Resource guide

Eight Sourced Reasons Why AI Is Bad

Eight objections to AI, each tied to a named source, for readers who want to push back on it.

Last updated October 15, 2026 1762-word guide Editor Ban the Bots

Key takeaways

Merriam-Webster named "slop" its 2025 Word of the Year, defining it as low-quality digital content usually produced in quantity by AI.
Anthropic agreed to pay $1.5 billion to settle an authors' lawsuit over pirated books used to train its AI.
A Stanford paper finds employment of workers aged 22 to 25 in the most AI-exposed jobs is 19% lower than it would otherwise be.
Google says AI Overviews cannot be turned off, and its only supported workaround is the Web filter.
Researchers at UC Riverside and UT Arlington estimate that training GPT-3 could evaporate 700,000 liters of freshwater.
Pew Research Center found 50% of U.S. adults are more concerned than excited about AI, up from 37% in 2021.
A preprint on ChatGPT and essay writing tested 54 participants on one task, so its findings are small and early.

The Main Reasons People Say AI Is Bad

Three reasons lead the case against artificial intelligence (AI): low-quality content, creative work used without pay, and power and water use. Jobs, AI built into products, scams, surveillance and effects on learning follow, each with its own evidence.

No source ranks these reasons, so the top three are our editorial choice. If you want to push back on AI, each reason below comes with a named source you can point to.

Low-quality content: Merriam-Webster named "slop" its 2025 Word of the Year.
Creative work used without pay: Anthropic agreed to pay $1.5 billion to settle an authors' lawsuit.
Power and water: a research group estimates that training GPT-3 could evaporate 700,000 liters of freshwater.
Jobs: Stanford researchers found employment of workers aged 22 to 25 in the most AI-exposed jobs is 19% lower than it would otherwise be.
AI in products: Google says its AI Overviews cannot be turned off.
Scams and deepfakes: an employee at Arup sent about US$25.6 million after a video call with deepfakes of colleagues.
Surveillance: a study by the National Institute of Standards and Technology (NIST) found higher false-positive rates for Asian and African American faces.
Effects on learning: an early Massachusetts Institute of Technology (MIT) preprint found weaker brain connectivity in people who wrote essays with ChatGPT.

Fake and Low-Quality Content

Merriam-Webster named "slop" its 2025 Word of the Year on December 15, 2025. The dictionary defines slop as "digital content of low quality that is produced usually in quantity by means of artificial intelligence." Merriam-Webster president Greg Barlow said people "want things that are real, they want things that are genuine."

Pew Research Center found that 76% of U.S. adults say it is extremely or very important to tell whether pictures, videos and text were made by AI or by people. Pew also found that 53% are not too or not at all confident they can tell.

None of these sources measure how much online content is made with AI.

For more on this reason, read our guide to AI slop.

Creative Work Taken Without Pay

In Bartz v. Anthropic, authors sued over pirated books that Anthropic used to train its AI.

In June 2025, the court ruled that training on lawfully acquired books was fair use. It sent the pirated copies to trial and did not treat them as fair use.

Anthropic agreed in September 2025 to pay $1.5 billion to settle. National Public Radio (NPR) reported about $3,000 per book.

The settlement website says the court granted final approval on July 20, 2026. First payments are expected on or before November 15, 2026.

The court did not rule that AI training is theft.

For the artists' side of this argument, read our guide to AI art theft.

Power and Water Use

The International Energy Agency (IEA) says data centers used around 415 terawatt-hours (TWh) of electricity in 2024. That is about 1.5% of the world's electricity.

The IEA projects that use will "more than double to around 945 TWh by 2030." It calls AI "the most important driver of this growth."

Researchers at the University of California, Riverside and the University of Texas at Arlington estimate the water used in training GPT-3. They say training GPT-3 in U.S. data centers could "directly evaporate 700,000 liters of clean freshwater."

The same paper on AI water use projects that global AI demand could account for 4.2 to 6.6 billion cubic meters of water withdrawal in 2027. That projection is a modeled estimate by the authors.

For the water side of this topic, see our guide to AI water use.

PJM Interconnection's independent market monitor, Monitoring Analytics, said that data center load growth is "the primary reason for recent and expected capacity market conditions."

Across the 2026/27 auction and the one before it, the monitor estimated that data center load accounted for $16.6 billion of $30.8 billion.

These auction figures don't show how the costs reach household electric bills. For the bill question, read our guide to whether data centers raise electric bills.

Jobs and Early-Career Workers

Stanford's Digital Economy Lab published "Canaries in the Coal Mine?" by Brynjolfsson, Chandar and Chen.

The revised version is dated August 12, 2026. It finds employment of workers aged 22 to 25 in the most AI-exposed jobs is 19% lower than it would otherwise be. The paper found no comparable gap for experienced workers.

Pew Research Center found that 64% of U.S. adults think AI will lead to fewer jobs over the next 20 years. Among AI experts, 39% said the same.

For the wider labor argument, read our guide to AI and jobs.

AI Forced Into Products

Google's help page says AI Overviews are a core Search feature. It adds that "Features cannot be turned off."

Google's only supported workaround is the Web filter. The help page says that filter "displays only text-based links without features like AI Overviews."

Pew found that about six in ten Americans want more control over how AI is used in their lives. Pew does not give an exact percentage for that figure.

For the steps, read our guide on how to turn off AI.

Scams and Deepfakes

In 2024, an employee at engineering firm Arup's Hong Kong office joined a video call with deepfakes of Arup's chief financial officer (CFO) and other staff.

The employee then sent HK$200 million, about US$25.6 million, in 15 transfers. Arup confirmed that "fake voices and images were used."

Police were still investigating, and no arrests were reported. The reports don't say any money was recovered.

For more on this kind of fraud, read our guide to deepfake scams.

Surveillance and Face Recognition

NIST tested 189 face recognition algorithms from 99 developers. Its study, published in December 2019, found higher false-positive rates for Asian and African American faces than for Caucasian faces.

NIST said the differences "often ranged from a factor of 10 to 100 times, depending on the individual algorithm."

The study measures how accurate the algorithms are. It does not measure how any government or business uses face recognition.

For the uses of this technology, read our guide to facial recognition.

Learning and Thinking

An MIT Media Lab-led study, with lead author Nataliya Kosmyna, split 54 participants into three essay-writing groups. The groups were ChatGPT, search engine and brain-only.

The researchers measured brain activity with electroencephalography (EEG). The ChatGPT group used a large language model (LLM), and the authors report that "LLM users displayed the weakest connectivity."

The authors also say ownership of essays was lowest in the LLM group. They conclude that LLM users "consistently underperformed at neural, linguistic, and behavioral levels."

This is a small, early study. It is posted as an arXiv preprint, and only 18 participants completed the fourth session.

The findings describe one essay task. They should not be stretched into claims about ChatGPT's effect on people in general.

What AI Is Good At

Pew found that 74% say AI should play at least a small role in weather forecasting. Pew also reports 70% for searching for financial crimes, 70% for searching for fraud in government benefits claims and 66% for developing new medicines.

Experts are more positive than the public. Pew reports that 56% of AI experts expect a very or somewhat positive impact on the U.S. over the next 20 years. Among U.S. adults, the figure is 17%.

Pew notes that its expert sample is unweighted and covers only experts who responded.

Demis Hassabis and John Jumper of Google DeepMind shared half of the 2024 Nobel Prize in Chemistry. The prize was for protein structure prediction, the work behind AlphaFold. David Baker received the other half.

The AlphaFold database held about 214 million predicted protein structures as of 2024.

Why So Many People Distrust AI

Pew found that 50% of U.S. adults are more concerned than excited about the increased use of AI in daily life. That share was 37% in 2021. Ten percent are more excited than concerned, and 38% feel both equally.

The survey ran from June 9 to 15, 2025, among 5,023 U.S. adults.

A separate Pew survey from August 2024 found 51% of U.S. adults more concerned than excited, compared with 15% of AI experts.

Pew found that 53% say AI will worsen people's ability to think creatively. Only 16% say it will improve that ability. For relationships, 50% say AI will worsen people's ability to form meaningful relationships, and 5% say it will improve that ability.

These figures show how people answered. They do not explain the reasons behind each answer.

Who Should Skip These Reasons

Readers who want a ranking of AI products won't find one in these eight reasons. Each reason is an objection tied to a source, so together they argue one side of a contested question.

Readers who want the case for AI should start with the section called What AI Is Good At. Readers who only want Google's AI Overviews gone should skip the reasons and go to our guide on how to turn off AI.

What Would Change Our Answer

The learning section would need rewriting if the MIT study appears in a peer-reviewed journal with different results. The jobs section would change if Stanford revises the 19% figure again.

If Google adds a setting that turns off AI Overviews, the section on AI in products would need rewriting.

A court ruling that treats training on lawfully bought books as infringement would change the creative-work section.

To start, open the section for the reason you care about most and follow the link to its hub page.

Sources

Frequently asked questions

▸ What are three reasons why AI is bad?

The three reasons listed first are low-quality content, creative work used without pay, and power and water use. No source ranks them, so that choice is our editorial call. Each reason is backed by a named source in the sections above.

▸ Why shouldn't you use AI?

Whether to use AI is your call. The eight objections above each come with a source. Pew Research Center found about six in ten Americans want more control over how AI is used in their lives.

▸ What is the biggest problem in AI?

No source ranks AI's problems, so any single answer is a judgment call. Low-quality content comes first in the list. Merriam-Webster named "slop" its 2025 Word of the Year. Unpaid creative work and power and water use come next.

▸ Why do people hate AI?

Pew Research Center measured levels of concern. It found 50% of U.S. adults are more concerned than excited about AI in daily life. Pew's figures show how people answered. They don't explain why.

▸ What are 5 negatives of AI?

Five negatives are low-quality content, unpaid creative work, power and water use, job risk for early-career workers, and AI built into products. Each has its own section above.

Latest related briefings

REGULATION POLICY

FCC’s $2B E-Rate Cut Threatens Internet Access for Millions

If the FCC ends the $2B E-Rate program, millions of students—especially from low-income families—could lose crucial internet access at school and home.

Oct 14, 20264 min read

Read analysis PARENTING EDUCATION

New AI Safety Guidelines Aim to Shield Kids Online—Will They Work?

IEEE released new AI safety guidelines to protect kids online, but their impact depends on adoption by tech companies and schools. Parents should stay vigilant.

Oct 14, 20264 min read

Read analysis JOBS LABOR