By Connor Jackson, Michael Frazier, Tyson Cavinder | Jul 1, 2026
Artificial Intelligence, love it or hate it, is here to stay. Two schools of thought have formed around the use of AI in journalism: the first camp believes that AI is the future, and the other believes that AI should be left in the past. There is a strong case for AI use when it comes to sorting and analyzing information, and it can be extremely useful for finding sources, but its weaknesses should encourage caution when using it.
Why is using AI a problem? Oftentimes, individuals find themselves treating ChatGPT or Google’s Gemini as some sort of “Super Intelligence”. If AI really is as smart as it seems, it makes perfect sense to rely so heavily on it. So… the question is, how accurate is AI?
Every year, Google processes FIVE TRILLION searches. With each of these searches, Google’s Gemini gives an AI response at the top of the search window. This AI overview is seen by many as a “stopping point” in their research. The New York Times partnered with Oumi to test the accuracy of these results.
“Oumi analyzed the accuracy of Google’s AI Overviews using a benchmark test called SimpleQA, which is widely used across the industry to measure the accuracy of A.I. systems….Oumi’s analysis focused on 4,326 Google searches. The company found that the results were accurate 85 percent of the time with Gemini 2 (October 2025) and 91 percent of the time with Gemini 3 (February 2026). (Collins et al.)” There has been clear improvement in a short time, but that’s still a shocking amount of inaccuracies. Roughly one in ten results on the AI overview will be wrong. Taking the five trillion searches per year, there are around 500 billion wrong results every year. That means that an estimated 951,294 wrong results are given every. Single. Minute.
Journalists may rely on Google for hundreds of questions per piece to quickly skim through what’s worth researching (World Association of News Publishers (Henriksson)). The number of inaccuracies in basic answers is highly concerning and misleading. More than that, journalists may collaborate with AI to speed up the writing process. From proof-reading to full on cowriting, this raises several problems.
On top of the issues that come with accuracy in search results, the average AI user also has to consider the misinformation that AI helps perpetuate, and in many cases, creates. In a 2025 study, NewsGuard (“NewsGuard One-Year AI Audit Progress Report Finds that AI Models Spread Falsehoods in the News 35% of the Time”), a leading expert in digital security and AI Accountability, found that top AI Chatbots would spread false information about 35% of the time when asked about controversial news topics. ChatGPT was one of the worst offenders– at a 40% rate, almost half the answers given were false. This issue is not intentional, rather it is a side effect of one of the worst uses of AI in journalism content farms.
Content farms are automated networks that mass-produce low-quality content. The AI that creates the content farm slop is able to tailor it to be easily picked up and shared by the Chatbots. The issue with the content these farms are putting out is said best by NewsGuard:“These websites… also tend to churn out dozens of articles a day and are often the originators of false claims.”
But why would someone want to put out a bunch of low-quality, false content? Money. NewsGuard, again, explains it best: “In many cases, the revenue model for these websites is… advertising under which the ad-tech industry delivers ads without regard to the nature or quality of the website.”
Though AI is fallible and often used for less than noble purposes, all is not lost, as Google has been using it to find and eliminate these content farms. AI is a tool like any other, and what matters most is how you use it, and make sure it is not using you.
Media literacy is a term that you have probably heard before–maybe in a school class, on the news, or from your parents–but what actually is media literacy, and how do you learn it?
First, what is media literacy? NAMLE, The National Association for Media Literacy Education, simply describes it as “the ability to access, analyze, evaluate, create, and act using all forms of communication.” For journalism, analysis is the most important. Analysis is the ability to break down a topic into its components to gain a deeper understanding of the whole. This is most important for those consuming media, as they need to be able to recognise bias and avoid echo chambers. The best way, though not the easiest, is lateral reading. Rather than only reading one website (or AI overview) and believing everything in it as gospel truth, lateral readers will find articles from all sides of the political spectrum, read them, then compare. The comparison allows for the lateral reader to recognise and catalog the bias, find false or misleading information, and find the facts based on the shared information in each story.
General media literacy is important, but what is even more important in the age of AI is the ability to discern media that is human-created vs. AI. Five years ago, this would be an easy task- AI couldn’t accurately generate fingers, backgrounds, or words- but now, it’s advanced. Photos and videos can be created that are indistinguishable by the human eye, and that’s where intuition comes in. If you see something that looks a little weird, a little off, a little different, put our evolutionary advantage to use, and check it out. Google’s SynthAI (“SynthID — Google DeepMind”) is the tool for the job, “SynthID embeds digital watermarks directly into AI-generated images, audio, text or video. The watermarks are embedded across Google’s generative AI consumer products, and are imperceptible to humans – but can be detected by SynthID’s technology.”
So if you ever see something that doesn’t look quite right, upload it to SynthAI, or ask ChatGPT to double check, then feel good knowing that you will not be the one to get got by that darn AI.
That darn AI is not all bad though. There are safe and effective ways to use it.
The Stanford Teaching Commons (“Understanding AI Literacy | Teaching Commons”) breaks down AI use into four main categories of literacy: functional (how AI works), ethical (the ethical side of using AI), rhetorical (using language to achieve goals in writing), and pedagogical (using AI to enhance teaching and learning). Reaching a novice level of AI literacy will take practice in several areas. Start off researching some basic terminology, learn some of the basics AI is capable of, and think about where the information is coming from.Then, make a decision on how you or your institution will feel about using AI. Knowing the tone of your writing, experimenting with different prompts, and learning different educational perspectives will greatly enhance a basic level of AI literacy.
Artificial Intelligence has become a major aspect of the modern world; this reflects heavily on the world of journalism in ways that are both uplifting and detrimental. So, how is this already impacting journalism?
According to Yahoo Finance, 82% of all journalists in the current day have utilized artificial intelligence in some capacity within their works. Generative AI is an extremely useful tool that can help enhance the process of journalistic creation. However, certain tools lean more towards AI creation than AI assistance. This can refer to grammar checks, image creation or even full writing.
An issue arises with this in terms of proper accreditation. Artificial intelligence lacks the ability to replicate the compassion that humans feel and sense towards one another. This results in accreditation of AI works being skipped over and passed off as complete human creation. Artificial intelligence works by giving answers that are pieced together responses of preexisting information. Failing to address the use of AI can lead to the decimation of careers. The crutch of AI can lead to transparency concerns, and reputation damage that not only affects the creator, but also the publisher behind such. This occurred in the case of Alex Preston, who was released from The New York Times for this offense.
The rise of artificial intelligence in the world of journalism has snowballed in the last decade. Away from the blurred line between assistance and creation, complete AI articles are beginning to dominate human created articles, not by quality, but by sheer number. In 2016, Wordsmith produced approximately 1.5 billion AI articles, which at the time, nearly equaled the number of human created articles. With the advances made in AI in the last decade, this number now exceeds the number of human articles. Said articles are produced for clicks, lacking human sentiment, creativity, and many times appear sloppy. If this trend continues, it is likely quality articles, that are already few and far between, will become nearly impossible to find.
Works Cited
Collins, Keith, et al. “How Accurate Are Google’s A.I. Overviews?” The New York Times, 7 April 2026, https://www.nytimes.com/2026/04/07/technology/google-ai-overviews-accuracy.html?eafs_enabled=false. Accessed 24 June 2026.
He, Jessica, et al. “Which Contributions Deserve Credit? Perceptions of Attribution in Human-AI Co-Creation.” 2025, https://dl.acm.org/doi/pdf/10.1145/3706598.3713522.
Miroshnichenko, Andrey. “AI to Bypass Creativity. Will Robots Replace Journalists? (The Answer Is “Yes”).” 2018, https://www.mdpi.com/2078-2489/9/7/183.
Muck Rack. “Muck Rack’s 2026 State of Journalism Report Finds 82% of Journalists Use AI.” Yahoo Finance, 2026, https://finance.yahoo.com/sectors/technology/articles/muck-rack-2026-state-journalism-140000040.html.
The National Association for Media Literacy Education. “What is Media Literacy? Media Literacy Defined.” National Association for Media Literacy Education, https://namle.org/resources/media-literacy-defined/. Accessed 26 June 2026.
“NewsGuard One-Year AI Audit Progress Report Finds that AI Models Spread Falsehoods in the News 35% of the Time.” NewsGuard, 4 September 2025, https://www.newsguardtech.com/press/newsguard-one-year-ai-audit-progress-report-finds-that-ai-models-spread-falsehoods-in-the-news-35-of-the-time/.
Sadeghi, McKenzie, et al. “Tracking AI-enabled Misinformation: 3,749 AI Content Farm sites (and Counting), Plus the Top False Claims Generated by Artificial Intelligence Tools.” NewsGuard, 23 June 2026, https://www.newsguardtech.com/special-reports/ai-tracking-center/.
“Understanding AI Literacy | Teaching Commons.” Stanford Teaching Commons, https://teachingcommons.stanford.edu/teaching-guides/artificial-intelligence-teaching-guide/understanding-ai-literacy. Accessed 26 June 2026.
Preston, Alex, and Emma Loffhagen. “The New York Times drops freelance journalist who used AI to write book review.” The Guardian, https://www.theguardian.com/books/2026/mar/31/the-new-york-times-drops-freelance-journalist-who-used-ai-to-write-book-review. Accessed 30 June 2026.
“SynthID — Google DeepMind.” Google DeepMind, https://deepmind.google/models/synthid/. Accessed 30 June 2026.
Henriksson, Temmu. “New survey finds half of newsrooms use Generative AI tools; only 20% have guidelines in place.” World Association of News Publishers, https://wan-ifra.org/2023/05/new-genai-survey/.
