They call it AI pollution. When content created by large-scale language models (LLMs) permeates across a domain like an oil spill, two outcomes will likely occur.
First, it will become more difficult to tell truth from fabrication, encouraging the charging and lowering of the quality of information provided by the Internet.
The second result is a clearing of the sky to some extent in the form of a walled garden of protected content opening up to users in premium, freemium, and other subscription models.
News website paywalls and YouTube Premium are early examples. But across vast swathes of the internet, pollution flows freely.
“Welcome to the world’s first fully AI-generated website!” Read the disclaimer on The Enlightened Mindset’s landing page.
The website has no ownership details or contact information, but lists hundreds of articles by “contributors” with names like Happy Sharer and Crazy Lee. The title is “Is Hagrid played by a robot?” Exploring the Possibilities,” “How to Be More Feminine: Speak Softly, Dress Modestly, and Gently Carry,” “How to Be in Danger: Learn Self-Defense, Carry a Concealed Weapon, and Network.” .
The story contains fictitious quotes attributed to real people. Hagrid’s story includes a statement by Dr. Reuben Binns, Professor of Artificial Intelligence at the University of Cambridge, about the drawbacks of having a robot play Hagrid. In the real world, Binns is an associate professor of human-centered computing at the University of Oxford.
A growing number of platforms use such approaches, but most do not provide a disclaimer.
lying down and waiting
Earlier this month, NewsGuard, which tracks misinformation by assigning trust scores and ratings to online news sources, announced that more than 700 AI-generated news and information websites operate with little or no human oversight. We identified that there was no disclosure whatsoever about the nature of the content. .
New websites are steadily being added to the list, but that valiant ongoing effort is likely just scratching the surface. NewGuard launched his AI Tracking Center in May and his team of content analysts worked on identifying and flagging such content, with 49 content in his first month. of such websites have been identified.
These platforms publish fabricated reports about political leaders and spread false claims about the deaths of celebrities. In November, one such site, GlobalVillageSpace.com, announced that Israeli Prime Minister Benjamin Netanyahu’s psychiatrist had died by suicide (the article was later tagged as satire).
In April, news platforms around the world reported the death of a Canadian actor named Saint Von Colucci. He reportedly underwent multiple cosmetic surgeries to resemble Jimin of K-pop band BTS, and died from complications. Colucci didn’t exist. All his images were generated by AI.
Al Jazeera called it “the first known case of AI being used to deceive a news organization en masse, marking the beginning of a new era of computer-generated fake news.”
Technology industry analyst Kashyap Kompela said that as the number of such websites proliferates, trusted sources such as trusted media brands need digital forensics teams trained to combat the misinformation they generate. He says he will have to organize the team in a hurry. “AI-powered fact-checking tools will also be built into future browsers. It’s a metaphor for the battle between good and evil that you often see in science fiction, with AI on both sides of the battle.”
Zombie websites, ghost products
Watch those domain names carefully. The old ones are gradually being revived. The Hairpin, a women’s website that was shut down in 2018, is also one of her more recently revived websites filled with AI-generated content. Headlines here include absurd things like “All Celebrities Have Small Real Teeth Beneath Their Big Fake Teeth.”
This version of Hairpin is run by Serbian DJ-turned-entrepreneur Nebojša Vojinović Vujo, who has reclaimed many abandoned news domains and flooded them with LLM-generated clickbait.
“In many cases, the revenue model for these websites is programmatic advertising, with the ad tech industry serving ads regardless of the nature or quality of the website,” NewsGuard said in a statement. “Unless brands take steps to weed out untrustworthy sites, their ads will continue to appear on these sites, creating a financial incentive to produce them at scale.”
Meanwhile, Amazon’s unnamed generative AI feature, launched in September to help sellers list and describe products, began to flop in January. The user ended up inputting the AI-generated text into the program, but the Amazon program was unable to remove it. As a result, products ranging from lawn chairs to sets of drawers appeared in Amazon search results with names that were essentially AI error messages. One product name says, “We’re sorry, but we can’t fulfill your request because it violates OpenAI’s usage policy.”
The error message was an initial bug and the listing has been removed. But they provide an indication of what the online world will become as we switch from human-generated clickbait to content generated for hits and advertising, without human involvement or oversight.
“Internet encitization has become so widespread that the term was chosen as the American Dialect Association’s Word of the Year in 2023,” Kompella points out. (The term was coined by internet activist Cory Doctorow in November 2022.)
Kompella added that this decline will only intensify as AI programs learn from AI content.
Google killer
The business of online advertising is also facing major changes, with new types of conflict emerging as search engines move from a listing approach that drives traffic to websites to a summary approach that drives the engines’ own AI programs. Probability is high.
For example, Google started monetizing by showing ads within summary search results in January, but other platforms are already doing better with summaries. Where search giants only respond with summaries, Perplexity AI responds to every query with detailed notes that include links to multiple sources. Consensus is the same for scientific papers. Arc Search’s Browse for Me feature in privacy-focused Arc Browser also blocks ads and trackers while displaying information and relevant links on customized web pages.
“They’re like your personal research assistants, streamlining and personalizing every answer,” Kompella says. “These are game-changers if you are looking for a response based on detailed qualitative analysis.”
These search services are costly for content creators, and we expect to soon see copyright and legal disputes over what constitutes fair use by such engines, but they are also costly for users. Possibly, Kompela says.
“The future Internet is likely to be a walled garden space,” he added. “As auto-generated content increases, hand-crafted content will become even more valuable, and premium material will become increasingly available and perhaps only viewable to paid subscribers.”
By the numbers: Bigger. Better?Take a look at our online lives
According to DataReportal.com, the average user spends about 40% of their waking hours on the Internet. Every minute, 6.3 million searches are performed on Google worldwide, and 6,944 prompts are entered into his ChatGPT. Take a look at other dramatic findings from his 2023 report from reference library DataReportal.com, research agency Statista, and cloud solutions company Domo.
*Between 2013 and 2023, the number of internet users worldwide nearly doubled, and the number of user profiles increased from 2.7 billion to 5.3 billion.
* India had 751.5 million internet users as of January, with an internet penetration rate of 52.4%.
* Some of the most online countries in the world include Norway and Denmark, with 99% internet penetration. Meanwhile, millions of people remain offline in countries such as North Korea, where internet penetration is reported to be among the lowest in the world at less than 10%.
* The average user spends 6.5 hours actively online each day. This equates to about 40% of your waking hours. His 2023 study from market research firm GlobalWebIndex, cited by DataReportal, found that leisure/entertainment accounts account for most of the time spent online, searching for information, staying in touch with loved ones, and spending more time online.
* The typical social media user currently spends 2 hours and 23 minutes per day on these platforms. This equates to more than a third of his total online time.
* The number of Google searches performed around the world is rapidly increasing. In 1999 there were 500,000 queries per day, in 2013 there were 2 million queries per minute, in 2022 there were 5.9 million queries per minute, and in 2023 there were 6.3 million queries per minute.
* ChatGPT currently holds the record for the fastest growing consumer application of all time. In January 2023, two months after launch, it had approximately 13 million unique visitors per day. That month, that number increased to 100 million monthly active users. In November 2023, CEO Sam Altman announced that ChatGPT’s weekly active user count had reached his 100 million mark.
* Digital ad spending more than doubled in five years, from $331 billion in 2018 to $720 billion in 2023. The majority of this spending is currently going to programmatic advertising. Programmatic advertising is a form of customizing a stream of ads for each user based on data. It is collected about your online activities, browsing history, spending habits and other similar parameters.
Check out Hindustan Times website and app for daily updates on fashion, health, festivals, travel, relationships, recipes and all the latest lifestyle news.
