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

Personally, I feel it would be a much smaller problem if Tumblr had an internal AI thing going on. What users REALLY don't like is that they have confided a post with one website, and that website just shared that post with a third-party because if opens up infinite possibilities.

If Tumblr can take your post and give it to OpenAI, they can take your post and give it to anyone, and that's the problem. Because for users, what they post is "between them and Tumblr" and not anybody else.

I'll even say more. Artists don't care if you scrape their art to make AI generated art with it. Because when this happens, it's "between artists and scrapers" and not anybody else, so it's fine. What they do care about is when people post that AI generated art on the internet, or publish it professionally, or do basically anything with it.

In other words, there's a sense of privacy when there is only two parties involved, no matter what is going on, but the instant a third party gets into the system, the first party will freak out, because then you lose that sense of privacy.


You're saying consent isn't transitive.

That's why they always ask for (or more frequently just take) "a paid up, non-exclusive, irrevocable, worldwide license"


I think "consent" is not the right word for this, because it's never simply "consent" it's always boundaries and expectations. Consent implies there's always one yes/no answer for a process. In practice the process is always so complicated that it would require countless yes/no questions, to the point nobody sane would want to deal with it.

Just look at cookie banners and imagine if we had a consent popup for every single thing that needs to be download to show a page: do you consent to download the HTML? Do you consent to download the CSS? Do you consent to download the Javascript? Do you consent to download the images? The user just wants the see the page. You have to make assumptions about how far that consent goes, so it's absolutely transitive, the problem is at what point it crosses the line.


So, please forgive the necessary parallel here.

Let's think about dating and imagine if we both had to consent to every single thing that needs to happen to get it on. Do you consent to dinner with me? Do you consent to coffee at yours? Do you consent to us getting our clothes off? Do you consent to ... you get the picture.

Thing is, we actually do this, but silently.

The "assumptions" you rightly speak of are the mutual trust mechanisms that allow two adults to make clear shortcuts without explicit words. We move at a certain speed. We check for feedback. We use clever signals. We base our trust model on the NSA definition that "Trust is the ability to do harm" in context where there is some symmetry. Sex is actually risky for both parties.

And no means no.

But in the digital world we're talking about a massive power asymmetry. When you're with a big corporation it moves from dinner to data-rape in a few milliseconds.

No means "yeah but harder".

Consent doesn't disappear because the process got complex. Human relations are more complex than any web page.

It's about whether we obey the rules and show mutual respect. Big-tech corporations absolutely do neither. and that's the problem here, not complexity.

What might level the game, is if when I visit your site with my web browser, you risk that I could do you some serious harm.


Unfortunately, it's not a human on the other side, it's a system, whether it's a system built based on business rules or a system built based on computer algorithms, everyone gets the same inflexible system. It's not logistically feasible to create a custom experience for each user based on how far they consent to every little thing, as the use cases are infinite: you may consent to download CSS on this site, but not on that site, or on that site, but not on pages that match this regex, etc. If you tried to do that, your product would immediately flop because most users would find it too annoying and complicated to use compared to any system that makes reasonable assumptions about consent.

I agree that many modern software have no respect to humans, but that's not because the developers didn't want to implement a yes/no consent dialog or several of them, it's because they think the only boundary they can't cross is what the law says they can't, and everything else is fair game.


Sorry late to reply after work today, but you really got me thinking AR

> everyone gets the same inflexible system. It's not logistically feasible to create a custom experience for each user based on how far they consent to every little thing,

Is it though? Maybe this is where AI is going to be a win - in dynmaic protocols and bespoke interactions. What if we leave my "AI web-browsing agent" to talk to your "AI web server" to quite literally negotiate a whole bunch of preferences. Usual game theory applies. If your bot tries to defect or fuck mine over, or my bot is too greedy or defective we both get a suboptimal outcome? Maybe I am happy to trade certain bits of data like location or budget but will withold or lie about others, say age and gender. In response I might not get all the info I seek. By making the value of our transactional relations explicit some interesting things might happen. Maybe if your Wbserver agent is is a bit too greedy mine will shop around to other sites etc. respects.


Then the AI would be the one making assumptions about your consent. When it gets it wrong and your boundaries are violated, everyone will just shrug and say "well but the AI said it was okay."

In my opinion, algorithms can not fix this. The developer needs to know clearly what boundaries they can't cross, and the only way for that to happen is if there are obvious penalties and deterrents. If Microsoft commits egregious privacy violations but it makes them more money than it costs them, why would they stop?


Why I use Vivaldi: workspaces and vertical tabs let me pretend I don't have a tab hoarding problem.

I know I could probably have this on Firefox with some extension, but I don't like the concept of extensions.


I find this ironic because you could say many of the problems with modern Windows (& Linux) GUI stem from distancing the design from the traditional desktop paradigm and seeking to mobile-fy everything, specially with large click targets.


>The filing includes comments from Eddy Cue, Apple’s services boss: >Microsoft search quality, their investment in search, everything was not significant at all.

Does this mean Microsoft doesn't spend a lot of money on improving Bing? That would be weird because they embedded Bing on Windows 11, and then embedded AI into Bing. I guess things changed since 2018?


I don't think either of those counts as an improvement. MS has a long history of integrating their otherwise not-so-competitive products into their OSes just to get people to use them. And wrt AI, well, things have definitely changed since 2018!


There's a "lot of money" and there's "a lot of money to Microsoft/Apple/Google".


This is such a terrible situation. We need depopulation but every country's economy seems to be based on infinite growth and can't deal with an aging population.

I'm scared the best case scenario will end up being "the world gets so hot it kills enough people to stop getting hotter." I don't know a lot about climate change but I don't thing the Earth will simply reset when enough people die, considering the melting ice caps won't just unmelt because of that.


> We need depopulation but every country's economy seems to be based on infinite growth and can't deal with an aging population.

There is no way of organising an economy that will allow it to support having the demographic of net producers significantly outnumbered by the demographic of net consumers. This is an economic problem, but only in the sense that if you want people to have access to good and service then you need people willing and able to produce those goods and services is economic problem. If you have a demographic collapse, then you don’t have that any more, no matter what system of economic management you’re using.

The idea that we need depopulation is also about as intellectually and scientifically valid as the voluntary extinction movement is, yet somehow a lot more people seem to think it has some credibility.


That doesn't seem to make sense because we're talking about how many people produce and consume things, not the rate of which they produce and consume things.

We just need the working age populace to produce more than everyone can consume. Automation was supposed to achieve that. In fact, I think it has already achieved that long ago?

I'll reword it in a depressing, pragmatic way. We can already produce enough to satisfy the NEEDS of the entire population. So if you only need X jobs to produce for everyone, but extra people are getting born because nobody can just set a limit to how many people are born, then what do these extra people do in their lives? What is going to be their job?

If they produced things, it would be a waste of resources, because the needs of the entire world are already satisfied. But that's exactly what they do. Do we really need fancy clothes, and fancy houses, and fancy food, and TV shows and music? Do we need art? We don't really "need" them, but we already produce enough that we can afford to have people whose jobs is just making these things. But by this logic we don't really need computers either, you can't eat them, so I guess I'm superfluous as well.

I guess the point is, how many of these "superfluous" things we should produce? If population goes down, some jobs won't get as many workers as before. I can't really say "artists are unnecessary, we should just get rid of the entire profession and replace it with Dall-E because it's more productive." In fact, I'm not sure there is any job we can just get rid off to state "we will birth no more people to do that job. It will have 0 workers assigned to it."

And this is of course just speaking theoretically, assuming you could control births and professions with precision, which we can not in practice. People are going to get born and they will do jobs that aren't as productive as other jobs and there is nothing we can do about it. But to say we can't produce enough doesn't make sense if we can't even define how much is "enough."


> If they produced things, it would be a waste of resources, because the needs of the entire world are already satisfied. But that's exactly what they do. Do we really need fancy clothes, and fancy houses, and fancy food, and TV shows and music? Do we need art?

The majority of society’s basic needs have been catered for since ancient Mesopotamia. The only significant innovation we’ve had since then is modern medicine. In regards to needs, everything else since then has just been fine tuning and improving access.

The majority of human development since that period has primarily been delivering for wants. The basic premise of your argument here is “how about you do without _everything_ that you want”.

This also ignores the fact that no, we cannot make “enough” of only the needs, and just remove all of the wants, because they have the same supply chain. You might argue that I don’t need the spark plugs in my car, but a farmer would certainly need the spark plugs in their tractor.


In terms of population reduction we're actually handling that by ourselves, as the absolute number of births peaked somewhere in 2014-2017 and started declining since.

Overall most of the population growth now is younger, larger generations replacing older ones, with the demographic pyramid looking like a sausage being filled with meat.

World population will peak by 2086 at ~10.4bln.


That's only a prediction, an extrapolation by the UN I believe. Many things can change until then and we don't know anything for sure. Even if it peaks by 2086, the population mass until then, with our current way of living, will have devastated the planet.


Yes, I'm using UN numbers, as they have the best track record.

Looking at past projections the population was typically slightly below the estimated figure, so the most likely scenario is that this will be the case here as well.

In any case I don't think that devastation will come to pass, as China's emissions pretty much plateaued already and India's are looking like they did as well. The west has been in decline in that regard for years now.

It's going to get worse - that's a given - but with the recent estimates giving a more precise number for emissions connected with land use change, global CO2 emissions might have plateaued, or even peaked a few years ago already.


That's still too many. Sure, it'll probably level, but why not get in front of that and find a good balance today instead of assuming it'll all just sort itself out magically?


It is sorting itself out "magically". If anything, we don't know what to do with ageing populations.

The previous estimate put the population in 2100 at 11bln and growing, but then China had its census, which revealed a very different demographic situation than anticipated and the pandemic happened, upending those projections.

Recently also India went below replacement fertility rate - that now makes the two most populous nations having crossed that threshold - also way ahead of previous projections.

Estimates of populations of African countries are most likely overstated, just like fertility rates.


We do not need "depopulation, as I specifically noted in my comment there is no issue with our population numbers being significantly higher than they are, let alone at our current levels. The issue is the kinds of technology that we are using and policies that we have, oil-based transportation, cow-based agriculture and to a lesser extent meat-based agriculture, food waste, coal-based energy generation, etc. We need to decarbonise the grid and transportation (we already know how to do this, it is only a matter of doing it), significantly reduce GHG emissions in agriculture through selecting different crops and livestock as well as increasing efficiency (this will need large scale public buy-in, so it needs to be led by the government), and put serious R&D into decarbonising heavy industry like steel and cement production. Once we get that far, we can focus on solving the remaining GHG issues like synthetic fuel for planes + high-speed rail replacements, and carbon capture & sequestration for what we can't solve in a cheaper way. "Depopulation" is an absurd, unrealistic, and unworkable proposal that does not actually solve the central problem and would actively make it worse. Population size is a huge part of what makes us able to achieve significant scientific and technological progress, "depopulation" would actively harm our efforts to restructure our economy and distribute technology as necessary to combat climate change.


Depopulation is the last thing we need. We are carbon based life forms. We can just sequester carbon in human bodies.


That's exactly right. We are taking carbon that was sequestered under ground over a period of hundreds of millions of years and releasing it back into the atmosphere in just a few hundred years. That genie will not go quietly back into the bottle.


Although there are ways to bring it back under the earth: Burry trees deeply underground. Which is of no economic value, but that might be a way, if we dig deep enough and do it on big scales.


I think that would depend a lot on whether depopulation led to rapid restoration of our lost wetlands and I am going to assume no one is making models of that because the wetlands issue gets so little attention generally.


We don't need depopulation. We just need to stop eating cows, riding cars and planes so much, and buying everything from across the world.

(and dozens if not millions other things, neither of which is harder than depopulation)


While this ist part of what we need to do, you are forgetting the industry sector, as the first and biggest polution source, that provides living standard to us, and that needs to be regulated the most for the biggest effect.


I concur. A lot of frontend problems wouldn't exist if HTML had something like XSLT. The fact there's no non-javascript way to reuse HTML code is a already a failure of DRY. Pretty much every web 2.0 technique so far has been just us trying to add DRY to a language that doesn't support it.


No, it's a post on X, formerly known as Twitter.


I guess we can officially shut down the internet.


Maybe he really, really, REALLY doesn't want to switch to Wayland.


People aren't even up to 2FA. Every day lots of people lose access to things because they used 2FA.


Dark Hero: The Revenge of the Failed Atoms

A 2 hour long liberal movie adaptation of Hollow Knight. Release date: before Silksong.


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

Search: