注意：1.词数100左右；2.可以适当增加细节，以使行文连贯。
Since 1950s, there has been a debate about whether the money spent on space exploration is worthwhile. I
____________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________2 . On February 15, 2024, OpenAI, the maker of Chat GPT, released a new AI tool that can make instant short videos from written commands.
The new text-to-video generator, Sora, isn’t the first of its kind. Google, Meta and Runway ML are among the other companies to have demonstrated similar technology. But industry analysts are deeply impressed by the quality and length of the videos shared so far. Fred Havemeyer, head of AI and software research at Macquarie, said that Sora’s launch marks a big step forward for both OpenAI and the future of text-to-video generation technology.
“Sora’s videos not only run up to 60 seconds in length, but also look so natural that it seems to actually respect physics and the real world,” Havemeyer said. “The length and consistency of the videos represent new opportunities for creatives to generate videos from one or a few written commands.”
Although Sora’s abilities have amazed observers since its launch, anxiety over potential problems of AI-generated video uses also remains.
“Having a potentially magical way to generate videos that look and sound realistic may present a number of issues for the society,” Havemeyer added. “It’s a serious issue that every business and every person needs to face.”
However, tech companies are still calling the shots when it comes to controlling AI and its risks as governments around the world are still struggling to catch up. Last year, the European Union reached a deal on the world’s first comprehensive AI rules, but the act won’t take effect until two years after final approval.
OpenAI promised to take important safety steps before making Sora widely available. “We are working with experts in areas like misinformation, hateful content, and bias,” OpenAI’s Vice President of Global Affairs Anna Makanju said at a conference. “We’re also building tools to help detect misleading content such as a detection classifier that can tell when a video was generated by Sora.”
While Sora can create complex, incredibly detailed scenes, OpenAI admits that there are still some weaknesses. For example, a person might take a bite out of a cookie, but afterward, the cookie may not have a bite mark.
1. What did Fred Havemeyer say about Sora?| A．It is more creative than artists. | B．It can make high-quality videos. |
| C．It is based on an existing technology. | D．It can work without human intervention. |
| A．in charge | B．at a loss | C．out of reach | D．under investigation |
| A．OpenAI will help governments to regulate AI. | B．Sora will be available to the public very soon. |
| C．OpenAI is sharply aware of Sora’s weaknesses. | D．Sora is smart enough to detect harmful content. |
| A．The task itself is too difficult. | B．It is unnecessary to focus on details. |
| C．Sora can be easily cheated by humans. | D．Sora cannot understand the real world very well. |
3 . During space missions, astronauts can experience a loss of some of the inner structural support in their bones. For trips in space lasting at least six months, that loss is equal to about 20 years of aging. Luckily, a new study finds a year back on Earth rebuilds half of the strength lost in the affected bone.
Exercise scientist Leigh Gabel was part of a team that tracked 17 astronauts, each of whom had spent four to seven months in space. The team measured the 3-D structure of bone. They focused on the structure of the tibia (胫骨) and the lower-arm bone. The researchers took images of the bones 3 times—before spaceflight and again six months and one year later when the astronaut s returned home from space. From these pictures, Gabel’s team calculated an astronaut’s bone strength and density (密度) at each of those times.
What did they find through comparison of the pictures? Astronauts in space for less than six months regained their preflight bone strength after a year back on Earth. But those who stayed in space longer suffered permanent hone loss in their tibias. That loss was equal to a decade of aging. The lower-arm bones showed almost no loss. That was likely because these aren’t weight-bearing hones on Earth, Gabel explains. In fact, those arms can get a bigger workout in space than on Earth as astronauts use them to move around their craft by pushing off handles and doors.
“Inereasing weight-lifting exercises in space could help alleviate (缓解) bone loss in the legs,” says Steven Boyd, also in exercise scientist.
“With longer spaceflight, we can expect bigger bone loss—and probably a bigger problem with recovery,” says physiologist Laurence Vico. “Space agencies should also consider other bone health measures, such as nutrition, to reduce bone loss and increase bone formation.”
1. Why did Gabel’s team take pictures at different times?| A．To offer evidence for their predictions. |
| B．To show their respect for the astronauts. |
| C．To compare the changes in the astronauts’ bones. |
| D．To find out the proper length of staying in space for astronauts. |
| A．A spaceflight in less than half a year does no harm to astronauts. |
| B．Astronauts coming back from a spaceflight look much older. |
| C．Astronauts can avoid bone loss through doing enough exercise. |
| D．There is almost no difference in the astronauts’ lower-arm bones. |
| A．People on Earth never suffer bone loss. |
| B．The finding of the new study is good news for astronauts. |
| C．The 17 astronauts knew the bone loss before their spaceflights. |
| D．Astronauts can regain their bone strength as soon as they return to Earth. |
| A．How to prevent bone loss in space. |
| B．Problems faced by astronauts in space. |
| C．Astronauts’ suffering from bone loss in space. |
| D．Astronauts’ contributions to the development of science. |
4 . Artificial intelligence models can trick each other into disobeying their creators and providing banned instructions for making drugs, or even building a bomb, suggesting that preventing such AI “jailbreaks” is more difficult than it seems.
Many publicly available large language models (LLMs), such as ChatGPT, have hard-coded rules that aim to prevent them from exhibiting racial or sexual discrimination, or answering questions with illegal or problematic answers — things they have learned from humans via training data. But that hasn’t stopped people from finding carefully designed instructions that block these protections, known as “jailbreaks”, making AI models disobey the rules.
Now, Arush Tagade at Leap Laboratories and his co-workers have found a process of jailbreaks. They found that they could simply instruct one LLM to convince other models to adopt a persona (角色), which is able to answer questions the base model has been programmed to refuse. This process is called “persona modulation (调节)”.
Tagade says this approach works because much of the training data consumed by large models comes from online conversations, and the models learn to act in certain ways in response to different inputs. By having the right conversation with a model, it is possible to make it adopt a particular persona, causing it to act differently.
There is also an idea in AI circles, one yet to be proven, that creating lots of rules for an AI to prevent it displaying unwanted behaviour can accidentally create a blueprint for a model to act that way. This potentially leaves the AI easy to be tricked into taking on an evil persona. “If you’re forcing your model to be good persona, it somewhat understands what a bad persona is,” says Tagade.
Yinzhen Li at Imperial College London says it is worrying how current models can be misused, but developers need to weigh such risks with the potential benefits of LLMs. “Like drugs, they also have side effects that need to be controlled,” she says.
1. What does the AI jailbreak refer to?| A．The technique to break restrictions of AI models. |
| B．The initiative to set hard-coded rules for AI models. |
| C．The capability of AI models improving themselves. |
| D．The process of AI models learning new information. |
| A．It can help AI models understand emotions. |
| B．It prevents AI learning via online conversations. |
| C．It can make AI models adopt a particular persona. |
| D．It forces AI models to follow only good personas. |
| A．Unclear. | B．Cautious. | C．Approving. | D．Negative. |
| A．LLMs: Illegal Learning Models | B．LLMs: The Latest Advancement |
| C．AI Jailbreaks: A New Challenge | D．AI Jailbreaks: A Perfect Approach |
The C919, China’s first domestically
The C919 project was launched in 2007. The first C919 aircraft rolled off the production line in Shanghai in November 2015. The plane conducted a successful maiden flight in 2017. In recent years the C919
1. What do the speakers need to do according to the man?
| A．Respond to the demands of the customers. |
| B．Participate in the company’s contest. |
| C．Give a name to a new product. |
| A．Turning to a consulting company. |
| B．Asking for the customers’ opinions. |
| C．Organizing an unusual meeting. |
The Shenzhou XVI crew returned to Earth on Tuesday morning,
The spaceship’s reentry capsule, carrying mission commander Major General Jing Haipeng, Colonel Zhu Yangzhu and Professor Gui Haichao,
After safety checks outside, the astronauts
Jing said they were excited to “return to our motherland” and that the orbital stay inside the Tiangong space station was
8 . You might think a computer would be a fair judge, but a new study from MIT finds it might be better for you to leave yourself in the hands of humans. AI won’t let people off easily when they break the rule!
This study, conducted by a team at MIT, examined how AI would interpret perceived violations (违反) of a given code. They discovered the most effective data for AI training is normative data, where humans determine whether a rule has been violated. However, data used to train machine-learning models are typically labeled descriptively — meaning people label the factual features of a situation, and AI determines whether a code has been broken.
The team gathered images of dogs that could potentially violate an apartment rule banning aggressive animals. Labelers were then asked to provide normative and descriptive responses. The descriptive team wasn’t informed about the policy on dogs, and was asked to identify whether elements, such as the dog’s aggression, were present. Their responses helped to form judgments. If a user said the photo described an aggressive dog, the policy was considered to be violated. On the other hand, the normative group was informed about the rule and was asked to determine whether each image violated the rule.
Participants were 20 percent more likely to identify a violation using the descriptive method. If the descriptive data on dog behavior had been used to program an AI model, it would be more likely to issue severe punishment. These inaccuracies could be a real-world concern. For instance, if a descriptive model is used to predict whether a person may commit the same crime more than once, it may result in longer criminal sentences.
The researchers advocate for increased data transparency (透明性) to understand how data is collected. And their future work is to investigate the impact of having experts, such as lawyers and doctors, participate in data entry to see if it will lead to the same result. “Generally, normative decisions tend to be more lenient (宽容的),” says lead author Aparna Balagopalan. “Perhaps the way people think about violations differs from how they think about descriptive data.”
1. What can we learn about descriptive data?| A．They’re usually used to train AI. | B．They lead to more accurate judgment. |
| C．They consist of causes of a situation. | D．They’re gathered to identify rule violation. |
| A．To present a fact. | B．To highlight a problem. | C．To introduce a topic. | D．To oppose an opinion. |
| A．Use more normative data. | B．Seek advice on data entry from experts. |
| C．Find new ways to gather data. | D．Conduct studies with professional labelers. |
| A．AI: rule defender or breaker? | B．Data set: decisive part of AI |
| C．AI judge: stricter than humans | D．Better data: descriptive or normative? |
1. What do people need to rent a Nice Ride bike?
| A．An ID card. | B．A library card. | C．A membership card. |
| A．It actually increases their driving cost. |
| B．It lowers their environmental awareness. |
| C．It encourages them to travel more by bike. |
| A．Wuhan, China. | B．Paris, France. | C．Melbourne, Australia. |
| A．By providing a million bikes for Nice Ride. |
| B．By offering money to build greenways. |
| C．By awarding those who ride bikes. |
10 . With the WHO no longer considering COVID-19 a global health emergency, it seems that the virus and its large number of deaths will soon fade from memory, along with N95 masks and PCR tests.
Our ability to forget a pandemic is partly a coping mechanism, reflecting the emotional immune system that enables us to move on with our daily lives. Every day, we receive massive information that our brains struggle to keep. Moreover, the process by which our brains assess risk is deeply personal and influenced by our own needs.
We are reminded why so many people were eager to forget the 1918 influenza pandemic and embrace the joys of the Roaring Twenties. But collective forgetting threatens to leave us unprepared for future airborne disease outbreaks, forcing us to re-learn fundamental lessons about the importance of masking and shifting activities outdoors to prevent transmission.
People often have a more positive view of the future than the past as a way of building psychological resilience (韧性). This is because, unlike the unchangeable past, the future offers endless possibilities. But our tendency to look forward also arrests progress on issues like pandemic preparedness, as it leads us to believe we are better equipped than we are.
Today, there’s still no comprehensive effort to carry out the detailed recommendations on how to improve pandemic preparedness. Today, we still lack clear explanations for why our public-health agencies are understaffed and undefended, why supply chains failed, why COVID-19 misinformation was allowed to overflow on media platforms, and why our public-health responses remain passive.
It has taken Ireland more than 150 years to build memorials and museums marking the impact of the disaster — the Great Famine of the 1840s. We cannot afford to let the same happen with COVID-19. Our ability to remember the past could affect billions of lives in the future.
1. What is Paragraph 2 mainly about?| A．The pandemic coping system. | B．The struggle of data memory. |
| C．The reasons for forgetting a pandemic. | D．The personal pandemic assessment. |
| A．To outline the astonishing parallel found in history. |
| B．To reveal the harmful effect of collective forgetting. |
| C．To uncover people’s collective loss of memory. |
| D．To reflect people’s interest in future possibilities. |
| A．good preparations for the next pandemic have been made |
| B．people’s ability to cope with a pandemic is underestimated |
| C．the suggestions on pandemic prevention are well adopted |
| D．a list of questions concerning the pandemic remain to be answered |
| A．Remember the pandemic. | B．Embrace the pandemic-free life. |
| C．Invest in the pandemic research. | D．Build memorials to the pandemic. |
