Articles

PopAI, short for Popular Artificial Intelligence, is a magazine dedicated to artificial intelligence. The readership of PopAI includes all literate readers who are interested in AI. Since artificial intelligence also requires....

According to the definition of Oxford Languages, " is a system of ideas and ideals, especially one which forms the basis of economical or political theory and policy."...

Consider many retinal images of different views of all practical scenes in which there is a stop sign that projects onto your retina in different angles...

Think of a situation when you are preparing for an exam. One student learns the process of solving problems step by step, while another....

What is M3D VLSI? Traditional chips are built in flat layers — Like a city made up of one-story houses. But real brains aren’t flat...

To avoid Post-Selection, machines must conduct conscious learning which requires brain-size neural networks. However, 2D....

Have you heard of the so-called “Great Leap Forward” in China? The entry of “Great Leap Forward” on Wikipedia wrote....

When you ask ChatGPT a question, it often replies even when it’s wrong...

If you're a high school student dreaming of building the next breakthrough in artificial intelligence

In modern AI, many models look extremely accurate on paper. Papers claim that a certain neural network gets “state-of-the-art” results with very low test error. But Dr. Juyang Weng argues...

Neural networks do not intentionally exaggerate their performance, but our training and reporting practices can unintentionally make them look like skilled performers when we are really showcasing the luckiest network after the lottery has already been run...

Numbers, in machine learning often look trustworthy. A model with "95% accuracy'' seems strong, objective, and scientific. But high scores can create a validation illusion...

Consider how to imagine the architecture of many current AI systems. Almost all published AI projects have an open skull, as shown in the following figure...

Dr. Juyang Weng, a researcher at Michigan State University, has made a startling claim: many of the world's most celebrated AI breakthroughs are based on what he calls "p-hacking" or "post-selection misconduct" — essentially a form of misconduct that is widespread in the field...

According to reports, on January 3, 2026, President Donald Trump announced that the U.S. military had launched strikes across Venezuela that culminated in the capture and arrest of President Nicolás Maduro (2013–2026) and his wife, Cilia Flores...

To understand the motivation of hidden abstraction, you need to first read my paper in the previous issue of the PopAI Magazine, titled "Open Skull Invites a Human Homunculus Who Does Post-Selection". In that article, I argued that an open skull not only attracts Post-Selection misconduct but also forces a human to hardcraft internal representations...

Modern artificial intelligence is often framed as a story of clean, linear progress: better architectures, larger datasets, and higher benchmark scores. Yet behind many celebrated results lies a quieter and more troubling habit — chasing the luckiest network...

Artificial Intelligence (AI) is increasingly being used to monitor students during exams, promising fairness and prevention of cheating. However, the growing reliance on AI proctoring systems raises serious concerns about privacy and ethical boundaries...

In modern AI, many models look extremely accurate on paper. Researchers claim that a certain neural network achieves "state-of-the-art" results with very low test error. But Dr. Juyang Weng argues that many of these results hide a serious problem: a technique called Post-Selection, which gives a dangerously false sense of good performance...

When ChatGPT confidently asserts a historical event that never occurred, or a Vision Transformer detects cancer in pure noise, the reaction is often amusement — until the consequences become real. AI hallucinations (fabricated outputs) and false positives (incorrect affirmative predictions) are not harmless anomalies; they pose serious risks across critical domains...

On June 10, 2026, I asked Gemini ``What is humans-in-the-loop in AI?'' Gemini replied ``Human-in-the-loop (HITL) is an artificial intelligence design pattern where a human actively reviews, modifies, or approves an AI's actions or outputs before execution. It acts as a crucial safeguard to ensure accuracy, safety, and accountability, particularly in high-stakes environments like ...

As I stated in my email to Juyang Weng at 4:02 on May 30, I don't claim expertise on what happened during high-profile competitions between AI systems and human players, such as the Go match between AlphaGo and Lee Sedol...

The world is built upon a foundation of money and finances, and is subject to a constantly shifting tide of profits and losses. Although many of us may not understand the inner workings of the financial world, what we do understand is...

Artificial intelligence is rapidly becoming a key part of cybersecurity. AI systems can help identify malicious software, detect suspicious network activity, and respond to cyberattacks faster than human analysts alone...

Imagine you are preparing for a final exam. Your teacher, looking to see how well you genuinely understand the material, promises to give you a brand-new set of questions you have never seen before---the test set...

We might be wrong about ``intelligence'' when we talk about Artificial Intelligence. Is it intelligence in the truest sense? Maybe we are misunderstanding what these systems do...

Artificial intelligence is the most popular technology used to perform complex tasks, analyze large quantity of data, and assist humans in everyday decisions ranging from writing code to medical research...

Imagine that a student tells you, “I scored 98% on my mathematics exam.” You might be impressed. But what if the student refuses to show the exam, refuses to say which questions they answered incorrectly, and only tells you about the questions they got right? Would you still be confident that the 98% represents their true mathematical ability...

I appreciate Prof. Jay McClelland’s care about the important issue of research ethics concerning AlphaGo and DeepBlue.
Letter 4
The following is my fourth letter to McClelland with regard to this subject, following the first three in [1]. As I stated, McClelland wished to have final words in the last issue. I agreed. From this issue on, McClelland and I will not have such a restriction, so that anybody can respond at any time. We will have direct email exchanges, which will lead to articles in this magazine...

Imagine applying for a job and being rejected by an algorithm you never see. You might assume the system simply compared your skills with the job requirements. But there is another, less visible question: how was the algorithm itself chosen?
A job-matching system may be built by training many different versions of a model...

PopAI, short for \textit{Popular Artificial Intelligence}, is a magazine dedicated to making AI concepts accessible to a broad audience, from students and researchers to policymakers and general readers. It publishes short, thought-provoking articles that connect artificial and natural intelligence, encouraging diverse viewpoints beyond conventional or incremental research...

PopAI, short for \textit{Popular Artificial Intelligence}, is a magazine dedicated to making AI concepts accessible to a broad audience, from students and researchers to policymakers and general readers. It publishes short, thought-provoking articles that connect artificial and natural intelligence, encouraging diverse viewpoints beyond conventional or incremental research...

PopAI, short for \textit{Popular Artificial Intelligence}, is a magazine dedicated to making AI concepts accessible to a broad audience, from students and researchers to policymakers and general readers. It publishes short, thought-provoking articles that connect artificial and natural intelligence, encouraging diverse viewpoints beyond conventional or incremental research...

PopAI, short for \textit{Popular Artificial Intelligence}, is a magazine dedicated to making AI concepts accessible to a broad audience, from students and researchers to policymakers and general readers. It publishes short, thought-provoking articles that connect artificial and natural intelligence, encouraging diverse viewpoints beyond conventional or incremental research...

PopAI, short for \textit{Popular Artificial Intelligence}, is a magazine dedicated to making AI concepts accessible to a broad audience, from students and researchers to policymakers and general readers. It publishes short, thought-provoking articles that connect artificial and natural intelligence, encouraging diverse viewpoints beyond conventional or incremental research...