AI-Skilled Workers Make 60 Percent More Across Sectors

By Katharina Buchholz, Data Journalist, Statista

PwC's Global AI Jobs Barometer has found that AI-skilled workers earn 62 percent more across the globe than their non-AI-skilled counterparts. The analysis looked at data across one million job ads in 16 different sectors and 27 countries and found that in consumer market jobs, differences were especially stark, with roles requiring AI skills paying 118 percent more than those which don't.

Other high pay gaps could be observed in the areas of technology, telecoms and media (+84 percent earnings for AI-skilled employees), energy, utilities and resources (+75 percent) as well as manufacturing (+73 percent). Below average gains were calculated for the public sector, where (government) employees only earned 16 percent more if they had AI skills. In the health industries, this number stood at 37 percent.

The report also found that the hiring of AI specialists has been rising sharply. While in 2012, 1 percent of jobs posted were for AI specialists, this more than doubled and stood at above 2 percent in 2025. The biggest share of AI jobs was listed in the area of tech, media and telecoms at 11.4 percent of all job listings. Hiring was up in all sectors, with shares of AI jobs reaching between 2 percent and almost 6 percent for the other sectors included in the chart. The exception was the healthcare sector with just around 1 percent of jobs posted having an AI component.

This post was originally published on Statista and republished here with permission.

Edited by Irfan Ahmad.

Read next:

• Being your own boss doesn’t always pay off: What 30 years of data reveal

• Google CEO Pichai’s Stanford Speech Accompanied by Student Protest, Visible in News Coverage but Not in Official Recording
by External Contributor via Digital Information World

Being your own boss doesn’t always pay off: What 30 years of data reveal

Xiaoying Wang, Wilfrid Laurier University and Seok-Woo Kwon, University of Calgary

More than 2.6 million Canadians work for themselves, and according to an annual RBC poll conducted in 2025, 59 per cent of Canadians aspire to own a business — the highest level since 2017.

The appeal is understandable. A recent poll found that nearly two-thirds of people feel they have plateaued at work and see owning a business as their next move. “Being your own boss” has become shorthand for freedom, control and finally getting paid what you are worth.

However, working for yourself does not reliably make people richer or happier. In our new study, we followed 12,686 individuals over three decades, from their late teens into their 50s, to see how self-employment actually played out over a their working lives.

The takeaway should give anyone weighing the leap pause. Simply being self-employed often left people no better off financially, and measurably less satisfied with life, than peers who kept a regular job. Whether the dream pays off comes down to how you do it, not whether you do it.

4 ways through a working life

Not everyone follows the same entrepreneurial path. By tracking people year by year through the National Longitudinal Survey of Youth 1979, run by the U.S. Bureau of Labor Statistics, we found four distinct patterns across adulthood.

Predicted probability of self-employment across adulthood for four distinct entrepreneurial career paths. (Seok-Woo Kwon and Xiaoying Wang), CC BY-ND

The largest group, about 69 per cent, remained in regular employment and rarely worked for themselves. We used them as the benchmark for everyone else.

Another 12 per cent tried self-employment in their 20s, then returned to regular jobs. About 13 per cent did the opposite, entering self-employment in their 40s and becoming more involved over time. A committed six per cent started young and remained self-employed throughout most of their careers.

The same decision — “work for yourself” — produced four very different lives, with different financial and personal outcomes depending on when it happened and what form it took.

It’s not whether, but how

When we looked at both financial and psychological outcomes, one pattern stood out. What set the financially and personally successful apart was not a brilliant idea or sheer grit, but the structure of their business.

People who built a formal, incorporated business — a registered company with its own legal identity — earned more and reported greater life satisfaction than people who never became entrepreneurs. People who simply worked for themselves as solo freelancers did not. On average, they earned no more than non-entrepreneurs and reported lower well-being.

Researchers have long noted that incorporated and unincorporated business owners operate very differently. Incorporated owners tend to run more ambitious and growth-oriented ventures compared to unincorporated owners.

In Canada, people with an incorporated business are more likely to plan to expand (37.6 per cent versus 22.6 per cent of unincorporated owners), and incorporated businesses are more likely to survive and typically earn more along the way.

Our results suggest the difference goes beyond income: the legal structure of a business is closely tied to whether self-employment improves lives or wears people down.

However, it’s important to note this is not a case of incorporate and watch your life improve. People who incorporate often start with advantages like more education, professional experience and skills.

The gap showed up even among people with similar backgrounds, but incorporation usually signals something about how a business was conceived from day one rather than acting as a simple fix after the fact.

Timing and the long game

Age also matters, but not the way popular accounts suggest. The image of the brilliant young founder who drops out of school to build the next big thing is mostly a myth, and experience turns out to be an asset.

Among people who built incorporated businesses, lifelong entrepreneurs earned the most, early starters reported the strongest life satisfaction and those who started in midlife struck the best balance between the two. The most rewarding path appeared to be building a formal business after accumulating skills, savings and industry experience.

Staying in business is not only about talent or timing, either. We found that people who grew up with books, magazines and a library card at home — a form of early-life advantage researchers call cultural capital — were more likely to sustain a business over the long term, even though they were no more likely to start one.

The skills and habits that keep a business alive for decades, it seems, may be formed long before anyone writes a business plan.

So should you take the leap?

None of this argues against self-employment. Instead, it suggests being clear-eyed about which path you are choosing. If the goal is greater financial security and a better quality of life, our findings point to building a registered company after gaining skills, savings and industry experience, rather than immediately after graduation.

Lifelong founders tend to earn the most, but for most people the midlife path offers the best balance of income and well-being.

If that is true for individuals, it may also be true for the systems meant to support them. Most programs are designed to help people launch, but far fewer help them incorporate, grow and survive.

Much of the funding that helps a business grow presumes it is already incorporated, leaving the solo self-employed — the group already earning the least — on the outside. If we want more people to reach the path that pays off, the support has to start earlier and last longer than the launch.

Xiaoying Wang, Assistant Professor of Strategic Management, Wilfrid Laurier University and Seok-Woo Kwon, Robson Professor in Entrepreneurship, University of Calgary

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Reviewed by Irfan Ahmad.

Read next: Google CEO Pichai’s Stanford Speech Accompanied by Student Protest, Visible in News Coverage but Not in Official Recording

by External Contributor via Digital Information World

The ‘right to repair’ movement has a point, but consumers should read the warranty fine print first

Wayne Fu, University of Michigan-Dearborn

Image: Sasun Bughdaryan / Unsplash

The “right to repair” movement is gaining steam as consumers push corporations to offer them more freedom to fix products – from cars to dishwashers to toys.

In April 2026, farm equipment maker Deere & Co. inked a US$99 million settlement in a class action suit over its prohibition on independent repairs to its increasingly high-tech equipment – another win for the movement. While the company didn’t admit wrongdoing, it will let farmers make more repairs themselves.

Equally significant, this case showed that the Federal Trade Commission, a lead plaintiff, may be more willing to protect consumers against the growing corporate control over servicing products after purchase.

Even President Donald Trump has weighed in. At an Oval Office event on June 4, 2026, he described existing restrictions as “strange” after he met with auto executives. “Nobody’s allowed to fix their car. … So I thought we’d do something about that,” he said, without offering details.

This push is understandable. As consumers want more reliable products, gaining the right to repair them with their own parts makes sense.

But they often overlook existing protections in their product warranties, which obligate the manufacturer to repair or replace if something goes wrong.

As a scholar focused on operational sustainability in supply chains, I have found that strong warranties aren’t just a safety net for buyers. They help companies build trust and stand out. Hyundai and Apple, for example, have used strong warranty programs to keep customers coming back for repairs within their own networks while maintaining profit margins. But many shoppers overlook this tool as the political momentum for the right to repair grows.

Back to the 1970s

Many automotive and electronics manufacturers have been making it harder for consumers to use parts not produced or authorized by the original manufacturer. For example, data can be transmitted back to the manufacturer in real time and flag a part from an independent supplier as incompatible.

Another common tactic is the use of “warranty void” stickers, which claim that repairs done by a third-party service will cancel the manufacturer’s product warranty. These practices have drawn widespread criticism for suppressing competition and encouraging planned obsolescence – and are among the main targets of right to repair advocates.

But consumers have a tool that’s widely underused: A 1975 law that prohibits voiding a warranty simply because an independent mechanic or a part from an independent vendor was used. This measure was designed to discourage the production or sale of low-quality products and sought to protect consumers from excessively restrictive coverage and bad-faith corporate negligence.

It was this law as well as several other consumer protections that the Biden administration’s FTC cited in 2024 when it warned several companies that they improperly restricted their warranty terms.

One reason consumers are largely unaware is that most find the text of warranty terms and the disclaimers difficult to read. And this isn’t an accident. Many manufacturers see warranties as sunk costs that should be avoided, and they have no incentives to clarify the terms or honor them.

Trust pays

Companies should rethink their approach to warranties – because it makes good business sense.

When manufacturers see they can no longer prevent third-party repairs, or offer warranties that are hard to redeem, it’s usually because they decide it’s more cost efficient to produce low-quality products. But that choice often cuts into the producer’s own profit while leaving the consumer worse off – and has a worse environmental impact, a recent study suggests.

A smarter option for manufacturers would be to establish a better, more efficient service network and offer more attractive warranty programs to retain customer loyalty.

Some companies have demonstrated that strong warranties pay off. Outdoor apparel maker Patagonia offers an “ironclad guarantee” to repair or replace its products for any reason. The “all mighty guarantee” offered by Osprey, which also manufactures outdoor gear, will repair or replace any damaged or defective product free of charge.

Then there’s retail giant Costco’s automatic extension of manufacturers’ warranties on major appliances and electronics – a major driver of its success.

Meanwhile, under the right to repair measures currently proposed, the post-purchase service and repair markets would likely get more competitive. New rules would let outside service providers and warranty companies gain better access to fix the products. Extended warranties and service contracts would then become even more prevalent, and manufacturers would need to become more vigilant.

Consumers also need more protection from the FTC, the top federal regulator tasked with consumer protection. Indeed, the right to repair movement reflects, in part, public disappointment that the government has failed to serve as a watchdog amid misleading corporate claims about warranty protections.

While the FTC has occasionally sued companies to protect consumer rights under the 1975 law, it has the legal tools to be more aggressive. Such a shift would not only change the corporate culture around warranties but send a message to consumers that warranties should work for them.

Enforcement, not just choice

If manufacturers embraced stronger warranty enforcement, consumers would benefit the most, provided they’re aware of what these protections entail. But manufacturers would come out ahead, too. By building up efficient service networks and offering more versatile warranties, they would remain competitive and foster customer loyalty, Hyundai being a good example.

The right to repair movement is in many ways a live policy experiment on expanding access to repair markets and giving consumers more choices. But more choices don’t necessarily lead to a better outcome. Coupling repair rights with stronger warranties and better enforcement is the best way a company can claim the prize it always says it wants: more satisfied consumers.

Wayne Fu, Associate Professor of Decision Science, University of Michigan-Dearborn

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Reviewed by Irfan Ahmad.

Read next: How Americans Feel About Sharing Their Data With AI

by External Contributor via Digital Information World

How Americans Feel About Sharing Their Data With AI

By Cloaked Team

You probably know that apps collect data about you. What you may not know is how many people are fighting back and how. A new Cloaked survey of 1,009 U.S. adults reveals that nearly one in three Americans have given an AI platform a fake name or fake birthday when asked for personal information. More than half have opted out of data collection, tracking, or targeted ads entirely. And nearly all of them keep using the same apps anyway.

That is the tension at the heart of this report. Americans do not trust AI platforms with their personal data. They feel monitored, powerless, and resigned. But they have not logged off. Instead, they are quietly pushing back in small ways, lying about who they are, opting out where they can, and covering their cameras just in case. This report examines what Americans are most afraid of sharing, which AI scenarios feel the most invasive, and what it would actually take to earn their trust.

Key Takeaways

• Fewer than 1 in 5 of Americans (18%) say they trust AI to keep their personal data secure.
• Americans are most uncomfortable sharing their Social Security number (88%), financial information (87%), and biometric data (74%) with AI.
• Over 2 in 5 Americans (41%) say they would leave a platform if they shared data with government agencies.
• Over 3 in 5 Americans (64%) believe AI is making decisions about them without their knowledge or consent.
• Nearly 1 in 2 Americans (44%) say they would pay more for a service that guaranteed their personal data would never be processed by AI.

Trust, Transparency, and the Price of Privacy

Americans are increasingly aware of how their personal data is collected and used, and many feel they have less control over it than in the past. The findings below highlight how attitudes toward privacy, trust, and data sharing are shifting as new technology emerges.

• Nearly 2 in 3 Americans say they have less control over their personal data today than five years ago.
  
  
• More than 1 in 2 Americans say they have accepted that companies know more about them than they are comfortable with.

• Boomers are the most likely of any generation to believe some loss of privacy is an unavoidable cost of modern technology (65%), compared to 54% of Gen Z.

• Gen Z is the most likely generation to feel powerless to protect their personal data from AI (59%), compared to 45% of Boomers.

• Nearly 1 in 2 Gen Z Americans (49%) say they would pay more for a service that guaranteed their data would never be processed by AI, compared to 37% of Boomers.

• More than 2 in 5 Americans said learning that AI was making decisions about them regarding credit, hiring, or insurance without their consent would cause them to leave a platform; it is the single most cited dealbreaker among those tested.

• Boomers are the most likely generation to say they would leave a platform if their data was shared with government agencies (51%), compared to 37% of Gen Z.

• While 52% of Americans feel comfortable with AI detecting fraudulent transactions, 59% feel uncomfortable with AI scanning their emails to personalize ads.

• Nearly 2 in 5 Americans are more concerned about AI collecting their children's data than their own.

Resistance, Evasion, and the Data Americans Refuse to Share

As AI tools become more integrated into everyday life, many Americans are drawing clear boundaries around what data they are willing to share. These findings reveal the situations that feel most invasive and the actions people are taking to protect their privacy.

• Nearly 3 in 4 Americans say a bank reviewing their social media posts to determine creditworthiness feels creepy or invasive, with Boomers being the most alarmed of any generation (82%).

• More than 2 in 3 Americans (70%) say an app that listens to their conversations to recommend products feels creepy.

• Gen Z is more likely than any other generation to say their trust in AI platforms have decreased in the past year (28%), compared to just 11% of Boomers.

• More than 1 in 2 Gen Z Americans (55%) feel uncomfortable sharing biometric data with AI platforms, compared to 42% of Millennials.

• Women are more likely than men to have physically covered or disabled a camera on a device due to AI privacy concerns (43% vs. 34%).

• Gen X is more likely than any other generation to have deleted an account or app over AI privacy concerns (56%), compared to 42% of Millennials and 43% of Gen Z.
• Meta AI users are the most comfortable of any group with AI handling their personal data (67%) and the most likely to trust AI to keep their data secure (36%)—nearly double the rate of ChatGPT users (20%).

Below are the AI tool users most comfortable with AI handling their personal data:

• Meta AI users are the most comfortable with AI handling their personal data (67%).
• Claude users are the second most comfortable (63%).
• Grok users are the third most comfortable (62%).

‍Below are the AI tool users most likely to trust AI to keep their personal data secure:

• Meta AI users are the most likely to trust AI to keep their personal data secure (36%).
• Perplexity users are the second most likely (31%).
• Grok users are the third most likely (27%).

Taking Back Control

Americans are not indifferent to what AI platforms are doing with their data. They are frustrated, resigned, and quietly fighting back in whatever ways they can, lying about who they are, opting out where it is offered, covering their cameras, and continuing to use the same apps anyway because not using them no longer feels like a realistic option.

The data from this survey does not describe a public that has made peace with the current state of AI and personal privacy. It describes a public that has run out of better options. But surrender is not the only path forward. Tools exist today that let you use the internet, sign up for services, and engage with AI-powered platforms without handing over your real name, your real phone number, or your real email address. Learn more about how Cloaked can help you take back control of your personal data at cloaked.com.

Methodology

A survey of 1,009 U.S. adults was conducted on behalf of Cloaked. Respondents were asked about their comfort levels with sharing personal data with AI-powered platforms, their trust in AI companies, actions taken in response to AI privacy concerns, and which AI-related scenarios they found most alarming. The generational breakdown was Gen Z (17%), Millennials (52%), Gen X (23%), and Baby Boomers (7%).

This post was originally published on Cloaked Research Hub and republished here with permission.

Edited by Irfan Ahmad.

Read next: 

• YouTube Expands In-App Video Sharing and Messaging to More Markets

• The consequences of relying on AI for accurate news
by External Contributor via Digital Information World

5 ways data centers endanger their local communities and the country as a whole

Neha Gour, George Mason University; Ed Maibach, George Mason University, and Luis Ortiz, George Mason University

Image Caption: Hosted by the Stanford Research Computing Facility (SRCF) on the SLAC campus, the U.S. Data Facility is the main hub of Rubin's data infrastructure. Credit: Olivier Bonin/SLAC National Accelerator Laboratory / NOIRLab. Licensed under a Creative Commons Attribution 4.0 International License.

Every internet search, streamed video and AI-generated response depends on a data center somewhere. Driven by rapid growth in artificial intelligence, cloud computing and cryptocurrency, data centers have become the backbone of the modern digital economy. But though their key role is in enabling virtual and remote experiences, data centers are physical buildings in real communities around the nation and the globe.

The United States hosts more than 4,000 data centers – more than any other country. The U.S. Department of Energy expects that, taken together, all U.S. data centers will consume as much as 12% of all U.S. electricity by 2028. In 2023, data centers consumed about 4.4% of total U.S. electricity – roughly 176 terawatt-hours.

In the U.S., Virginia has more data centers than any other state – over 600, two-thirds of which are in the northern Virginia suburbs of Washington, D.C. In 2023, the state’s data centers consumed about 26% of Virginia’s total electricity supply – a higher share than in any other state.

We study science communication, climate science and public health, so we wanted to understand how data centers in Virginia affect the people who live near them and the broader public.

We found that the data centers that already exist affect nearby residents and the nation as a whole in five main areas: air quality, water quality, noise levels, land use and energy costs.

Air pollution

Data centers generally operate 24/7 and consume enormous amounts of electricity, which must be generated somewhere – either near the data center or farther away.

When fossil fuels are burned to generate that power, they emit a wide range of air pollutants, including those linked to lung disease, cardiovascular disease, stroke and neurological conditions. They also emit heat-trapping pollution that causes global warming and climate change, which, in turn, worsens air pollution further.

Generating power for U.S. data centers in 2023 emitted the equivalent of 2.2% of the nation’s greenhouse gas emissions. Other air pollutants emitted from fossil-fuel combustion are associated with increased risk of ADHD and autism in children and risks of Parkinson’s and Alzheimer’s diseases in older adults.

Unless the energy powering data centers comes from clean energy sources, such as solar, wind or geothermal, generating that electricity also pollutes the air. People who live near fossil-fuel burning power plants, whether in communities that also host data centers or in distant states, are exposed to air pollution. And during electrical outages, on-site diesel generators kick in, releasing large amounts of air pollution that can harm data center employees and nearby residents alike.

Water consumption and pollution

Data centers require vast quantities of water to cool their servers. Globally, they are projected to consume between 4.2 billion and 6.6 billion cubic meters of water annually by 2027. In the United States, data centers already rank among the top 10 industrial water users.

In northern Virginia, data center water use has risen sharply. In Loudoun County alone, just northwest of D.C., potable water use by data centers more than doubled between 2019 and 2023, while facilities across northern Virginia consumed nearly 2 billion gallons of water in 2023.

This demand can strain local rivers, aquifers and municipal water systems, even in regions like the mid-Atlantic that are not usually prone to drought, but especially in regions like the U.S. Southwest that face persistent droughts.

Noise pollution

Data centers’ continuous operation means that cooling systems, including air chillers and cooling fans, generate a persistent humming sound around the clock – as do any generators that are in use to provide power.

In northern Virginia, some residents have complained about an industrial-scale “drone” or “hum.” Measurements at the data centers that were the subject of complaints found noise levels were between 40 and 59 decibels on residential property.

Those noise levels are quieter than a conversation with someone 3 feet away and not loud enough to damage people’s hearing or violate local noise ordinances. But they are close to levels the EPA says reduce people’s ability to work, sleep and exercise. Some people have complained that data center noise has given them trouble sleeping and concentrating, and some have said they avoid using their homes’ outdoor spaces, where the noise is louder.

Land use and community well-being

Data center expansion often targets land near green spaces, agricultural areas or rural communities where developers can secure affordable land with access to existing electricity supplies.

Converting green space into industrial facilities can diminish health benefits associated with being in and near natural environments, including opportunities for physical activity and improved mental well-being.

In Virginia, residents living near data center construction have reported increased exposure to truck traffic and diesel exhaust, which can contribute to respiratory and cardiovascular health risks, especially in children and older adults. While these effects are typical of large construction projects, they can be amplified when several data centers are clustered together.

In places like Prince William County, Virginia, developers have proposed data centers on roughly 2,400 acres of undeveloped land in the Rural Crescent, an area designated by the county’s planners to remain relatively undeveloped. Those data centers could transform open space and rural farmland into industrial zones, disrupting communities with long-standing ties to the land.

Rising energy costs

As data centers increase electricity demand, they put upward pressure on energy prices across the grid. A 2024 Virginia legislative report found that the state’s typical residential electricity bill could rise by $14 to $37 per month by 2040 because of grid strain tied to data center growth – a 9% to 25% increase over current average bills, and a figure that does not factor in potential inflation.

These higher costs are paid by all consumers, but they place a greater burden on families that are most economically distressed, who also tend to have more health problems. Lower-income families spend a higher share of their budget on electricity, and when bills rise, the consequences can include reduced access to adequate heating and cooling, increased risks of heat-related illness and cold-related cardiovascular stress, as well as difficult choices between paying for energy and food or healthcare.

What can be done

Many of these health harms can be mitigated through better planning and design.

Increasing the share of renewable energy used to power data centers would help reduce air pollution and associated health harms.

Using recycled water in targeted systems that cool individual server rows or racks rather than whole buildings can significantly reduce cooling energy demand, with some studies estimating reductions of up to 29%.

On noise, a Leesburg, Virginia, data center reduced low-frequency tonal noise by reengineering its fan mounts.

And on energy costs, requiring large-scale data centers to cover more of the grid costs they create could help protect residential customers from higher electricity bills.

The world’s digital infrastructure runs through data centers, and that is not changing. We believe that expanding this infrastructure without protecting the health of surrounding communities is an unacceptable option.

Neha Gour, Ph.D. Candidate in Science Communication, George Mason University; Ed Maibach, Distinguished University Professor Emeritus of Communication, George Mason University, and Luis Ortiz, Assistant Professor of Atmospheric, Oceanic and Earth Sciences, George Mason University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Reviewed by Irfan Ahmad.

Read next: Is Regulation Holding Back Tech in the EU?

by External Contributor via Digital Information World

Is Regulation Holding Back Tech in the EU?

By Felix Richter, Data Journalist, Statista

Two years after unveiling Apple Intelligence at WWDC 2024, Apple’s most wanted AI feature – a smarter and more capable version of its digital assistant Siri – is finally around the corner. Unless you live in the EU, that is.

On Monday, Apple officially announced Siri AI, an “entirely new version of Siri,” which Apple describes as “profoundly more capable and personal.” Originally planned to be launched alongside Apple Intelligence in late 2024, the AI-powered Siri was marred with problems, delayed several times and ultimately rebuilt from the ground up. The new Siri will be available as a beta to English-language users later this year, with one notable exception: the European Union.

As Apple announced in a separate press release, Siri AI will not be coming to the EU for the foreseeable future due to the Digital Markets Act’s interoperability rules, which would require Apple to grant other virtual assistants a degree of system-level access, which the company deems unsafe and unacceptable. It’s not the first time that Apple has held back a release in the EU to ensure compliance with EU rules and, judging by the strongly worded press release, the company is not happy about it.

The first suite of Apple Intelligence features, released in the U.S. in late 2024, was also delayed several months in the EU and other companies have encountered similar problems when trying to comply with EU laws. Back in 2023, when Meta launched Threads, a text-based social media app linked with Instagram, the EU was also excluded due to GDPR and DMA-related concerns. Back then, it was impossible to sign up to Threads without linking an Instagram account, a practice likely non-compliant with EU rules. It took several months for those issues to be resolved and, when Threads finally launched in the EU in December 2023, users had the option to create an account from scratch.

In recent years, the EU has made a name for itself by repeatedly going toe-to-toe with U.S. tech on issues ranging from tax evasion to privacy laws and antitrust rules. And while this shows that even the most powerful companies in the world must comply with the rules, it has also created a situation where EU users are left behind increasingly often when it comes to the latest tech or digital services. This could create backlash against regulation, which is why it’s important that the EU demonstrates its will to cooperate with tech companies to find solutions, as it has in the past.

This post was originally published on Statista and republished here with permission.

Reviewed by Irfan Ahmad.

Read next:

• Users trust AI and human fact-checkers equally, but for different reasons

• New research from a multi-institution consortium finds major AI models often omit religious perspectives in responses
by External Contributor via Digital Information World

Users trust AI and human fact-checkers equally, but for different reasons

By Jonathan F. McVerry, The Pennsylvania State University

Image: AI-assisted, generated for illustrative purposes by DIW

Users tend to trust artificial intelligence (AI)-powered fact-checkers as much as human fact-checkers, but for different reasons, according to a new study led by Penn State researchers. The researchers said there is no definitive “winner” when comparing the two fact-checking systems, because users see distinct strengths and weaknesses in each.

In their study published in Media Psychology, participants tended to trust AI more for large-scale scanning tasks, like identifying “red flags” in social media posts. They trusted humans for more nuanced fact-checking that requires piecing together evidence or interpreting complicated situations.

“There's a very clear distinction that emerges from the study that AI is considered good at low-level linguistic features, like identifying telltale signs that something is not credible,” said author S. Shyam Sundar, Evan Pugh University Professor and James P. Jimirro Professor of Media Effects at Penn State. “Humans are seen as being better at corroborating evidence from multiple sources.”

The research team first conducted a pretest to identify six news headlines that varied in credibility. Two hundred and ninety-one participants residing in the United States were then shown those headlines in simulated social media posts via an application created for this study called FactDeck. Some posts were labeled as fact-checked by an AI system and others by human fact-checkers. Participants saw one of three types of explanations:

• Evidence-based: The system labeled the post false with a reference to the information that contradicted the post.
• Feature-based: The system flagged suspicious wording or unusual phrasing.
• “Black box”: No explanation was given for why the post was marked false.

The researchers focused on “machine heuristics” — mental shortcuts people use when evaluating AI, based on stereotypes about machines. They found that while participants assumed AI systems were objective and accurate, they also distrusted AI for lacking human judgment. When it came to determining which system garnered the most trust, first author Mengqi Liao said the two opposite perspectives offset each other.

“Some studies only compare AI versus human fact-checkers, to find out which is trusted more,” said Liao, assistant professor at the University of Georgia, who completed her doctoral studies with Sundar at Penn State. “They get a lot of inconsistent results. That’s why we proposed a competing hypothesis that showed how positive and negative views of both can coexist and cancel each other out.”

Liao added that users preferred some explanation rather than none — the “black box” option.

“We want to provide enough explanation to users that helps them better understand how the system makes a specific decision. It may help them calibrate their trust,” Liao said. “They're not just relying on the system’s decision. They can also make a judgment based on how the system reached the decision.”

The findings suggest that effective fact-checking tools should not only provide accurate results but also explain how those results are reached. Liao said that fact-checking programs should help people recognize what AI systems are good at and what they're not, rather than relying on users’ own naïve, outdated notions of machine capabilities.

Sundar said this is increasingly important as AI fact-checking becomes more necessary. Human fact-checking can’t keep up with the volume of misinformation on social media today.

“The ideal situation would be a human-AI collaboration, but it's not always possible for humans to intervene and check for evidence from multiple sources,” Sundar said. “So, we are going to have to, at some level, completely automate this whole fact-checking business, and rely on AI, which can be much better at efficiently sifting through evidence from multiple sources than humans, despite what people think.”

Sian Lee, assistant professor at the University of Mississippi, who earned a doctorate in informatics from Penn State in 2024; Annie Dooley, a doctoral student at the Ohio State University, who earned a master’s degree in media studies at Penn State; and Aiping Xiong, assistant professor of privacy and cybersecurity informatics at Penn State, were authors on the paper as well.

This post was originally published on Penn State News and republished here with permission.

Reviewed by Irfan Ahmad.

Read next: New research from a multi-institution consortium finds major AI models often omit religious perspectives in responses
by External Contributor via Digital Information World