Gartner Predicts Search Engine Volume Will Drop 25% by 2026, Due To AI Chatbots and Other Virtual Agents 93
Gartner: By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents, according to Gartner. "Organic and paid search are vital channels for tech marketers seeking to reach awareness and demand generation goals," said Alan Antin, Vice President Analyst at Gartner. "Generative AI (GenAI) solutions are becoming substitute answer engines, replacing user queries that previously may have been executed in traditional search engines. This will force companies to rethink their marketing channels strategy as GenAI becomes more embedded across all aspects of the enterprise."
With GenAI driving down the cost of producing content, there is an impact around activities including keyword strategy and website domain authority scoring. Search engine algorithms will further value the quality of content to offset the sheer amount of AI-generated content, as content utility and quality still reigns supreme for success in organic search results. There will also be a greater emphasis placed on watermarking and other means to authenticate high-value content. Government regulations across the globe are already holding companies accountable as they begin to require the identification of marketing content assets that AI creates. This will likely play a role in how search engines will display such digital content.
With GenAI driving down the cost of producing content, there is an impact around activities including keyword strategy and website domain authority scoring. Search engine algorithms will further value the quality of content to offset the sheer amount of AI-generated content, as content utility and quality still reigns supreme for success in organic search results. There will also be a greater emphasis placed on watermarking and other means to authenticate high-value content. Government regulations across the globe are already holding companies accountable as they begin to require the identification of marketing content assets that AI creates. This will likely play a role in how search engines will display such digital content.
I predict 75% chance that Gartner is wrong (Score:5, Insightful)
I mean literally about anything. I have no idea why people pay them to prognosticate on the future.
Re: (Score:2)
> strong words for someone posting no data...
Are you even being serious? A document from October 2020 certainly has the overall perspective and hindsight of what happened.
You just googled and pasted in the first thing that matched your confirmation bias without any thought.
Published online 2020 Oct 19. doi: 10.15252/emmm.202013171
PMCID: PMC7645374
PMID: 33073919
COVID19 pandemicrelated lockdown: response time is more important than its strictness
Re: (Score:2)
"Sweden's no-lockdown COVID strategy was broadly correct, commission suggests"
https://www.cbc.ca/news/world/... [www.cbc.ca]
Initially I was searching for negative media articles about how Sweden stopped lockdowns early in COVID, but this will do.
Keep in mind CBC funded by a leftist Canadian government, but they do reference the original study.
Re: (Score:1)
> See now that wasn't so hard was it?
The only thing scientifically proven hard... is my dick... from watching you scramble.
All the sensationalist media bullshit about how Sweden ended lockdowns and it was going to be a disaster, blah blah blah...
It's the same garbage every year. The media is always working over time to push an agenda that anyone can see isn't true. Which is why it's a lot of work.
You can be a pretend pragmatist and a condescending twat but the lockdowns were bullshit. Even king fuckfa
Re: (Score:1)
Strange to complain about "uniqueness" of "control group nation" is that it in fact served as a "control group nation" because it's a liberal Western democracy within the EU, functioning under a sufficiently similar legislative, cultural and economic system to serve as control for the global test we did. The main difference was that they didn't lock down, whereas the rest of us did. So we have a "hypothesis - lockdowns and masks have a meaningful impact, test subjects: most of Western democracies, control g
Re: (Score:2)
Strange to complain about "uniqueness" of "control group nation" is that it in fact served as a "control group nation" because it's a liberal Western democracy within the EU, functioning under a sufficiently similar legislative, cultural and economic system to serve as control for the global test we did.
Never mind that they have an incredibly low population and even more importantly, a super low population density https://database.earth/populat... [database.earth] which makes it not ideal at all for use as a control in the context of studying the spread of really any disease but particularly an airborne one.
Outside of that of that it makes a perfect control though.
Re:I predict 75% chance that Gartner is wrong (Score:5, Informative)
Today, science has proven beyond reasonable doubt that they didn't have a statistically meaningful impact on spread,
Nope, wrong. Stop listening to Faux News. What you think you're citing has never been peer reviewed [factcheck.org].
But one working paper posted online in January — and not peer-reviewed — has gotten a lot of attention in conservative circles for its conclusion that “lockdowns have had little to no effect on COVID-19 mortality.” The paper, which is an analysis of other studies, has been touted as a “Johns Hopkins University study,” but it’s not a product of the university’s Bloomberg School of Public Health, whose vice dean — among other public health experts — has criticized the paper.
“The working paper is not a peer-reviewed scientific study,” Dr. Joshua Sharfstein, vice dean of the Johns Hopkins Bloomberg School of Public Health, said in a Feb. 8 statement sent to us in an email. “To reach their conclusion that ‘lockdowns’ had a small effect on mortality, the authors redefined the term ‘lockdown’ and disregarded many peer-reviewed studies. The working paper did not include new data, and serious questions have already been raised about its methodology.”
Sharfstein said that early on “when so little was known about COVID-19, stay-at-home policies kept the virus from infecting people and saved many lives. Thankfully, these policies are no longer needed, as a result of vaccines, masks, testing, and other tools that protect against life-threatening COVID-19 infections.”
Further, and for more insight [snopes.com] on this fake study:
While many media reports on this working paper noted that "lockdowns only reduced COVID deaths by 0.2 per cent," this may give readers a false impression of what this working paper actually found. The common definition of a "lockdown" is a mandatory state of isolation. In terms of the pandemic, many people would take "lockdown" to refer to a requirement for people to stay inside their homes (not attending public events, school, going to restaurants, or leaving for any other non-essential reason.)
This paper, however, defines a lockdown as "the imposition of at least one compulsory, non-pharmaceutical intervention (NPI)." This means that this study interprets a mask-wearing requirement as a "lockdown," even if that requirement did not prevent a person from visiting public spaces.
Still further, study after study has shown that lockdowns, social distancing, and mask wearing unequivocally reduced infections and deaths [newscientist.com].
If lockdowns didn't work, why is it every single time people gathered [livescience.com] in groups [vice.com] some of them contracted covid [washingtonpost.com] and some died [go.com] from it? It's almost as if keeping people separated would make more sense owing to the facts.
Re: (Score:1)
lol great response to a well sourced and though out comment.
just snark your way out of thinking about your priors at all. best not to challenge your brain too much.
Re: (Score:2)
Re: (Score:1, Troll)
>If lockdowns didn't work, why is it every single time people gathered [livescience.com] in groups [vice.com] some of them contracted covid [washingtonpost.com] and some died [go.com] from it?
Wait, you don't remember Summer of Love, when thousands routinely gathered to scream next to one another for hours and no one contracted covid?
The thing you linked as debunk btw? That was a government panel that needed to exonerate UK government because it was one of the most draconian ones, looking at a rather miff
Re: (Score:2)
Have you forgotten how many people refused to wear masks despite them being required? A true lockdown, where everything is closed so you have nowhere TO GO AND MEET PEOPLE in the first place is vastly different from an honor-system of "Please wear a mask so you don't accidentally kill people" that certain people took as literally Nazi Germany levels of oppression.
Re: (Score:2)
Re: (Score:2)
Nope, wrong. Stop listening to Faux News. What you think you're citing has never been peer reviewed.
Is it your assertion something is wrong and can be summarily dismissed because it has not been peer reviewed? Shall the huge body of CDC's pandemic era studies none having undergone peer review be considered wrong and subject to summary dismissal as well?
This paper, however, defines a lockdown as "the imposition of at least one compulsory, non-pharmaceutical intervention (NPI)." This means that this study interprets a mask-wearing requirement as a "lockdown," even if that requirement did not prevent a person from visiting public spaces.
Policy questions are the most salient as they directly inform real world impacts of policy. If people are going to organically tend to pursuit NPIs on their own initiative in response to perceived risks anyway then papers contrasting retaliative outcomes
Re:I predict 75% chance that Gartner is wrong (Score:4, Insightful)
I'm curious, how many times on Slashdot have you been shown to be wrong with your covid claims as is the case now with the many replies (complete with backing citations) below your post here?
Re: (Score:2)
For starters, you're conflating "lock downs did nothing in the context of COVID" with "lock downs caused other problems non COVID problems" which are two separate issues. Furthermore, you use phrases like "well documented" but can't be bothered to provide citation (which should be easy if these things are well documented) as those responding to you have done and it can't be due to the time it would take as you have no problem taking the time to write lengthy posts. It's also interesting that you're taking t
Re:I predict 75% chance that Gartner is wrong (Score:4, Insightful)
Re: (Score:2)
75% of the prediction might be wrong, but for Google they can't just say 'Gartner has 75% wrong on the prediction of the complete dismissal of our company.' If a doctor is wrong 75% of the time, but has 25% that it's right, your child would die, would you take the chance?
Re: (Score:2)
I think you are saying Google has to respond to something like this.
I suggest they are, by diversifying their business and investing in LLMs. It may not be enough - in fact, it probably isn't enough if Gartner is right, but they're at least doing something.
Re: (Score:2)
Google has responded quite a while ago, long before Gartner said this. They have their own LLM, Gartner's statements actually seem to lean heavily on Google's own assessments. I.e. Gartner isn't actually pulling this one out of their ass, but instead more or less repeating what other players in the search field including Google and Microsoft reps have been saying for a while.
Google's LLM is Bard which is generally considered to be much more prone to hallucinations than ChatGPT and generally less helpful. Th
Re: (Score:3)
Google has responded quite a while ago, long before Gartner said this. They have their own LLM, Gartner's statements actually seem to lean heavily on Google's own assessments.
It's also a comment on the continuing enshittification of google's own search engine. Google search hallucinates results all the time, based on what it "thinks" I want, instead of what I typed into the fucking search box. OK, these may not be hallucinations but a sale of my search to the highest advertising bidder, but it's still shit, and less useful than just giving the user the results that they searched for.
Re: (Score:1)
They have recently announced that they will at least try to clean out some of the "spam and AI results" from your search results.
I won't hold my breath considering the recent piratewires story on how google ended up in the mess with Gemini.
Re: (Score:1)
But I know that Google will make the results of their LLM useless for search as they have for their traditional search engine.
Google results have been increasingly untrustworthy over the last 12 years. The first page is frequently completely useless.
So I'm hoping for an opensource LLM that isn't corrupted by greed to the point of being worthless.
Re: (Score:1)
Meta's LLM model is open source.
Re: (Score:2)
Strictly speaking as prognosticators go, 25% accuracy isn't all that bad.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
> why people pay them to prognosticate on the future.
Did you ever notice these types of folks sometimes contradict themselves?
1. Decide what you want to do.
2. Pay them for access to a study that agrees.
3. Execute your plan.
4. If it fails, blame your reliance on the #1 forecaster.
4a. If it succeeds take credit.
That's totally worth $1200 of company money.
Re: (Score:2)
I mean literally about anything. I have no idea why people pay them to prognosticate on the future.
The numbers are a bit random (could be 5%, could be 75%), but they're saying the same thing I've been saying for months, that LLMs will meaningfully reduce search engine traffic.
Of all the tech companies freaking out and making layoffs I think Google is the one that should really be concerned. LLMs are the one thing that can really bite into their search dominance.
Here comes AI Product Placement (Score:3, Insightful)
Businesses will need to train AI to insert product placement into their answers to maintain ad revenue.
You: What is a detergent?
ChatGPT: A detergent is a substance or compound used for cleaning. It's typically added to water to help remove dirt, grease, stains, and other impurities from surfaces. Detergents such as Dawn work by reducing the surface tension of water, allowing it to more effectively penetrate and lift away grime. They often contain surfactants, which are molecules that have both hydrophilic (water-attracting) and hydrophobic (water-repelling) properties, enabling them to surround and solubilize oily or greasy substances in water. Detergents are commonly used in household cleaning products such as Tide, Palmolive, and Cascade dishwasher tabs.
Re:Here comes AI Product Placement (Score:5, Funny)
Exactly. My example was:
'What time is it in Paris?"
"It is 3pm in Paris. Here are current airline ticket and hotel prices, I've already booked your flight and added it to your calendar. There will be a rental car waiting for you on arrival. This service brought to you by Orbitz!"
Re: (Score:2)
welcome to paris Arkansas
For some things, perhaps (Score:4, Insightful)
The trouble with search engines now is that the data is getting old. Say you're trying to find a solution to a programming question for a current API version. What you find is stuff that's several years out of date that doesn't actually work now. AI often gives you similar responses. The result looks legit but when you try the code, it's wrong. And if you tell the AI that it's wrong, it gives you a condescending response "You're right, that blah blah function call doesn't exist. Here's another bullsh*t response."
AI is trained on out-of-date stuff. It doesn't know that it's out of date.
Re: (Score:2)
I have been trying for years to learn Python. There are many tutorials. I have yet to get through lesson 1 without getting errors.
Re: (Score:3)
That has been a huge issue for years. The worst is when they don't put dates anywhere so I can't tell how old it is. I'm constantly hitting outdated parameter and config options.
Then there are the people who ask a question then go "I solved it!" without posting the solution. I hope there is a place in purgatory for them at least.
Re: (Score:2)
Re: (Score:2)
Then there are the people who ask a question then go "I solved it!" without posting the solution. I hope there is a place in purgatory for them at least.
Yes, with a sign saying “I managed to get out by”.
Re: (Score:2)
And yet when you need to find something that happened ten or twenty years ago you have to wade through all the remotely similar things that happened in reverse chronological order until you get to what you were thinking of.
Re: (Score:2)
The trouble with search engines now is that the data is getting old. Say you're trying to find a solution to a programming question for a current API version. What you find is stuff that's several years out of date that doesn't actually work now. AI often gives you similar responses. The result looks legit but when you try the code, it's wrong. And if you tell the AI that it's wrong, it gives you a condescending response "You're right, that blah blah function call doesn't exist. Here's another bullsh*t response."
AI is trained on out-of-date stuff. It doesn't know that it's out of date.
Given the issues you’ve raised regarding search engines alone, I’d say we humans have been trained to accept outdated bullshit responses.
AI “learned” from that. Marketing called that “Intelligence”. I’d blame them, but they certainly know how well we’re trained to accept outdated bullshit..
I can see it (Score:4, Insightful)
As I've been searching lately, AI has been cutting in to my searches. When a search engine decides to throw AI at what I've asked for it, I search on a different search engine; I'm looking for matching WEB PAGES, not some computer's interpretation of what I've asked for.
I'm old-fashioned. I search for content, not computer guesses at content that might contain what I asked for. Today, I looked for information on a store chain having financial issues. Google decided I was looking for "closest to Chicago", for the Los Angeles-based chain. duckduckgo at least didn't prioritize Chicago store locations over the recently-reported corporate problems.
That sort of thing will cut my searches by at least 25% going forward, as they increase the bad data. I'm getting too old to deal with the [censored] "help".
Re: (Score:3)
Agreed. What I want in a search engine is the ability to search for phrases, have "must includes", "must not includes", date ranges, and domain filters. After that it's nice to have them put effort into down ranking SEO crap.
Google is failing on all fronts. The most carefully constructed query can ignore phrases or exclusions if the algorithm feels like it, and the battle against SEO spam is more or less lost. Then you throw in the ads and remove the censored stuff - don't try looking directly for anyth
Re:I can see it (Score:5, Insightful)
Plus with AI I *HATE* playing "Guess the right wording". I feel like I'm playing a text adventure from the 1980s. "Pull the knob"? "Push the knob?" "Push the knob on the wall"? wtf!?
Re: (Score:2)
Plus with AI I *HATE* playing "Guess the right wording". I feel like I'm playing a text adventure from the 1980s. "Pull the knob"? "Push the knob?" "Push the knob on the wall"? wtf!?
Im not sure if I should laugh at the 80s reference and my childhood memories, or cry on behalf of Progress realizing we call that shit “Intelligence” now.
Re: (Score:2)
I can see the opposite. Last searches where AI got me better results then keyword search on google:
Re: (Score:2)
What's really pissing me off with search is no longer having pages of results and having the abominable "scroll to see more results" UI crap.
Previously I would do a search then start looking at about page 4 as that would get you past most of the shitty promoted, irrelevant, results on the first few pages. Nowadays you simply can't get to the actual results you want as you lose the will to live before you can scroll enough results (by which time the browser tab has started creaking too).
"AI" is just the ret
Gartner, McKinsey, et al (Score:4, Insightful)
You have to love these consulting houses. They sell super-expensive reports that either contain banal generalities, or else they make predictions of the obvious. But the language is pretty and there are lots of buzzwords. CxOs then select what they need from the reports to justify the dumb decisions they want to make anyway.
Cynical here, because our assigned desks were just rationalized away based on a "future of work" report from some McKinsey. I've hear that it looks "much neater" now. Well, duh, we've gone WFH-to-the-max and most of the desks have literally never been used. Probably not what management intended, but what they got.
Re: (Score:2)
AI cannot replace search until hallucination ends (Score:2)
The issue right now is that if you ask AI anything, what it gives you might be made up or wrong, so often you have to check some parts in a search engine anyway to get a definitive answer.
A recent attempt at asking AI how to do something led it to giving me three frameworks that were wholly fictional to use - which I only found out after a lot of googling to try and find where the frameworks AI had mentioned were.
AI can help amplify other things, not saying it can't be useful, but I'm not sure it will be re
Only if AI gets MUCH better (Score:2)
The quality of the results I get from searching vs AI are worlds apart. I sometimes go to services like ChatGPT looking for specific technical information and the results I get are usually vague at best "To create a certificate first create your certificate..."
Re: (Score:2)
"Traditional search" is typing your query into a text box and getting back whatever Google's algorithm thinks matches. That could very well be replaced by an interface where you type in a more natural language query, you still get web page hits, and they're maybe selected a little more intelligently. That last thing may or may not be good for you.
Chatting with a chatbot is an engaging UI, but it's not a very good one for getting accurate results, and it's certainly not the only way of using AI to seach.
Re: (Score:2)
"Traditional search" is typing your query into a text box and getting back whatever Google's algorithm thinks matches. That could very well be replaced by an interface where you type in a more natural language query, you still get web page hits, and they're maybe selected a little more intelligently. That last thing may or may not be good for you.
Considering the enshitification of google search results through many mechanisms including AI based relevancy botting and ad placements it would need to be pretty good to just undo the last 4 years of slide. My guess is it’s just going to lubricate the slope instead of enable any sort of traction toward content we actually want.
Re: (Score:2)
So, a lot of this depends on what happens in the courts and the marketplace - a lot of money wi
Re: (Score:2)
Don't hold your breath. Google doesn't think their search is worse now than it was before. Quite the opposite. It's got many more sponsored links now.
What will they ever do? (Score:2)
Re: (Score:3)
Won't someone think of the poor marketers?
I do. Often. Not particularly fondly.
Desperation (Score:2)
If this does happen, I predict it will be because search engine results will get much worse, prompting people to desperately used an AI Chatbot as a last resort. I already have to go directly search Stackoverflow or even Youtube for what I want because Google's results are next to useless.
It's actually just the evolution of search (Score:2)
Re: (Score:3)
I do wish ChatGPT and others would footnote their answers and have a bibliography of sources used in the answer.
They literally don't know. It's not how language models work.
You could have them go through a process to try to figure out if information in their answer is a static factual claim or dynamically generated data and then search a copy of the training data for matches. But then you're just using another search engine.
Re: (Score:2)
They literally don't know. It's not how language models work.
That doesn't stop Bing. And the results are exactly what you would expect from an LLM: they definitely look like references from a distance.
Re: (Score:2)
I do wish ChatGPT and others would footnote their answers and have a bibliography of sources used in the answer.
Bing/Copilot does this. I think Google does as well.
It's really money in motion (Score:2)
Money in Motion 101 - Insist, push for, tsunami fear spreading - It's the usual playbook to get you to spend money, companies to buy things, technology replacement, new government regulations, push voting blocks, ...
Gartner, Blackrock, Vanguard, Fidelity, Wall Street, Lobbyists, Bureacrats, Election Candidates, NTO, non-profits need money to be in motion to survive and justify their jobs.
If there's a stable IT and computing environment for 20 years, with no big changes needed, Gartner has no strategy consul
Predictions are hard... (Score:3)
...especially about the future
I predict that search will evolve
I hope it gets better, and makes it easier to actually find what I want
I fear that it will get worse, with a flood of scams and ads
Whatever happens, I suspect there will be a lot of unexpected surprises
I predict AI will replace Gartner (Score:4, Funny)
I predict AI will replace Gartner. And unlike most situations in which AI has enshittified things, nobody will notice in this case.
Re: (Score:3)
You don't think there'd be an appreciation for the increase in quality?
Re: (Score:2)
I predict AI will replace Gartner. And unlike most situations in which AI has enshittified things, nobody will notice in this case.
How could anyone notice? They use the same principals to come up with the same responses. There is not an ounce of original thought in either.
That depends (Score:3)
The one thing that will determine this is what happens first: AI finally producing reliable results or people dumping AI because it doesn't produce reliable results.
Re: (Score:2)
The results are already reliable enough for most people, for what they search for.
"Who's that one singer who sings baby got back"
"what's the best color to pair with green"
"who's the president"
Just add in sportball stats and they'll never use a search engine again. Instant curated reality...
Re: (Score:2)
The one thing that will determine this is what happens first: AI finally producing reliable results or people dumping AI because it doesn't produce reliable results.
You don't understand what is going on here. Even AGI will not produce 'realiable results'. This current AI absolutely CAN be useful, just not the way people want it to be useful. What that means is you can ask it for patterns and it can give them to you. Is that a useful trick? Yes. Will it give you reliable 'answers'? LOL, no.
Already (Score:2)
I'd be really surprised if that's not true already.
Unless someone's trying to find very precise original sources or technical data, train/plane/bus schedules, or news, or something to buy, search engines have become almost completely useless due to (primarily) ad placement and the enshitification of the Internet due to corporate sites and social media having taken over the majority of information silos. You don't have many free hosting sites anymore, so people just use social media groups to keep track of t
Re: (Score:2)
>The top 100 questions are similar - a lot of stupid "what time is it", "when is mother's day", and "how many ounces in a cup" type questions
I often use Google search as a calculator, because it's actually faster than opening the calculator app... especially if you need to include unit conversions.
Already pretty useless. (Score:2)
Obvious prediction is obvious. (Score:2)
Not only do I use Google less, but I'm even more grateful to not have to use online forums.
With AI you can get a sane, coherent answer to the question you actually asked.
Or ... (Score:1)