Stenberg now seem to be experiencing health problems. So I don’t think the headline is that inaccurate.

But yes, a better headline could have been chosen. Just as you don’t have to give every major security vulnerability like Heartbleed or Dirty Cow a special name. But people often don’t do that because they want to motivate users to click. When I publish a blog post, I sometimes don’t use completely neutral titles either. And I don’t make a penny from it.

But instead of complaining about the title, we should rather talk about the actual problem and try to solve it. Because, unfortunately, the comment is right in terms of content. Besides, it’s a comment and not an article.

[-]

Valkhir@reddit

The headline says

Open-source developers are working themselves sick on AI bugs

"AI bugs". As if AI was responsible for the bugs.

Not "AI-submitted bug reports" or "bug reports submitted with the help of AI" or anything remotely accurate. I'm sorry, but it's just plain misleading.

[-]

daemonpenguin@reddit

Maybe English isn't your first language? The title is both clear and accurate.

[-]

ImNotABotScoutsHonor@reddit

No, it isn't. What they are saying is correct.

"AI bugs" makes it read as if it were bugs introduced by AI.

[-]

daemonpenguin@reddit

That's what the headline indicates. Nothing about the headline is inaccurate.

[-]

Infinity-of-Thoughts@reddit

The headline doesn't make any mention of the bug reports being of bad quality.

The headline literally just implies that "there are so many bug reports now, and not enough open source developers to handle it, leading to them being overworked".

The article literally also says that it has taken a toll on Steinberg's health:

This leaves little time for the project's other further development and takes a toll on Stenberg's health. His wife, as the Swede blogs, has expressed her concerns about his long working hours and the imbalance between work and leisure for the first time.

[-]

daemonpenguin@reddit

I'm in the same boat. Almost all the reports I've received recently are from AI bots, most of them inaccurate. I'm considering shutting down all of my public code repositories (there are about a dozen) just so I don't receive slop.

[-]

SanityInAnarchy@reddit

Data processing by advertising providers including personalised advertising with profiling - Consent required for free use

That seems incompatible with the GDPR, and it's unlike pretty much any of these other consent dialogs I've seen. Here's the archived version.

[-]

Kevin_Kofler@reddit

Unfortunately, courts ruled that this extortionary practice is legal. The GDPR only requires there to be a way to refuse cookies, it does not require that way to be free. (According to the court rulings, this practice also does not legally constitute extortion or anything else illegal.) Making it pretty useless. Extortionary cookie banniers have now become the industry practice in newspaper and magazine websites and online newspapers and magazines.

[-]

JimmyRecard@reddit

It is almost certainly illegal. GDPR requrires that the method to decline cookies must be as easy as the method to accept them. In no universe is having to pull out a credit card as easy as accepting cookies. However, EU courts have been reluctant to enfoce their own laws because for the most part, the sites using this are newspapers who are already struggling to keep their head above the water.

When Facebook tried it, they got smacked.

https://noyb.eu/en/noybs-pay-or-okay-report-how-companies-make-you-pay-privacy
https://en.wikipedia.org/wiki/Consent_or_pay

[-]

cafk@reddit

In Germany and Austria this has been ruled as a legal & valid approach, based on local law.
As it's easy to not visit a page - there is no mandate that the content has to be accessible without consenting or paying.
And not visiting a page is an easy way to ensure that you don't have to accept the cookies unfortunately.
Similarly to how in the 90s "I'm 13 or younger button" consent banner redirected you to online children's media and didn't grant you access to the site.

[-]

JimmyRecard@reddit

GDPR is a directive, meaning that ot applies directly and uniformly across all of EU, and it overrules local laws where in conflict.

[-]

ficiek@reddit

This is completely not how this works.

[-]

JimmyRecard@reddit

No, that's directive. Directive = binding goal for the national legislation to achieve. Regulation = legally binding in every member state when they go in force, and overrule national law when in conflict. Incredible how confidently incorrect people love to be on this website.

Regulations are in some sense equivalent to the legislative acts of the member states, in the sense that what they say is law and they do not need to be mediated into national law by means of implementing measures. As such, regulations constitute one of the most powerful forms of European Union law and a great deal of care is required in their drafting and formulation.

When a regulation comes into force, it overrides all national laws dealing with the same subject matter and subsequent national legislation must be consistent with and made in the light of the regulation. While member states are prohibited from obscuring the direct effect of regulations, it is common practice to pass legislation dealing with consequential matters arising from the coming into force of a regulation.

https://en.wikipedia.org/wiki/Regulation_(European_Union)

[-]

cafk@reddit

It's a directive to create a law, which German DSGVO is.
The implementation of the directive has room for interpretation and first needs to be escalated through local law & legislation by someone who is willing to spend money on lawyers for principles, in order to either change local laws or to be able to escalate the issue to European level.
EU isn't as powerful as you assume it to be, Germany still has their right wing appeasing border controls under emergency law, which contradict EU freedom to roam directive, some regional municipalities are suing against it, but we've seen decisions go in bith directions.

and it overrules local laws where in conflict.

It cannot overrule existing laws that a country already has, even if conflicting with a directive. EU has no way to change local law, but gives guidance with a relatively lax timeline for implementation and enforcement is down to local levels.

[-]

JimmyRecard@reddit

Sorry, I wrote directive when, I meant regulation.

Regulations are directly binding in every member state.
https://european-union.europa.eu/institutions-law-budget/law/types-legislation_en

https://en.wikipedia.org/wiki/Regulation_(European_Union)

When a regulation comes into force, it overrides all national laws dealing with the same subject matter and subsequent national legislation must be consistent with and made in the light of the regulation. While member states are prohibited from obscuring the direct effect of regulations, it is common practice to pass legislation dealing with consequential matters arising from the coming into force of a regulation.

[-]

cafk@reddit

And from article 288:

A directive shall be binding, as to the result to be achieved, upon each Member State to which it is addressed, but shall leave to the national authorities the choice of form and methods.

So if a regulation is loosely worded and gets translated to national legislation and law, there can be differences.

Which enables the Leave/Pay/Accept approach in German legal definition of DSGVO which is the law implementing GDPR.

[-]

JimmyRecard@reddit

Regulation and directive are two different types of EU legislation. Those transposition rules only apply to directives, not to regulations.

A directive is EU telling countries what's their goal, and countries writing their own legislation to achieve it.
Regulations are directly binding without any further transposition (as long as they don't regulate outside of the areas where EU has supremacy, and they don't infringe on the country's constitution)..

Please educate yourself on EU legislation.

[-]

cafk@reddit

DSGVO is the implementation of GDPR regulation, which allows the leave/pay/accept approach handling.

Again, the article 288 describes how EU regulations can be implemented by countries.
If a regulation has holes, those may be translated to the law which may seem against the intent.

DSGVO is the implementation under Article 288 of the GDPR in Germany and thus the German interpretation of the regulation, with additional clarifications included in Bundesdatenschutzgesetz the that was the German predecessor.
It contains some aspects which are noticeably more strict compared to GDPR, others that clarify vague definitions from GDPR to German law.

It's not about understanding EU law, but how the countries implement the law, which in some cases allows this interpretation.

[-]

JimmyRecard@reddit

Regulations don't need implementation. They're automatically legally binding. Try again.

[-]

Jean_Luc_Lesmouches@reddit

GDPR requrires that the method to decline cookies must be as easy as the method to accept them.

But it is easy to avoid cookies: leave the website. What the gdpr forbids is "you're on our site, so we assume you've already accepted cookies for this page" which was the norm before

[-]

JimmyRecard@reddit

That's explicitly prohibited. GDPR Article 7(4) says:

When assessing whether consent is freely given, utmost account shall be taken of whether, inter alia, the performance of a contract, including the provision of a service, is conditional on consent to the processing of personal data that is not necessary for the performance of that contract.

So, such consent based on pay or okay screens is not freely given within the meaning of GDPR, and is not valid.

This means that you can refuse service if the lack of data prevents you from proving service. The classic example here is being able to refuse service if you're a delivery company and data subject refuses to give you their address.

When it comes to running ads, tracking individual users is nor necessary as you can run static ads or ads based on the context of the article (same way that physical newspapers have been doing for more than 100 years).

[-]

Jean_Luc_Lesmouches@reddit

the performance of a contract, including the provision of a service

News websites do not provide you with a contract or a service if you're not a subscribed customer.

[-]

JimmyRecard@reddit

If that's the case, terms of service aren't valid. I actually agree with you that such implicit browsewrap or clickwrap contracts shouldn't be legal, but that's not the position of the websites since they asert that their terms of service are valid contracts.

[-]

Jean_Luc_Lesmouches@reddit

Those are different things

[-]

dutch_connection_uk@reddit

I mean given the constraints of the business model I imagine the result of that would be that you have to make an account and submit credit card info to have any access, even if you accept cookies?

[-]

Declination@reddit

Facebook offers a "free" service. Ostensibly, newspapers have a business around selling their content and I think that is the difference. There cannot be a free standing right to get other peoples products for free. But if you are giving away something you don't then get to claim you have an interest.

[-]

JimmyRecard@reddit

That would alleviate the concerns of rejections not being as easy as acceptance.

I think the issue is still the monetary payment. Labouring at your place of employment for x number of hours is not as easy as making a free account and entering your credit card (without being charged).

[-]

trueppp@reddit

Nobody is forcing you to use the website.

[-]

CardOk755@reddit

They don't want people to read their shit? I won't read it. Everyone is happy now, right?

[-]

Annual-Advisor-7916@reddit

I don't remember any ruling that it's legal. I remember a prolonged legal battle here in Austria though (Standard newspaper).

[-]

Kevin_Kofler@reddit

The archive.org link still requires you to use the developer tools to remove the "sp-message-open" scroll block.

Here is an archive link that also rips that out: https://archive.ph/8mx3E

[-]

SubmarineWipers@reddit

Devs will need to realize that writing applications (esp. network facing apps) in languages as unsafe as C is only generating tons of unnecessary work, that safe languages prevent. C is nearly 60yo, its time to let it die everywhere but in tiny MCUs.

"We adopted Rust for its security and are seeing a 1000x reduction in memory safety vulnerability density compared to Android’s C and C++ code"

https://byteiota.com/rust-memory-safety-2026-82-better-performance-proven/

[-]

Lonely_Fig5352@reddit

nice try microslop diddy

[-]

Kevin_Kofler@reddit

LLM AI is a scourge that destroys our planet with its unbounded energy hunger, hikes up prices for energy, RAM, and SSDs to astronomical levels, and makes human software developers stupid (as shown in several studies, even one by Anthropic themselves) and sick (as in this case).

[-]

MatchingTurret@reddit

LLM AI is a scourge that destroys our planet with its unbounded energy hunger, hikes up prices for energy, RAM, and SSDs to astronomical levels

Reminds me of this:

In 50 years, every street in London will be buried under nine feet of manure

[-]

thoughtcriminaaaal@reddit

You are projecting current problems in a linear fashion into the future. Things usually don't work that way.

Literally no one in the AI space has any solution to making AI better than to feed it more synthetic data, train it for longer, and use more tokens. Their gamble is that they get AGI or the whole market collapses, because none of them run a profit.

[-]

ZorbaTHut@reddit

This isn't even remotely true, there are constant innovations and improvements.

[-]

MatchingTurret@reddit

So? I don't care whether OpenAI collapses or not. That was not what I commented about. If the AI bubble bursts, the problem is solved.

[-]

NonStandardUser@reddit

I agree with both sides; thus, I think you should think of kevin's comment as a sign of the AI bubble popping and the "something" stopping. In which case, you don't need to react!

[-]

MatchingTurret@reddit

In which case, you don't need to react!

People are reacting. The problems are obvious and a solution will be found.

I don't claim to know what the solution will be, but I'm 100% sure that "unbounded energy hunger, hikes up prices for energy, RAM, and SSDs to astronomical levels" will not go on, either through a tech solution (preferably) or through a burst bubble with all its consequences.

[-]

NonStandardUser@reddit

Well I'm saying you don't need to go around saying stuff to people. I've clarified this point in the comment 😁

[-]

ThePoisonDoughnut@reddit

Your entire logic rests on the assumption that humanity isn't about to reckon with: - the worst energy crisis in history - the increasing inhabitability and land infirtility of large and highly populated parts of the planet - geopolitical tensions resolving in unfavorable ways to the precarious supply chain that allows these technologies to be produced at all (Taiwan is currently free real estate for China right now if they decide to take it, so there goes TSMC) - many other completely unsustainable aspects of the current status quo that I should probably not bring up here because this isn't a politics sub

So many of these dialectics could individually pop this AI bubble when they inevitably resolve unfavorably towards any hope of this going the way you want it to go; are you so busy tokenmaxxing that you're oblivious to all of them?

[-]

MatchingTurret@reddit

are you so busy tokenmaxxing that you're oblivious to all of them?

LOL. That's not me, I assure you. I have used Gemini on my phone, though.

[-]

Kevin_Kofler@reddit

Yet all the predictions have either come true or even been surpassed, because politicians have done absolutely nothing to counter the crisis for all those decades.

[-]

MatchingTurret@reddit

I honestly can't understand this kind of dooming. It's obviously not true. Billions have been lifted out of poverty, smog in large parts of the world is a thing of the past...

[-]

kat-tricks@reddit

our planet is accelerating towards 2 degrees of global temperature increase!! ecosystems are collapsing, people are being displaced now. Are we watching the same news? Or are you one of those rich people who thinks the world is better because everyone in their neighbourhood has a new lexus?

[-]

MatchingTurret@reddit

I'm looking at things like this: GDP per capita (current US$) - China which single-handedly lifted a billion people out of extreme poverty. People forget that in living memory 55 million people starved to death there.

Or this: 🌆 Los Angeles in the 60s & 70s: The Smog Crisis That Changed California Forever | History of LA

[-]