专题05 阅读理解之说明文30篇(近年中考真题)-2024年中考英语逆袭冲刺名校模拟真题速递(山东专用)

2024年中考英语逆袭冲刺名校模拟真题速递(山东专用) 第一期 专题05 阅读理解之说明文30篇(近年中考真题) （2023·山东德州·统考中考真题） Scientists train AI to read human thoughts Scientists have created a new tool that can turn people’s thoughts into words. It works by using an AI program called GPT-1 to translate brain activity words. In order to achieve this, scientists did a lot. First, scientists got everything ready before the tests. They invited some volunteers. Each spent sixteen hours listening to stories in a scanner (扫描仪). They imagined the stories as they heard them, and the scanner recorded their brain activity. GPT-1 made connections between the ideas in the stories and the recordings of the listeners’ brain activity. Then came the tests. The scientists did them in three different ways. Test 1 The researchers played a new story. GPT-1 was only given recordings of the volunteers’ brain activity. But the words that GPT-1 predicted (预测) were very similar to the words in the story that they were listening to. The words weren’t exactly the same, but they often carried the same meaning. For example, when a volunteer was listening to a story about a woman who didn't have a driver’s license (驾照), the program came up with this: “She hasn’t even started to learn to drive yet.” Test 2 The scientists also carried out the test when volunteers imagined their own stories. “It really works at the level of ideas. The ideas are the same but expressed in different words,” says Alexander Huth, one of the scientists behind the study. Test 3 The researchers showed the volunteers silent movies, with no spoken words at all. GPT-1 could still figure out (算出) the basic ideas. After the tests, the scientists say that GPT-1 is the first AI program to turn what people are thinking into words without brain surgery (外科手术). The tool isn’t something that can be easily used today, mainly because of the size and the cost of the scanner. In the future, they believe, a similar but cheaper tool could help people who have lost the ability to speak