Zhonglin Wang started playing table tennis only four years ago. But last year, he and other researchers came up with a clever way to improve their game: Build a smart table.

Now they’ve just introduced a prototype to the public. It can measure where a ball lands, how fast the ball’s going and where it’s headed. The table can do that because its innovative wood surface forms the top layer of a novel self-powered sensor. The data it acquires could help players to perform better, says Wang.

Wang is a materials scientist at the Georgia Institute of Technology, in Atlanta. He describes the invention as a machine that can learn to analyze how someone plays. Importantly, the new smart table doesn’t need a battery to detect the ball. When a ball hits the wood, it sets off a chain of events that can both generate a small electric current and record the ball’s behavior.

Wang is an expert at inventing devices that provide their own power. In 2012, he invented a triboelectric(摩擦生电的) nano-generator. He calls it TENG, for short. Even if you’ve never heard of “triboelectricity”, you know the phenomenon. Most of us call it static electricity. The first TENG that Wang put together in 2012 looked like a sandwich. It had layers of charged materials separated by a small gap. When the sandwich bent or changed shape, its layered materials rubbed against each other. This caused electrons to start building up on one side.

What makes the new game table truly unique is its use of wood as one of the TENG’s layers. As it turns out, Wang explains, “Wood is an excellent triboelectric material.” Building a full-size game table would require a large grid of wood TENGs. Wang says that’s on his drawing board as a project for the future.

Inspired by sea cucumbers, engineers have designed tiny robots that rapidly shift between liquid and solid states. On top of being able to shape-shift, the robots are magnetic and can conduct electricity. The researchers put the robots through an obstacle course of mobility and shape-changing tests in a study.

Where traditional robots are hard-bodied and stiff, “soft” robots have the opposite problem; they are flexible but weak, and their movements are difficult to control. “Giving robots the ability to switch between liquid and solid states equips them with more functionality,” says Chengfeng Pan, an engineer at The Chinese University of Hong Kong who led the study.

The team created the new shape-shifting material by putting magnetic particles (磁性粒子) in a metal. Researchers tested the material’s mobility and strength in a variety of contexts. With the aid of a magnetic field, the robots jumped over gaps, climbed walls, and even split in half to cooperatively move other l objects around before combining itself together.

“Now, we’re pushing this material system in more practical ways to solve some very specific medical and engineering problems,” says Pan, On the biomedical side, the team used the robots to remove a foreign object from a model stomach and to deliver drugs on-demand into the same stomach.

“Future work should further explore how these robots could be used within a biomedical context,” says Pan. “What we’re showing are just one-off demonstrations, proofs of concept, but much more study will be required to find how this could actually be used for drug delivery or for removing foreign objects.”

ChatGPT is back in the headlines again as OpenAI, the company behind the incredibly popular chatbot, has now introduced a powerful new language model – GPT-4.

First announced on March 14, GPT-4 goes beyond the capability of the previous language model known as GPT-3.5. The most impressive new feature is its ability to use images as prompts (提示). Its image-based outputs can be as simple as describing the contents of a refrigerator. But they can also be complex like interpreting a series of images to explain what a joke is supposed to be.

But GPT-4’s advancements don’t stop there. The new language model can code (编程) classic games like Tetris (俄罗斯方块) in a matter of minutes, produce the code and layout for a website based on a hand-made drawing, and change the tone and style of its responses. And, perhaps to the worry of teachers, GPT-4’s ability to pass tests has greatly topped the previous version’s. According to CNN, GPT-4 passed a simulated (模拟的) law school bar exam (律师资格考试) with a score in the top 10 percent of test takers, unlike GPT-3.5 which only scored at the bottom 10 percent.

GPT-4 will also see greater use as a tool to power other apps and services. Duolingo, a US educational technology company, plans to use GPT-4 in their upcoming Duolingo Max function to help users improve their language studies through dynamic AI conversations and answer explanations. Another organization looking to use the software is Be My Eyes, which wants to introduce a digital assistant to help visually impaired (视障的) users better understand the world around them. Even the government of Iceland plans on creating a five-year fund to support the digital preservation of the Icelandic language using the software.

However, there are still pitfalls (隐患) in the new language model. GPT-4 still struggles to be completely accurate with its responses, especially with newer information after September 2021. GPT-4 also still occasionally “hallucinates (产生错觉)”, meaning that it will insist that wrong information is true.