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...