4 ms·

Agree with your first paragraph and disagree with the second: Google really has it coming. They bred the SEO industry that has essentially turned websites into

Agree with your first paragraph and disagree with the second: Google really has it coming. They bred the SEO industry that has essentially turned websites into SPAM. There isn’t a single website that doesn’t have you scroll through lengthy SEO-optimized bullish.. content. As for ChatGPT “stealing” content: I’d rather take the view of aggregating human knowledge as a positive. Imagine ChatGPT aggregating, summarising, weighing and dumbing down academic content. And if as a result all that knowledge became accessible by a broad audience. What could this do for the average person’s scientific literacy? Or fake news for that matter?

palata 4y ago
> What could this do for the average person’s scientific literacy? Or fake news for that matter? I guess not much better than when the media say "A study proves that X": it doesn't mean anything, it doesn't mean the study has any value, it doesn't mean that there is a scientific consensus. Yet people routinely consider X to be proven (see what happened in the early days of the pandemic). Add ChatGPT to the mix, and you don't even have the links to the original studies. I can totally imagine it being even worse.
Nathanba 4y ago
>As for ChatGPT “stealing” content: I’d rather take the view of aggregating human knowledge as a positive It may be positive for humanity in general but that wasn't really the point that was made. It's a bad incentive for the individual knowledge creators. I suppose we could argue that it's high time that knowledge is less artificially guarded behind indiviual humans and that it gets a standardized format and available for instant query. That "format" is therefore the knowledge graph/db of the chatgpt trained model. Maybe the answer here is that the AI programmers need to figure out a way to link to the original creators so that you can at least look at the source of the knowledge. Surely that should be doable. Because I may trust the ChatGPT3 bot a lot but maybe I also want to check out the forum link where it thinks it accumulated the trained data piece that resulted in its answer. And if that list of links is extremely long then so be it, show me ranked source links of 500 websites. It will be no different than looking at a google search.
- feanaro 4y ago
  Less guarded behind individual humans, but more guarded behind individual corporations. This is the real problem.
anileated 4y ago
Aggregating human knowledge with LLM is a negative, unless it provides attribution. Humans are selfish, we are incentivised to do the work because we take pride in it. There's no pride if you are producing for LLM operator who then charges people to provide your work as an answer.
davidy123 4y ago
I don't think that "Google has it coming" for this. In fact I think they support the open web much more than competing alternatives (Facebook, etc) who want to create walled gardens that require auth (identity). It's up to content providers if they want to completely target SEO and create poor quality content, and some are constantly gaming the system. What Google is doing is looking for more meta content, which is a very beneficial for everyone. meta keywords didn't work, and SEO goes beyond them to provide precise information about a site's content. Anyone can use SEO techniques for good and bad, it's by no means proprietary. It's the general Web that didn't keep up and provide better solutions for quality and precise content, solving problems like "This blog is not about avocados" or links in a sidebar about avocados that lead search results for avocados to that page. Microformats and other approaches were suggested, but they were basically too complicated and tools didn't support them. Where Google has fallen down is not providing more search/browsing tools to allow individuals to tune sites and keywords. There are many obvious opportunities in Google search, news, etc that would make the experience amazing compared to the picking-through-the-repetitive-garbage we see today. I suspect there is a battle between their tech and marketing about this, and I also think Google's ML team are using our garbage-picking to tune their algorithms rather than letting people just choose what they want to see. That is disgraceful, and they won't change it until there is competition which makes me lose a lot of respect for Google, but I don't blame SEO directly for this and since SEO is based on standards anyone could use their own "Web reader" to add a layer of filtering without special agreements, unlike content by companies that try to create walled gardens. edit: comment about "web reader."
PodWaver 4y ago
I don't think it'a good thing it all. There are multiple reasons for this One is for something to have quality you need to support people making it. And the AI doesn't make stuff, it just repackages it. If there was no one on the internet writing about 'x' topic the AI couldn't give you an answer. And by pretending AI does actually write it we are lowering the support for the real creators which makes them less likely to write. They need to eat, and it's great if they can get their bread by writting because they do their job and get their bread. If they no longer can do that, they will put hours into something else which gives us less writing. Second one is that AI will probably in my eyes be the leading subject of making sure there's more fake news and less scientific literacy. This would be too big to sum up in short text form. But how I see it the important stuff happens because of perspective/context changes. But the web and AI programs have next to no context. If I go on the web I get a small glimpse of someone's view, before it gets mushed together with someoene elses short view, and then an ad, and some completely unrelated paragraph about some event happening on the other side of the globe, and so on and so on. It's like studying something in a circus. There's no context in a circus outside of chaos and I think the same applies to technologies we build and are building. There's lot of data, but very little communication or meaning happening. Now imagine how the world looks if majority of what is written is made by machines we've build to spin around existing (already crazy with no context) written word and produce more of it?
notahacker 4y ago
> What could this do for the average person’s scientific literacy? Or fake news for that matter? Reduce it, by providing concise, well-written and authoritative sounding but often inaccurate answers to their question (increasingly likely to be based on 'fake news' as the corpus expands) without source or context instead of requiring people to read up on the background or basic principles to figure it out for themselves.

palata 4y ago

> What could this do for the average person’s scientific literacy? Or fake news for that matter? I guess not much better than when the media say "A study proves that X": it doesn't mean anything, it doesn't mean the study has any value, it doesn't mean that there is a scientific consensus. Yet people routinely consider X to be proven (see what happened in the early days of the pandemic). Add ChatGPT to the mix, and you don't even have the links to the original studies. I can totally imagine it being even worse.

Nathanba 4y ago

>As for ChatGPT “stealing” content: I’d rather take the view of aggregating human knowledge as a positive It may be positive for humanity in general but that wasn't really the point that was made. It's a bad incentive for the individual knowledge creators. I suppose we could argue that it's high time that knowledge is less artificially guarded behind indiviual humans and that it gets a standardized format and available for instant query. That "format" is therefore the knowledge graph/db of the chatgpt trained model. Maybe the answer here is that the AI programmers need to figure out a way to link to the original creators so that you can at least look at the source of the knowledge. Surely that should be doable. Because I may trust the ChatGPT3 bot a lot but maybe I also want to check out the forum link where it thinks it accumulated the trained data piece that resulted in its answer. And if that list of links is extremely long then so be it, show me ranked source links of 500 websites. It will be no different than looking at a google search.

feanaro 4y ago
Less guarded behind individual humans, but more guarded behind individual corporations. This is the real problem.

anileated 4y ago

Aggregating human knowledge with LLM is a negative, unless it provides attribution. Humans are selfish, we are incentivised to do the work because we take pride in it. There's no pride if you are producing for LLM operator who then charges people to provide your work as an answer.

davidy123 4y ago

I don't think that "Google has it coming" for this. In fact I think they support the open web much more than competing alternatives (Facebook, etc) who want to create walled gardens that require auth (identity). It's up to content providers if they want to completely target SEO and create poor quality content, and some are constantly gaming the system. What Google is doing is looking for more meta content, which is a very beneficial for everyone. meta keywords didn't work, and SEO goes beyond them to provide precise information about a site's content. Anyone can use SEO techniques for good and bad, it's by no means proprietary. It's the general Web that didn't keep up and provide better solutions for quality and precise content, solving problems like "This blog is not about avocados" or links in a sidebar about avocados that lead search results for avocados to that page. Microformats and other approaches were suggested, but they were basically too complicated and tools didn't support them. Where Google has fallen down is not providing more search/browsing tools to allow individuals to tune sites and keywords. There are many obvious opportunities in Google search, news, etc that would make the experience amazing compared to the picking-through-the-repetitive-garbage we see today. I suspect there is a battle between their tech and marketing about this, and I also think Google's ML team are using our garbage-picking to tune their algorithms rather than letting people just choose what they want to see. That is disgraceful, and they won't change it until there is competition which makes me lose a lot of respect for Google, but I don't blame SEO directly for this and since SEO is based on standards anyone could use their own "Web reader" to add a layer of filtering without special agreements, unlike content by companies that try to create walled gardens. edit: comment about "web reader."

PodWaver 4y ago

I don't think it'a good thing it all. There are multiple reasons for this One is for something to have quality you need to support people making it. And the AI doesn't make stuff, it just repackages it. If there was no one on the internet writing about 'x' topic the AI couldn't give you an answer. And by pretending AI does actually write it we are lowering the support for the real creators which makes them less likely to write. They need to eat, and it's great if they can get their bread by writting because they do their job and get their bread. If they no longer can do that, they will put hours into something else which gives us less writing. Second one is that AI will probably in my eyes be the leading subject of making sure there's more fake news and less scientific literacy. This would be too big to sum up in short text form. But how I see it the important stuff happens because of perspective/context changes. But the web and AI programs have next to no context. If I go on the web I get a small glimpse of someone's view, before it gets mushed together with someoene elses short view, and then an ad, and some completely unrelated paragraph about some event happening on the other side of the globe, and so on and so on. It's like studying something in a circus. There's no context in a circus outside of chaos and I think the same applies to technologies we build and are building. There's lot of data, but very little communication or meaning happening. Now imagine how the world looks if majority of what is written is made by machines we've build to spin around existing (already crazy with no context) written word and produce more of it?

notahacker 4y ago

> What could this do for the average person’s scientific literacy? Or fake news for that matter? Reduce it, by providing concise, well-written and authoritative sounding but often inaccurate answers to their question (increasingly likely to be based on 'fake news' as the corpus expands) without source or context instead of requiring people to read up on the background or basic principles to figure it out for themselves.