.... I find the AI overviews very useful. It sometimes takes a lot of time going through a lot of irrelevant junk to find a specific answer to a question. The AI overview, with an artfully crafted prompt, often cuts right to the chase.

acting like it is nowadays by forcing it on their users without consent, and wasting countless environmental resources in the process.

.... You can ignore it. Just scroll down. You might even be able to find a search site that doesn't have it at all. But scroll down is probably the easiest.

Here you go: https://www.democraticunderground.com/10143677089#post10

No AI Duck Duck Go

AI-generated answers..and at the point that the AI-generated answer appears, resources have already been wasted anyhow, so simply choosing to not read it doesn't legitimize it.

If you really don't think forced use isn't happening, you also must not be aware of what's been happening on social media (I recently quit using it largely because of this). Unless you live in the EU, there is no real opt-out from it. Just for starters, Meta (the parent company of FB and IG) is treating all publicly posted content--and now, even private DMs on IG, which used to be encrypted--as tools for training AI. I could go on and on.

It is increasingly being forced on employees in the workplace as well, with no real recourse.

I think you're thinking of this as just a search engine issue, which it absolutely is not. It is also worth noting that many of the search engines that do give you the option to disable it intentionally make it very hard to find let alone enable that option.

I use NoAI DuckDuckGo and already mentioned it in a reply to someone else, but I do appreciate you promoting it as well!

don't know whether it's correct. If you do that checking, you haven't saved much time. And even if you do the checking, you don't know how much much important information was left out by the AI filtering what you learn.

... since they provide links in the summary itself. And it isn't necessary or important in all cases. For instance, I asked about some of the predatory habits of the red shoulder hawk that I have nested in my yard. Not exactly putting together a legal brief.

I also find it useful for counter fact checking. For instance, a DU post claimed that Susan Collins had multiple affairs, including the man who became her husband while his wife was dying. I googled this and got an AI summary that corrected the DU post. So, instead of fact checking the AI, let the AI do some fact checking.

Try it, you might find it more useful than you have been lead to believe.

to fact-check yourself to fact-check others.

Since the AI companies themselves advise people to check genAI results - not because they're good or responsible companies, but because they're trying to shift liability for AI models they know will make mistakes to the user - it's careless at best not to check. And worse than careless if you relay "information" from an AI to others, especially if you don't let them know your source is a chatbot.

.... "fact check" it? If so do you fact check the facts that you checked? Maybe you spend a lot of time doing recursive, circular, and probably redundant fact checking.

In other words, "some guy on the internet" or "some page on the internet" is not good enough by the standards you seem to be advocating, and, in the end, you would end up doing as much work with a regular search as you would with an AI search.

This exaggerated, paranoid level of fact checking that is supposedly required for an AI search result is not actually necessary and, if it were, would be just as required for a regular search. In both cases the same level of caution is required, based on the purpose and seriousness of the question, the complexity of the subject, and the background knowledge that the searcher brings to the activity.

In many cases the level of effort required for a reasonable level of caution is much less for an AI answer than for a "regular" search, because the AI search provides a link to several resources for each point in the answer. You can sometimes make an assessment just by knowing the source, often, for instance, Wikipedia. And you can click on the links and judge for yourself if something is sus.

This whole "AI is evil" panic is far too exaggerated and overly judgemental.

to content the AI is supposedly referencing. That's also a huge problem in AI slop papers turned in by cheating students, and AI slop papers submitted to medical and scientific journals. Apparently impressive fake references.

And even if the AI by happenstance (since they aren't reasoning) gets something right, what they get right is often ripped off almost verbatim from websites the AI companies repeatedly steal content from, sites they're depriving of traffic and revenue. AI search is killing the internet, expropriating its content for the profit of a few billionaires, and when you use AI overview you're helping them do that.

.... internet search, the topic of the OP. Not student essays, which is, as you say, cheating and also self destructive. Or medical or scientific papers. Or legal briefs. All of which have problems.

AI is probabilistic, but probabilistic is not happenstance. In some contexts it can become a form of consensus, and in a web search consensus can have value.

results untrustworthy, and the more people use it, the dumber and more dependent they'll be.

.....the host will die, and it will become obsolete. We will all go back to some golden age, and become smart again.

But I doubt that. It's just a new thing and new things often cause unwarranted concern, and even panic and fear.

Search: Mathematicians protesting calculators

WARNING - AI Response:

The most prominent historical protests against calculators by mathematicians and educators occurred in the 1980s. These protests were primarily led by educators and traditionalists who feared that the introduction of calculators would cause students to lose foundational mental math skills and numerical understanding.

The 1986 NCTM Protest - The most famous of these demonstrations occurred during the April 1986 annual meeting of the National Council of Teachers of Mathematics (NCTM) in Washington, D.C.

The Catalyst: The NCTM released recommendations advising the integration of calculators into math classrooms, even in early elementary grades.

The Protest: A group of concerned math educators and mathematicians—most notably led by John Saxon, author of popular math textbooks—picketed outside the meeting.

The Arguments: Protesters carried signs warning that "calculators are crutches that will cripple our kids' minds" and argued that rote memorization and manual calculation were essential to cognitive development. They argued students shouldn't use calculators before mastering mental math and basic arithmetic.

The Lasting Debate - This historical standoff represents a classic clash between embracing new technology and preserving traditional educational methods.

The Pro-Calculator Stance: Educational bodies like the NCTM argued that calculators freed students from getting bogged down in lengthy computations, allowing them to focus on higher-level mathematical problem-solving and concepts.

Modern Parallels: This historical friction over calculators is frequently compared to modern debates over artificial intelligence and tools like ChatGPT in the classroom.

as genAI.

And FWIW, there are already studies, including from Microsoft, showing AI use dumbs down users.

Btw, please don't post AI slop on a message board for discussions between people. I hope you haven't been posting a lot of AI-generated messages here.

....will continue to do so, as as irony, or other considerations may demand.

I gave you fair warning! Any factual whoppers?

know better.

.... you might as well complain about YouTube.

based on theft. Two tremendously bad decisions right there.

.... through any means, that could be trusted with the assurance that it would not make mistakes. There is no risk free source of information, or risk free access to information. In the case of browser search, AI does not seem to me to be out of the norm.

If you, I, and the world itself are still around in ten or so years, we should get together and compare notes on what went wrong and what went right.

See you then!

I really enjoyed reading what you had to say in these exchanges not only because it was well written and well reasoned, but because you stayed calm and did not rise to the insults that were being not so subtly thrown at you. One of my frustrations with DU at times is that there are a great many people here who seem uncomfortable with having their ideas challenged, no matter how politely you do so. It can be quite an echo chamber, especially with regard to the topic of AI, which seems to have so much hysteria surrounding it.

Thanks again.

.... more aggressive in expressing themselves are at least trying to be civil and add to the conversation in positive ways. Maybe that doesn't always work out, but It's a lot better here than elsewhere, even if not perfect!

Perhaps that’s why I felt compelled to compliment you for the manner and tone you used in your responses.

I hope you have a pleasant day

You can't count on any of it to be accurate. That may not always matter with casual searches, but for anything where accuracy is of paramount importance, you have to fact check every bit of it.

I found the AI overviews helpful at first, but if I'm looking for something citable to support a point I want to make (which is often the case), the overview often feels like an obstacle between me and the specific piece of information I'm looking for.

Ymmv, of course.

.... When they first came out, I noticed some real whoppers, but not as much anymore.

I also use some terms in a very technical sense, and get responses that are focused on the "common" meaning of the term. I then have to rephrase the prompt to get to the meaning that I intended. But it takes the form of a dialog, which seems to make it more productive than retyping the search over and over again.

As you say, other's mileage may vary, a lot depends on what you are trying to accomplish, and, as always, different strokes for different folks.

I find them to be useless.

That's just my experience, though.

Search results at one time "cut right to the chase" before Google destroyed it.

.... Googles evolution dose not match up with mine.

Different strokes for different folks.

...

... and it does not align with my own personal judgement. Maybe you could explain further. What sort of objective criteria are you basing this assessment on? What exactly is "plain" search? In what way, and when, did Google destroy it?

I used Google heavily for more than 25 years. I used it well enough to take advantage of AdSense in its early days, so I'm not speaking as a casual user. My experience has been that I used to be able to find obscure information quickly. Now I have to wade through optimized content and sponsored results. Queries that once yielded specific pages now return generic high-authority sites while small independent sites are harder to locate. The amount of irrelevant material has increased, so more time and effort is required to achieve the same result.

That's the subjective part of my answer.

I'm not claiming that my experience proves the case, of course not. I'm saying that my experience led me to suspect that something had changed. The experience is subjective. The underlying question—whether Google Search became less effective at helping users find relevant information efficiently—is objective and testable.

But the broader question isn't subjective. Researchers have documented the rise of SEO-driven content, Google has spent more than a decade issuing updates aimed at combating low-quality and manipulative results, and the search results page itself has evolved from "ten blue links" into a far more commercial and feature-heavy interface. Whether search has become less efficient at helping users find relevant information is an empirical question. My experience suggests the answer is yes.

Search results became increasingly optimized for Google rather than users. This isn't controversial. It's one of the defining developments of the modern web. SEO evolved from helping sites be discoverable into an enormous industry devoted to understanding and exploiting Google's ranking incentives. Researchers have documented this repeatedly.

A recent longitudinal study found that highly ranked pages tended to be more heavily optimized but often judged to be lower quality, concluding that SEO can work against users' perceptions of expertise and quality.

Is Google Getting Worse? A Longitudinal Investigation of SEO Spam in Search Engines
https://downloads.webis.de/publications/papers/bevendorff_2024a.pdf

Google itself recognized the problem. The entire history of Google's major updates tells this story.

• Panda (2011) targeted "content farms."
• Penguin (2012) targeted manipulative linking.
• Helpful Content updates targeted low-value content.
• March 2024 updates explicitly claimed they would reduce "low-quality, unoriginal content."

Google has spent over a decade trying to undo problems created by incentives within its own ecosystem.

Google Changes Search Algorithm to Oust Content Farms
https://www.cmswire.com/cms/web-content/google-changes-search-algorithm-to-oust-content-farms-010325.php

The search results page itself changed dramatically. Google used to be "ten blue links." Academic analyses of archived search pages show that the results page evolved into something much more complex: ads, featured snippets, shopping modules, maps, knowledge panels, direct answers, Google-owned verticals, AI summaries.

Researchers describe modern search results pages as "feature-full" interfaces that increasingly provide answers directly rather than simply pointing users to external sites.

The Evolution of Web Search User Interfaces -- An Archaeological Analysis of Google Search Engine Result Pages
https://arxiv.org/abs/2301.08613

There is evidence that many users perceive declining quality. Even major newspapers have covered it. The Guardian summarized the debate this way:

"Critics argue that Google increasingly surfaces spam, clutter, and commercially motivated content, while struggling to combat SEO-driven degradation." Importantly, the article also notes that measuring search quality is difficult because results are personalized and constantly changing.

‘Google says I’m a dead physicist’: is the world’s biggest search engine broken?
https://www.theguardian.com/technology/article/2024/jul/20/google-is-the-worlds-biggest-search-engine-broken

The irony is that trying to use Google Search to find objective evidence that Google Search has become worse turns out to be harder than it should be. Twenty years ago, that sentence would have sounded absurd. Today, many people immediately understand what it means.

Google Medical Update: Why Is the Search Engine Decreasing Visibility of Health and Medical Information Websites?
https://pmc.ncbi.nlm.nih.gov/articles/PMC7068473/

Is Google making search worse to sell more ads?
https://journalrecord.com/2025/02/20/is-google-making-search-worse-to-sell-more-ads/

The Continuous Log of Google Search Changes
https://uberall.com/en-us/resources/blog/the-continuous-log-of-google-search-changes

How Search Engine Optimization (SEO) Grew from Nascent Stages to AI
https://www.researchgate.net/publication/378633575_How_Search_Engine_Optimization_SEO_Grew_from_Nascent_Stages_to_AI

Study Shows Decline in Google Search Quality and Reveals Path for Generative AI Adoption
https://synthedia.substack.com/p/study-shows-decline-in-google-search

That is the objective part of my answer.

.... that's a lot to digest, I'll have to work my way through it.

One note, and this IS a subjective judgement, is that the explosion of clutter in search results is what I find I am able to avoid when I say that AI "cuts to the chase." I get a very focused and consist answer without having to piece together bits and scraps from pages that otherwise just repeat the same low quality information over and over.

Again, thanks for taking the time to put together a very comprehensive answer to my request. I'll start looking into the sources you provided.

I appreciate you considering what I had to say.

That's a fair distinction, and I think it may explain why people have such different reactions to AI search.

I don't disagree that AI can cut through the clutter. In fact, I suspect its popularity is partly evidence that the clutter exists. If plain search were still delivering concise, relevant results as efficiently as it once did, there would be much less demand for AI summaries in the first place.

Where my hesitation comes in is that I often use search not just to retrieve an answer, but to discover what I didn't know to ask. I may be looking for obscure sources, minority viewpoints, independent experts, or details that don't fit neatly into a single synthesized response. In those situations, the clutter is frustrating, but the process of exploration still matters to me.

So perhaps we're describing two different use cases rather than contradicting one another. For common questions, AI may indeed "cut to the chase." For exploratory research into the long tail of human knowledge, I still miss the older style of search that more readily surfaced unexpected and independent sources.

I've encountered summaries that were almost unintentionally funny, they were so wrong.

They become frustrating when one of my students uses them to answer homework questions, and they're shocked - shocked, I tell you! - when they lose points because an AI-generated answer is inaccurate.

to find what you're looking for, which is stated as factual, but might very well be slanted at best or totally wrong. The logical answer to that is simply to click the Google search hits themselves.

Plus....it's not necessary to search the web. The point of the OP. It's just a multi-billion dollar boondoggle of epic proportions. A solution desperately seeking a problem to solve that doesn't exist. No worries if it weren't for the fact that the boondoggle is sucking up vast amounts of resources to give you incorrect info. Enjoy higher electric bills for the "convenience" that isn't.

however "artfully crafted" the prompt is. And sending corrections into Google (per their request) doesn't do anything - I know, I've done it over and over again (researcher).

Google and their paying advertisers want. I agree that the results are often good ones. But it gives Google too much influence over the results they want to influence. AI is only as good or as accurate as the programming that produced it. I believe the court is ruling in favor of objective results that are not influenced by Google's agenda. Purely objective results may not be achievable, but AI makes it possible to put a pretty heavy thumb on the scale.

We can see this happening by comparing, for instance, the Chief Twit's Grok to other AI engines. He even proclaims his violations of the Overton window as a "feature, not a bug."

I'm not sure how to enforce a reasonable "Overton window" in the age of the internet, regardless of the search engine used. As you note Google and all the others use their influence over search results, AI or otherwise, for their own ends. I tend to trust organizations that are in it for the money a bit more, as opposed to those explicitly aiming for political influence. But my searches are typically for basic facts, not opinions based on the facts. The predatory habits of red shoulder hawks is probably not a big target for political manipulation.

One irony about the court ruling is that we don't actually know what the court ruling was. As explained in the article, the authors used Google to translate the ruling from the German to English! So that nefarious AI was in the loop! Who knows what malicious misrepresentation it introduced!

that accompany the results rather useful. But even basic scientific facts related to global warming or alternative energy could be targets of political influence. always best to triangulate the truth using two or more sources that are generally trustworthy.

That's why I quit them.

They could have been working to improve the efficiency of their search engine as measured by relevance of search results and energy use of the processes; instead they dumped this monstrosity on us.

Of course we are not Google's customers. Google's actual customers are corporations (including Google itself) that want to sell us their shit. An enlightened search engine would ignore all the shitty web sites and that would be bad for Google's business.

AI "search" is just another notch ratcheted up in the enshittification of search technology that Google and others have found profitable due to the increased ad revenue it generates.

When I joined DU in 2009, I was a tech optimist, believing that websites like this and social networks like Facebook would enable a true marketplace of ideas. Barack Obama is said to be the first social media president.

Unfortunately in retrospect I should have been less naive. I remember seeing plenty of ill-informed blogs, personal websites, and chainmails in the 2000s and early 2010s. But those were the kid's meal compared to the sewage on modern day Facebook or Twitter.

As was written in a blog in 2016:

I switched to DuckDuckGo and haven't looked back. No AI bullshit shoved down my throat, no tracking, far fewer ads, and more!

has it disabled by default (noai.duckduckgo.com)

They now have a No AI extension you can download there as well. It makes an admirable effort to block AI-generated images too, though I've seen quite a few slip through the cracks. But it's a hell of a lot more than what most search engines are doing.

That is until they enshitify. Let's hope not.

because Trump wants them? That NETCHOICE running them is just all the tech bros. They trash Bernie and AOC but most MAGA people don't want the GD things either.

1. They don't want to pay any taxes or startup costs (they'll amortize everything)
2. They want first dibs on local water supplies in massive amounts and pay almost nothing for it
3. They want first dibs on all the municipal power sources (electric, solar, etc.) and pay as little as possible for them
4. They are willing to bribe any local politicians to get the above 3 items.

Their buildings are ugly warehouses, usually set out on the middle of nowhere. The people they hire will be paid low wages with no benefits and they'll be given little job security or special training. When the data center starts to lose money, they'll just shut down and leave their hulking mess for others to clean up.

will be 600 MILLION gallons per year, and I'll bet they've lied about that number. That works out to 1.64 million gallons per day 365 days per year. The want to tap the Ogallala aquifer for the water which even now farmers have realized in the last few years how they fucked up doing massive amounts of pumping and lowering that aquifer to dangerously low levels which took millions of years to create.

Also, there's lots of flat open space (which they need) and the acreage costs are lower than city properties. But mainly it's the failure of zoning commissions to keep them from coming in. When it's easy to bribe the local authorities - i.e. the people who allow the building permits and such - it's clear why they're doing this.

And yes, in many cases the farmers are screwed.

to ban AI from my searches. In my opinion AI is completely untrustworthy. Often it contradicts itself multiple times in the same search.
Maybe this will force Google to let users choose if they want to be subjected to stupidity called AI.
.

a lot has to do with how precisely one composes their query. I love the sort of tool it has become for things like trying to remember the name of a TV series. I’m not concerned with incorrect info. I’ll know the name when I see it.

I google the name of a local woman, and it said she'd been in jail for a DUI.
An AI summary of her life followed. It got very confusing, till I realized that they'd combined the
bio information of 2 different women who shared a name. And the only photo
was of the woman who had NOT gotten a DUI.

via AI as fact. It’s clearly incapable of that kind of discernment.

without AI.

I know what it takes to find information, and ChatGpt has been very valuable to me in the last year. I also have good judgement about real vs fake. Many people don’t.

... are the AI techbros trying to defend and justify stealing the intellectual property of others – without recompense or even crediting them – to train their AI trollbots. Some of my work has apparently been used to such ends, but if I see a penny on the dollar when the lawyers are done gutting any punitive damages awarded, I will be truly amazed.

actor and filmmaker they care about has probably had work they put a lot of effort into stolen to train AI. Stolen by tech bros who really don't care about any of that work and knowledge and culture they've expropriated, but simply want to profit from AI tools intended to create imitations of it when used by people who don't have that knowledge and talent.

A lot of people still aren't aware that the entire generative AI industry is built on theft.

And unfortunately there are apparently a lot of people who find the pretense of knowledge and talent and skills they don't have so tempting they'll use AI even when they're aware of the theft, and they try to find excuses for it, and try to show off their fraudulent AI-enabled "accomplishments."

People don't need AI to write or to be creative in all the many ways humans can be creative. But AI peddlers want them to believe they do.

It's such destructive tech.

Like the saying goes, why buy the cow when you can get the milk for free? 25 years ago there was the controversy over illegal free music downloads on Napster. And today we see two lines of thought, one that news websites shouldn't put up paywalls (hello, how will the reporters get paid for their work?), another that AI devalues knowledge work.

The ones responsible for that theft belong in prison. Their companies should be sued out of existence, and all the illegally trained AI models should be destroyed.

The tech oligarchs own Trump and Republicans

Now, imagine AI "learning" from Trump, Republicans, MAGAts, FOX, and other rightwing nuts and fascist sites.
Further, imagine that AI "learns" from these sites. What does AI actually learn and have the "controllers" "learned" AI how to tell falsehoods?
If AI is getting info from sites that lie, it's not too far fetched to understand that AI CAN, but will it be allowed, to intentionally. ..LIE? And if it does, who is responsible?
Since it seems most of the AI owners are right or slanted right, it may follow that they will allow possibly partisan fibs. Should we mention CBS or Fox?
Now the media and the Right have another tool to misinform an already gullible public.
SO...
Since the majority of Americans already don't know if a politician or corporation is lying...
HOW DO WE TELL if AI lies?

Is your ice cream flavored with real vanilla or 4-Hydroxy-3-methoxybenzaldehyde made from petrochemicals?

Like, if their lips are moving....


It's much harder to find one telling the truth about--- just about anything!

If we had a nation of well-informed internet users, AI could be an asset when used purposefully. However, the reality is we are a nation of post-literate, post-intellectual, eternal juveniles who expect to be told what to think.

I'm so sick of the answers I get from Google's AI. I end up screaming at it.

For simple things, it usually gives me what I need. I wouldn't depend on it for important research.

And found they are behind. Hours later it will give me the correct answer.

It's got yo wait for posters and news organizers to post reports before it can give the correct info I guess.

I'm very impressed with the clarity of this ruling. Most judges don't have the world knowledge to figure these things out. However, I would like to add an element that is floating out there.

Most people don't want to say it out loud, but AI is not controllable, period, full stop. In computer science terms, AI is not deterministic, which is to say that it is impossible to know exactly how these very large models will react. And perhaps they share that with organic brains. We can't be sure a dog that looks docile will not bite under some set of circumstances.

For the past 25 years, we have talked about "algorithms." An algorithm is essentially a program: If this, then do that. Algorithms can be very complex, but they are deterministic. It is always possible to find the point in the algorithm where things go bad. There is no "program" with AI. The neural networks are "trained", and with proper training, they hopefully develop tendencies that are more good than evil. But we can see vast differences between Grok (which was largely trained on the Twitter cesspool) and Claude, for example. Grok tends toward the evil, antisocial side, where Claude may be more humane and altruistic by nature.

But the main point is that, when there is a bad result, you can't "fix the algorithm". All you can do is try to throw more training data that reduces the odds of that bad behavior.

and it's amazing how many AI fans never think this through.

Let's say you train a neural network to differentiate between a dog and a mouse. You give it 10,000 pictures of dogs at eh Westminster show, dogs at a dog park, dogs playing frisbee, etc. And you do likewise with 10,000 pictures of mice. Then you show it a picture of a skunk. It will identify the new picture was either a dog or a mouse, and be 100% certain in its decision.

I heard of another case recently where a woman got a Ring alert that there was a bear at her front door. It was actually the woman herself. I haven't been able to find that video, so it might be apocryphal, but AI is messed up enough for that to be true.

Then there's this:

We've done it for 30+ years and somehow found what we're looking for.

At lest with a regular search one has to go to a site to look at the information.

This is an in depth review of the AI chatbot phenomena in two parts. I found it informative and fairly disturbing. It also made clear to me one basic problem with AI overall... mostly around how AI has no real way to judge the validity of the information it takes in and regurgitates. People still have to be the arbiters of the content produced and AI isn't being sold in that way.