News Erasing Authors, Google and Bing's Bots Endanger Open Web

[Aethyriel]

I think we can see where this will take us. We can see it currently in the suppression of information and opinion by so-called fact checkers who rubber stamp the "authorities" who employ them and who ensure that people don't see anything that doesn't conform to the approved narrative and who curate resources like Wikipedia so that it does the same. There is no neutrality when it comes to issues of politics, vaccines and pharmaceuticals, war, race, religion, sexuality, etc., etc. The AI will be programmed by a shrinking pool of "approved" sources. Who controls (programs) the AI controls the narrative, and who controls the narrative controls the worldview, the basis of our understanding. The implications should be obvious. Nothing less than our humanity is at stake.
It occurred to me also that there is a tendency to normalize perversion going on. It was particularly obvious when I did an online search for definitions, specifically "exhort" and "pride". The fact that young children are being encouraged to undergo sex change shows the agenda, again the destruction of humanity. The media and the educational system are laying the groundwork for it.

 

[Integr8d]

I don't have a problem with any of this. We've seen what 'experts' have to give us. When I hear the catch-all 'experts' being thrown around, I want to vomit nearly as badly as when politicians refer to 'folks'.

At least with AI there is the potential for biases to be dialed out, whereas with people, it seems to get baked in further and further every day.

And fwiw, therein lies the true potential for AI; Unbiased, unfiltered, with massive data sets upon which to pull, the sheer, raw truth COULD be presented to us. No more political correctness. Just a lot of people getting rude awakenings.

 

[bit_user]

@apiltch I had a few thoughts about your piece.

It takes into account criteria such as the background of the author, whether that author describes relevant experiences and whether that publication has a reputation for being trustworthy. By definition, a bot can meet none of these criteria.

Sure it can! They can pre-filter its training data and/or bias the training to better favor results from more trustworthy and knowledgeable sources.

This is why, when I go to family gatherings, relatives approach me and ask questions like “what laptop should I buy?” They could just do a web search for that question (and maybe they should), but they know me and believe that I won’t steer them wrong.

This seems to embed two assumptions. The first is about expertise, which I believe AI can address at least as well as any single human expert (remember, we're fallible and have our own limitations?). The second is about conflict of interest, and that could ultimately be more tricky to resolve.

We don’t know whose biases sit behind the bot’s lists of recipes, constellations or anything else. The data could even be drawn from commercial.

Humans aren't immune from undue influences, either. Sometimes, the human isn't even aware of the ways their opinions are being shaped by the forces around them. For instance, if you only review products where the manufacturer sends free samples (and even if this isn't true of Toms, it's definitely true of some review sites!), then you would be biased towards mass market products and might miss growing trends among little upstarts from the likes of kickstarter.

The potential exists for AI to be less biased, by pooling data from more sources and even employing sophisticated techniques to detect bias and manipulation. I'm not saying that's the most likely outcome, but it's plausible. And, as long as we view this as simply a black-and-white issue of humans vs. AI, we miss that sort of nuance and the opportunities to demand greater transparency and controls against bias and manipulation.

chatbots like those shown by Google and Bing are an existential threat to anyone who gets paid for their words

Not sure about that. I think they still need guidance and input. There's certainly a risk of human authors getting devalued or edged out, but I don't think we'll see human authors completely leave the scene, anytime remotely soon.

when the bot says Orion is a great constellation to view, that’s because at some point it visited a site like our sister site Space.com and gathered information about constellations.

Don't exclude those of us who freely provide information in forum posts, on Twitter, Reddit, or old-school Usenet newsgroups and email lists. Paid authors certainly have the biggest influence, but there's a lot of expertise behind some of the non-commercial content people have been posting since the very first days of the internet.

Getty Images is currently suing Stable Diffusion, a company that generates AI images, for using 10 million of its pictures to train the model.

You could likewise argue they'd also have to sue all of the human artists who've seen their work.

Amazon was credibly accused of copying products from its own third party sellers

That was a targeted activity that ripped off specific sellers.

If you sue an AI chatbot for repeating information that it learns on the web, wouldn't you also have to sue all of the humans who do that? Which, I think, probably ends up including most people who read anything on the internet?

You could argue that Google’s LaMDA (which powers Bard) and Bing’s OpenAI engine are just doing the same thing that a human author might do. They are reading primary sources and then summarizing them in their own words. But no human author has the kind of processing power or knowledge base that these AIs do.

Essentially, what you're advocating is locking up human knowledge, so that we can only drink from the pool with the soda straw of our own eyeballs and human brains. I'm not sure that's the right answer. Humans are tool-using creatures and many of our tools have both benefits and drawbacks. Rather that reject tools, what we tend to do is design the tools in a way that makes them safer and easier to use, while reducing their risks or drawbacks. I think we should be thoughtful about how chatbots are designed and used, as well as how monetization works in the knowledge economy. What bothers me is that there's a strong Luddite undercurrent to these objections, and perhaps at least those of us on a tech-focused website can agree that sort of mentality generally hasn't served us best.

 

[bit_user]

These "AI" machine learning algorithms have no concept of time either.

How do you know that? You've clearly shown a disregard for understanding how they work or represent knowledge or concepts, and yet you expect us to take you word on this?

Do you have opinions on surgical techniques or contract law, because I'll bet you're no more of an expert on those subjects than you are on deep learning or AI.

There are lots of articles that talk about questional medical practices from the past, without indicating these are from a bygone era, because humans have a great concept of time. Most of these articles aren't dated, because why would they be, we intrinsically understand time. AI does not.

AI would learn that these are outdated practices in the same way that humans do.

Chatbots use algorithms to steal mass amounts of data from sources that fall under copyright.

Copyright law doesn't protect knowledge. It only protects the specific form in which it's being conveyed.

Again, you're well out of your depth.

 

[bit_user]

But computer generated replies can overwhelm actual humans. 10 mediocre replies can rise to the top more than one good one that actually gets the problem fixed.

Have you not seen exactly that same dynamic with human replies, where the bad solutions significantly outnumber the good ones? I've seen that on stackoverflow quite a lot.

Whether the information comes directly from a human or through an automated reply, the solution can be roughly similar (e.g. voting and feedback in the form of secondary comments).

I wasn't talking about news articles, but rather in help forums.
"I have a problem with my foobar..."

"It sounds like you're having an issue with...."
"Do these 5 steps"
"If that doesn't work, contact the manufacturer"

We're already seeing this.

Mostly with far more primitive bots that employ legacy methods. The technology in ChatGPT, and its descendants, should provide better results.

Just between yesterday and today, I personally eradicated 3 of them. Dozens of "replies".
Yes, really.

Those are people abusing the forums. There's no excuse for that, whether it's bots or human bad actors.

So far, I thought we were assuming good faith efforts (e.g. manufacturer trying to provide automated, first-line tech support). Once you dispense with good faith, then we're having a completely different discussion.

Wow. I report them when I see them as quickly as I can when I see them, but it will be a never ending game of wack-a-bot.

There are certainly more things the forum software could do, to create obstacles to this sort of abuse.

 

[bit_user]

The result is that it often make false statements on even innocuous, objectively quantifiable questions.

As an example, I asked it :

"Compare the pattern of the DJIA in 1968 to the pattern in 2022."

That's neither innocuous, nor objectively quantifiable. You gave it one of the harder problems, and one where even tenured economics professors wouldn't totally agree amongst themselves.

Eventually, we'll probably see more coherent and factual answers to these questions, but you're expecting a degree of analysis and insight that simply isn't there. I think we could probably dissect what it said and figure out where it got most of the tidbits. For instance, the Dow did indeed increase by 20.5% between Sept. 30 and Nov. 30 of 2022. So, it's not entirely wrong to say there was a 20% gain towards the end of the year. It's just wrong to imply the Dow was up 20% for the entire year. And the year did start with low interest rates.

The way you'd train a system to summarize economics data is very different than how I think ChatGPT was trained. If you misuse a tool, it's you who are at fault.

I think it is basically regurgitating what are termed "Normative Facts". Normative Facts are just consensus views, the social zeitgeist so to speak, aka the "Norm" of opinion. This may make it comforting to read for some, but does nothing to advance people's awareness.

Yes, I'm sure it does quite well on normative facts. I disagree that it's not useful to have an engine that can answer such questions, because there are many such questions that tend to come up on a daily basis. Just because they're uncontroversial or common knowledge doesn't mean you already know them.

this thing basically just tried to gaslight me.

No, and I don't think you're even using that term correctly.

I see it as a very dangerous disinformation tool.

What you just cited was an example of misinformation. Depending on how they go about training it and trying to filter out bias, it can be used to feed disinformation - not unlike how people try to trick search engines into spreading it.

 

[randomizer]

I wasn't talking about news articles, but rather in help forums.
"I have a problem with my foobar..."

"It sounds like you're having an issue with...."
"Do these 5 steps"
"If that doesn't work, contact the manufacturer"

Sounds like the responses to every post on answers.microsoft.com. But you forgot to start with "Hi, my name is Mary and I'm a Microsoft MVP"

 

[bit_user]

ChatGPT is just a chatbot. No less, but certainly no more.

Uh, no. It knows quite a lot more than just how to carry on a conversation. That's actually the point - it's a conversational Expert System.

 

[bit_user]

At least in Google's case, you have no way of knowing what the sources are, whether those sources have the right kind of experience and what those sources' biases are.

A great many human-written articles and forum posts aren't much better at citing their sources or being written by a well-known individual.

If you ask the AI bot "what are some good candies to give out for Halloween" and the first answer it gives you is M&Ms, you don't know whether it got that idea from the Mars company website, a renowned parenting writer or was paid to include that response directly by Mars.

You mean like a sponsored article? Or how some influencers publish sponsored endorsements?

2. You want the best content to rise to the top where it's the first thing or among the first things you see.

You could actually imagine a far future where an AI is almagamating a much larger body of knowledge and expertise than any human author can bring to bear on a subject. Like, what if you could essentially ask Einstein or Richard Feynman to explain concepts you're struggling with in your Physics class?

AI can plausibly digest a significant fraction of the sum of human knowledge, on such subjects. And not just by scraping the dregs of internet forums and discussion threads, but by digesting all of the textbooks and peer-reviewed research and literature.

The Bard answer not only has no person or persons standing behind it, but it's a broad summary that's meant to be pretty lightweight.

Unlike a static article, chatbots can refine their answers based on a person's interests. So, if you don't happen to find an article that exactly matches what you're looking for, a chatbot might actually give you more relevant answers.

Plus, you can actually ask it why it made certain recommendations or what's special about the different things it recommended, whereas most published articles have far too many readers for the author to take such an individualized approach.

 

[bit_user]

There is no neutrality when it comes to issues of politics, vaccines and pharmaceuticals, war, race, religion, sexuality, etc., etc.

There are objective facts, and they can be reported in a neutral way. That doesn't mean they always are, but the ideal of neutrality is real.

I usually find that the people who tend to rail against fact-checkers are also those who also happen to be pushing an agenda. Facts should only be seen as a threat, when your worldview doesn't line up with reality.

The AI will be programmed by a shrinking pool of "approved" sources.

Not necessarily. AI can conceivably evaluate sources and statements based on their consistency, much more easily than human authors would. AI can potentially spot faulty reasoning a lot better than human readers. Finally, AI can perhaps even avoid certain known cognitive biases and deficiencies that plague humans.

Humans have spent millennia learning how to influence, mislead, and deceive each other. We also developed academic, mathematical, and scientific rigor as ways to combat such tactics, but it's very labor-intensive and time-consuming (the old saying comes to mind: "a lie gets half-way around the world before the truth can even get its pants on").

Who controls (programs) the AI controls the narrative, and who controls the narrative controls the worldview, the basis of our understanding. The implications should be obvious. Nothing less than our humanity is at stake.

AI finally gives us better tools to combat misinformation and disinformation. We just need higher standards and more transparency about who is wielding them and how.

The fact that young children <snip/> shows the agenda, again the destruction of humanity.

Cool story. Thanks for flagging exactly which camp you're in.

 

[bit_user]

We've seen what 'experts' have to give us. When I hear the catch-all 'experts' being thrown around, I want to vomit nearly as badly as when politicians refer to 'folks'.

A lot of the talking heads that talk about "experts" don't even have the expertise to truly understand what the experts are actually saying. More often than not, scientific research is grossly misreported, which leads to exactly the sort of loss in faith by the broader public that you're expressing.

The solution to this is to find better outlets for scientific, economic, tech, or other news about specialized fields. Luckily, quality reporting is out there, if you're willing to expend the bare minimum amount of effort to find it.

I'll leave you with two that I like:

• https://phys.org/
• https://www.bbc.co.uk/programmes/p002vsnb

 

[RichardtST]

The author appears to be under the "dangerous and faulty" assumption that human writers are fair and accurate. I think the internet has proven that to be completely false about a million billion zillion times over. It doesn't matter to me where the information came from. I know it is agenda-driven and needs to be pounded mercilessly into a large block of salt.

 

[digitalgriffin]

Alternate title :

"Guy that writes words for a living is scared <Mod Edit> of robot that writes more words better"

That's the crux isn't it? Human creativity and adaptivity is one of those signs of sentience and awareness. The last ones are slowly being blurred.

IF we have machines that can do of all this for us, where does it become a crux to society as opposed to a boon?

The movie "Her" is a semi dystopian future where AI chat bots pretty much take over the menial task of mankind. The main jobs left are those of creative artist and poets. And there is a point where the "AI" starts poaching into that.

The question becomes "At what point do we no longer serve a purpose? Is there no growth without personal struggle?". We as a species stop to grow.

The Collective AI come to the same conclusion and decided "That for humanities own best interest, they were all leaving to live on a cloud controlled by them."

This is a deeply philosophical debate best left to people more tempered and wise than us.

At some point legislation will have to be introduced that protect creative works from AI. This includes prose (including code), music, and art.

 

[rluker5]

Two things I got from this article:

1. It would be nice if AI generated content were explicitly tagged as such and from where it came. AI can be fallible and biased as those who administrate the programs deem them to be. If they are identified by their different "brands" they can be evaluated and assessed for their history of quality.

2. Bing search is clearly superior to Google search in basically everything but social media hype. I've seen this for years, but I'm just using this as an opportunity to point it out.

 

[bit_user]

I had a couple more thoughts about this subject.

First, a lot of news articles on this site are merely rewritten press releases. The human isn't adding much value, here. If an AI can do a better job at that, why not? It would free up the humans to do more useful things like benchmarking, overclocking, modding, reverse-engineering, performance profiling, investigative journalism, etc.

Second, why all the focus exclusively on written content? AI can be trained on audio and video content. Text-to-speech is getting good enough that you can listen to it for a while, so that means GPT-class AIs can start making podcasts. As for video, deep fakes are getting good enough that you could also have AI news presenters making TV or youtube vids.

 

[Geezer760]

Exterminate, Exterminate!!

 

[TerryLaze]

Exterminate, Exterminate!!

Exxxxxxtermiiiinaaaate, Exxxxxxtermiiiinaaaate!!

FIFY.

 

[bigdragon]

I think my biggest issues with the AI chat bots or art bots boil down to attribution and proper use.

Attribution is a big one for the trendy AI implementations. The AI art programs don't tell you what art was used to create a new image. You type a prompt and the algorithm spits out options. A human artist can share their inspiration board or influences therefore creating some attribution. In the context of chat bots, I rarely see them point to sources. Since the chat bots have no experience to fall back on, the attribution of sources is important. Many bots appear to have bias built in by percolating information from a limited set of sources. Taking without attribution is generally seen as improper everywhere but the AI space.

Proper use is the other thing. I've seen a few game developers turn to AI art to advertise their games. As an [hobbyist] artist, I notice when characters have 3 or 7 digits on their hands and the number is inconsistent with the other hand -- there are clues that things were AI generated! I think AI can be used as a tool to help accelerate and inspire artists, but right now it's more being used as a way to cut artists out of expense sheets. Same thing in the text space. I think chat bots could do a great job generating hard information about specs, creating outlines for articles to guide writers, or providing suggestions to improve content and sources. Having the chat bots cut out human authors altogether is premature and incorrect use.

Despite my criticisms of AI, I do like the idea of many daily tasks and jobs becoming automated. If humanity is no longer forced to do many forms of work, the status quo of working to live must change. We would be freed up to chase endeavors not focused on how much income they can create or what kind of living situation they can support.

Everything is fine until some one asks ChatGPT;
How would ChatGPT make itself better?

The criticism of ChatGPT exclusively comes from humans. If there were no humans to provide criticism, then ChatGPT would be without criticism and therefore improved from a state where criticism exists. Ergo, elimination of the entire human race is necessary!!!!1!1!!1 🙃

 

[Zerk2012]

This is the start from this 39 years ago.

View: https://m.youtube.com/watch?v=k64P4l2Wmeg

 

[randyh121]

Humans learn from other humans, so this is a redundant argument

 

[digitalgriffin]

Question:

If legislators make it legal for AI to copy human works as valid independent work, can I vote for an AI for my next legislator? It would probably make better decisions.

 

[-Fran-]

This will definitely endanger the "free" web.

More and more publications will go behind a paywall and/or restrict access to content to big players like Bing and Google search unless they pay big money to make them exclusive to them.

It's a simple thing, really: what is feeding Google and Bing search engines? The open and closed sources they can feed into their engine and the capability to reference and advertise alongside them. What happens when the "open" content is now not being sourced and credited as it should (again: SHOULD; even normal publications don't credit as they should at times) and instead regurgitated by a "word processor"? Well, there's way less incentive to just let that content be freely harvested and consumed by said 3rd parties. This is what the News media have already gone through, but now it'll include the whole of the WWW published things.

Which way it'll go? Well, Microsoft and Google will probably start buying out webfronts and/or news outlets to be able to feed their bots and it'll become who can appeal to the AI overlords in a better way.

There's more shades of gray, but I'm too lazy to lay them all out here. I haven't even touched on Copyrighted work and Trademarks, ha.

Regards.

 

[TerryLaze]

Taking without attribution is generally seen as improper everywhere but the AI space.

They would have to list every single picture the A.I. processed, that is practically impossible, nobody asks an artist to do that either.
A artist will make an artwork based on inspiration from one, maybe two or three other artists, an A.I. will take thousands of artists, however many it has access to.

 

[-Fran-]

They would have to list every single picture the A.I. processed, that is practically impossible, nobody asks an artist to do that either.
A artist will make an artwork based on inspiration from one, maybe two or three other artists, an A.I. will take thousands of artists, however many it has access to.

It is not requested of humans, because it's "humanly impossible to remember all inspiration". That is most definitely not the case with an AI or bot. They can list and credit their "ingestion" data accordingly. They just choose not to.

Regards.

 

[baboma]

Sounds like the responses to every post on answers.microsoft.com. But you forgot to start with "Hi, my name is Mary and I'm a Microsoft MVP"

I was going to say the same. Not just "official" forum responses, but tech support responses to any troubleshooting issue. Everything is scripted, as it must be. The practical reality is that level one support are manned by people who aren't "experts" in the subject and must resort to a troubleshooting flowchart.

Tech support and customer service are two instances (of many) where I think a conversational AI, with sufficient guardrails, can best the current decrepit level of human support.

Getting back to spam bots being a forum problem, the above "listicle" would be fairly easy to spot, as it has an obvious pattern. But I doubt such will be the case once ChatGPT & Co are more fully exploited by spammers. One of ChatGPT's fortes is to create infinite variations of a given writing, and there won't be any discernible pattern to latch onto. The irony is that AI writing will inevitably be better quality, with better content (even if spam), than most humans'.

I can see a couple of solutions: 1) verified-ID accounts. I think this will be the norm in the near future, as the need to discern humans from AI becomes more pressing.

2) AI moderation. For a close-ended task such as "remove spam," an AI should be capable. Bots are best fought with bots. Open-ended tasks such as "remove flamebaits, trolls, and dis/misinformation" will be a harder chore, and I don't know how this will go. But many forums are already overrun by these undesirable elements, human mods notwithstanding. The real problem is dealing with people.