It’s speculated libgen is in there as well.
It was an interesting time, because (from memory) 2000 was still "It's pretty complicated for average person to create and post content on the web" and transitioned to 2010's "There are highly polished, easy to use, free SaaS solutions that a trained dog could use." With mobile versions of same in the next few years. (All dates mass market penetration, not first release)
There was a direct correlation between ease-of-publishing and less intellectual discourse on the web.
"Info decay" was definitely a good way to put it. It wasn't that people were necessarily dumb or angry, but that there was a lot more honestly incorrect information online.
Before, there were crackpots, but it was pretty easy to tell they were crazy.
Having seen both, the banal wrongness is far worse.
I think that's one reason wikipedia really took off ~2005+. People were craving an easier way to get even roughly-right information.
blockchain?
all that pointless CO2 for nothing
So: We are already there, all without ChatGPT... ChatGPT (with its cool ways to make it nondeterministic, like: replacing tokens with synonyms and raising the temperature. Or: retraining it...) will make things worse definitely.
Probably the way to go is for {academia,sects,orgs with sense for humanity,...} to build stores of attributable information and then start shunning the branches which produce bullshit (this is what scientific publishing should do already.... it doesn't... even without ChatGPT there's already templated articles strewn around the corpus and noone cares). The main issue there is, that this means that they will also have to stop using a lot of technology (and buy a 32nm foundry...), because many orgs are probably already poisoned.
This was the first result: https://applications.messergroup.com/co2-applications
There were a number of other relevant results. Papers on various applications of CO2, forum posts, etc. Nothing on emissions.
Sure google is a lot worse these days than it used to be, but it still seems plenty usable to me and I have no issues finding what I want with some basic operators.
The problem imo is that Google was good enough that basic operators were not required, as such the avg Joe is amazed when finding out that such operators exist, and Google doesn’t really promote the advanced search page these days.
Sure this talks a lot about emissions, but it also seems to include a detailed breakdown of various industrial uses for CO2, with numbers, projections, statistics and all.
First result for “co2 industrial usage numbers”, without quotes - just like the sibling comment.
You need to specify atleast a few keywords to clarify what you’re looking for - search engine nor llm chatbot nor HN commenters can read your mind…
It's one of the most common patterns I come across. I search word X but find a bunch of word Y stuff that's not what I'm looking for, because word X and Y appear together a lot in Y's context. So I add -Y to the query.
It's become a tool that basically anyone can at least get some information out of. Googling used to be a skill that non-technical people struggled with.
Many of my coworkers feel the same. And this started way before GPT was conceived.
Besides, your post somehow feels like generic technology-angst to me. It seems like whenever we invent something new, there are those people that predict this is finally our demise...
GGPT is a tool, like every other technology. Use it wisely, and you will likely benefit. Use it without engaging your brain, and you will like see negative effects.
ChatGPT is a symptom that perhaps finally raises discussion of the root problem above the noise. Yeah, ironic.
In other words, it's more of a social than a technical issue.
And without commercial buy in, I can’t see why anyone apart from computer geeks and scammers would want to use it. The first question your average users would ask is “why isn’t $SHOP on there”.
Much as I dislike the modern web, it’s what the people have chosen in exchange for free services (like email) and convenience (like online shopping).
And to repeat my earlier point, an EULA isn’t going to keep your vision clean. Malicious actors aren’t going to give a hoot about any EULA. And any business that might take an EULA seriously would simply choose not to use such a minority platform anyway.
Also if this “EULA” is targeting how companies use your platform, then it’s not an EULA. What you’re talking about is a different type of contractual agreement.
I'm not sure more, better rule are going to stop a rule engine from participating. If anything, formalized rules merely makes it easier to train the machine learning system du jour to follow them.
My modest proposal is to go the other direction - fill the entire internet with garbage data so that the generative models trained on it become useless. Clearly there's nothing that could go wrong with this plan.
The masses are still on mainstream platforms, but more and more people are nauseated by the instagrams of this world.
Sooner or later some group will realize it becomes necessary to dissolve the bands which have connected them with another, and you'll have a formal split.
In short, I think people are wired to do better in smaller groups and I welcome this balkanization, albeit with the caveat that the communities can interconnect and grow/shrink/die/emerge organically to suit the needs of the people within them instead of commercial investors.
But on the other hand it's hard to deny that the mainstream internet is getting worse and worse. I grew up with the internet as my nanny, but I'm moving closer and closer to the idea that smaller groups focused on specific interests really do provide better experiences.
Perhaps it was never meant to be, and the sweet spot is just living your life completely offline and using the Internet for Wikipedia and Whatsapp.
[0]: https://www.goodreads.com/quotes/7444411-it-s-like-in-the-gr...
This sounds like a legit use for ChatGPT :-)
Like how facebook or github account depend on your phone number, or free trials depend on your credit card number.
I would argue this is a problem stemming from the way "AI" is presented rather than anything to do with "AI" itself.
That is to say, fucking nobody trusts autocorrect to be correct and "AI" is really just a more complex form of autocorrect.
But people hear "AI" and think AI as in Transformers, Terminator, R2-D2, etc. They're intelligent, right? And that's where the presentation is doing significant harm. There is nothing intelligent, let alone sentient, about "AI".