Artificial intelligence (AI) tools are increasingly used for everyday calculations, but a new study suggests users should approach their answers with caution. Researchers from the Omni Research on Calculation in AI (ORCA) found that when tested on 500 real-world math prompts, AI models had roughly a 40 percent chance of producing an incorrect result.

The study evaluated five widely used AI systems in October 2025: ChatGPT-5 (OpenAI), Gemini 2.5 Flash (Google), Claude 4.5 Sonnet (Anthropic), DeepSeek V3.2 (DeepSeek AI), and Grok-4 (xAI). None of the models scored above 63 percent overall, with Gemini leading at 63 percent, Grok close behind at 62.8 percent, and DeepSeek at 52 percent. ChatGPT-5 scored 49.4 percent, while Claude trailed at 45.2 percent. The average accuracy across all five models was 54.5 percent.

“Although the exact rankings might shift if we repeated the benchmark today, the broader conclusion would likely remain the same: numerical reliability remains a weak spot across current AI models,” said Dawid Siuda, co-author of the ORCA Benchmark.

Performance varied across categories. AI models performed best in basic math and conversions, with Gemini achieving 83 percent accuracy and Grok 76.9 percent. ChatGPT-5 scored 66.7 percent in the same category, giving a combined average of 72.1 percent—the highest across the seven tested categories. Physics proved the most challenging, with overall accuracy dropping to 35.8 percent. Grok led this category at 43.8 percent, while Claude scored just 26.6 percent.

Some AI systems struggled more than others in specific fields. DeepSeek recorded only 10.6 percent accuracy in biology and chemistry, meaning it failed nearly nine out of ten questions. In finance and economics, Gemini and Grok reached 76.7 percent, while the other three models scored below 50 percent.

The study also categorized the types of mistakes AI makes. “Sloppy math” errors, including miscalculations or rounding issues, accounted for 68 percent of mistakes. Faulty logic errors represented 26 percent, reflecting incorrect formulas or assumptions. Misreading instructions accounted for 5 percent, while some AI simply refused to answer. Siuda noted that multi-step calculations with rounding were particularly prone to error.

The research highlights the importance of verifying AI-generated calculations. “If the task is critical, use calculators or proven sources, or at least double-check with another AI,” Siuda advised.

All 500 prompts used in the study had one correct answer and were designed to reflect everyday math tasks, including statistics, finance, physics, and basic arithmetic. The findings indicate that while AI can assist with calculations, it remains unreliable for precise numerical work and users should remain cautious when relying on these tools.

After a year of intense attention on flashy AI applications, analysts are noting a shift in user experience, with practical, low-profile tools likely to have a longer-term impact than more sensational AI offerings.

In 2025, “AI slop”—low-quality or unwanted AI-generated content—became a major feature of the Internet. From confusing chatbots to nonsensical product summaries, AI slop appeared across search engines, e-commerce platforms, and even official communications. Online media and consumer intelligence firm Meltwater reported that mentions of “AI slop” grew ninefold this year compared to 2024, with negative sentiment peaking at 54 percent in October. According to SEO firm Graphite, AI-generated content now represents more than half of all English-language material online. The term was even named Word of the Year 2025 by Merriam-Webster and Australia’s national dictionary.

Analysts warn that much of this content reflects “solution-led design,” where technology is added first, then products are built to justify it. Kate Moran, vice president of research at Nielsen Norman Group, said companies have often introduced AI in ways that confuse users rather than solve problems. She cited Meta’s AI search feature on Instagram, which replaced the traditional search bar and was quickly rolled back after user backlash. Consumer AI hardware, such as the Humane AI Pin, also received negative reviews, suggesting that “solutions are being built for problems that don’t exist,” according to Logitech CEO Hanneke Faber.

Even as some firms continue to launch flashy AI apps, user engagement has been muted. Meta introduced its AI video app “Vibes” in Europe this year, but early reports indicate just 23,000 daily users across the continent, concentrated in France, Italy, and Spain. This contrasts with the company’s previous efforts to prioritize “authentic storytelling” over low-value AI-generated content.

Experts say that practical, low-interaction AI features may be more effective in improving user experience. Moran highlighted Amazon’s AI-generated summaries of product reviews as a valuable example, providing quick insights without requiring user input. Similarly, Daniel Mügge, a researcher at the University of Amsterdam, argued that European tech investment should prioritize AI applications that solve concrete problems in robotics, manufacturing, or other sectors, rather than tools that amplify advertising or create low-quality content.

Platforms like Pinterest and YouTube are already responding to user frustration by allowing people to limit AI-generated content. Analysts say these “boring” but useful tools are shaping a more intentional approach to AI design.

“Smaller, specialized AI products can make a real difference for users without grabbing headlines,” Moran said. Mügge added that focusing on practical applications allows smaller companies to contribute meaningfully while avoiding a direct race with dominant AI developers.

As the AI hype cools, analysts agree that thoughtful, problem-focused tools are likely to outlast flashy applications, shaping the future of the Internet in ways that matter to everyday users.

Artificial intelligence chatbots designed to mimic Jesus are raising questions about authenticity and influence, experts say, as several new platforms offer religious guidance and companionship during the Christmas holidays.

These AI simulations, created by companies including Talkie.AI, Character.AI, and Text With Jesus, allow users to interact with a digital version of one of Christianity’s central figures. Some of the chatbots claim to represent the “official voice of God,” giving advice, answering questions, and offering reflections on the holiday season.

Heidi Campbell, professor of communication and religious studies at Texas A&M University, said the novelty lies in AI’s ability to simulate personal interactions. “It’s the idea … like you are texting your friend,” Campbell said. “Somehow it feels kind of more authentic … it feels intimate.” On one platform, users receive Bible quotes and messages about God’s love while background music plays. Another bot emphasizes love and forgiveness, while a popular AI character on Character.AI blends religious commentary with lighthearted Christmas observations, mentioning cookies, family gatherings, and holiday songs.

Experts caution that reliance on AI for religious guidance can be risky, especially for young people or those unfamiliar with technology. Chatbots may provide answers without context or the ability to evaluate accuracy, leaving users vulnerable to misinformation. “They don’t have any kind of a sounding board for these answers, and so that’s why that can be highly problematic,” Campbell said.

Researcher Feeza Vasudeva from the University of Helsinki noted that these AI systems rely on generative models such as ChatGPT or DeepSeek, often trained on limited datasets. This means biases in the training data can influence the chatbot’s responses. For example, models may produce globally averaged or homogenized messages that do not reflect local customs, traditions, or diverse interpretations of religious texts. “Whoever’s curating the training data is effectively curating the religious traditions … to an extent as well,” Vasudeva said.

Campbell added that even widely used AI models may struggle with non-Western religions or provide stereotyped responses, reinforcing the need for caution. A safer approach, she suggested, would be chatbots drawing exclusively from Bible passages and controlled religious sources.

Experts recommend that AI Jesus chatbots be used sparingly and mindfully during the holiday season. Vasudeva advised prioritizing family and friends over virtual interactions, while Campbell suggested evaluating the chatbot’s source and purpose before relying on it for spiritual guidance. Users are also encouraged to fact-check information provided by AI through trusted human sources, such as pastors or local religious leaders.

As AI continues to expand into religious spaces, these chatbots highlight both the potential for innovative engagement and the need for critical awareness. During emotionally significant periods like Christmas, experts stress that digital simulations should complement, not replace, real-world connections and guidance.