Source: Science and Technology Daily, CCTV News, China Science News, Global Science, etc
The "leopard" core sample may be new evidence of ancient microorganisms on Mars
Source: CCTV News According to the official website of the United States Space Agency, the United States "Perseverance" rover recently collected a unique red core sample with "leopard prints" near the Jerzero crater on Mars, which is the 22nd core sample collected by the "Perseverance" and was named "Cheava Falls" by the scientific team. According to its members, on Earth, these speckled features on rocks are often associated with the microbial fossil record in the subsurface. The hematite contained in the continental sedimentary rocks participates in the chemical reaction, which will cause white spots on the red rocks, and the iron and phosphate released by the reaction will form a black halo around the white spots, thus forming "leopard" spots. This reaction can serve as an energy source for a class of microorganisms that rely on chemical energy, and the presence of the above features on rocks suggests a possible association with microorganisms. Therefore, their preliminary analysis suggests that this "leopard print" may indicate the presence of microorganisms in ancient times on Mars. The scientific team is still considering other possible causes of the spot, but to reach a definitive conclusion, rock samples need to be brought back to Earth for study with laboratory instruments.
https://news.cctv.com/2024/07/27/ARTIMyMLgHjuuxvHHCf9BY8H240727.shtml
The first high-resolution map of the bottom of the Antarctic ice shelf
来源：Filip Stedt/University of Gothenburg
The Dosen Ice Shelf, as part of the West Antarctic Ice Sheet, maintains the stability of the West Antarctic Ice Sheet. Studies have predicted that if the West Antarctic ice sheet collapses, it could lead to a 3.2-metre rise in mean sea level. A study published ·in Science Advances recently described the shape of the base of Antarctica's West Antarctic Ice Sheet's Dotson Ice Shelf in detail for the first time, according to Nature news.
Using a robotic submersible equipped with sonar, the research team measured the topography at the bottom of the Dosen Ice Shelf, as well as the temperature and salinity of the seawater beneath the ice shelf. Unexpectedly, the research team found 75 teardrop-shaped structures at the bottom of the ice shelf, ranging in length from 20 to 300 meters and an average depth of 14 meters, which exist only on the western side of the ice shelf. The researchers speculate that friction between the bottom of the ice shelf and the seafloor may have caused the water to spiral, which accelerated the local melting of the ice shelf and formed these "teardrops". But more observational evidence is needed to confirm this hypothesis. Earlier this year, the research team redeployed the submersible to conduct further investigations of the Dosen Ice Shelf and the nearby Thwaites Glacier. But near the end of the mission, they lost contact with the submersible (valued at $3.8 million). The research team says that there is a great deal of risk in transporting expensive equipment under hundreds of meters of ice, which is why this data is valuable and rare. https://news.sciencenet.cn/htmlnews/2024/8/527478.shtm
OpenAI推出AI搜索引擎SearchGPT
According to the United Kingdom "Guardian", United States Consumer News and Business Channel and other media reports, on July 25, local time, OpenAI announced that it is testing a new artificial intelligence (AI) search tool SearchGPT, which can access information from the Internet in real time, aiming to provide users with more timely and accurate information. The search engine starts with a large text box that asks the user "What are you looking for?" "Instead of returning a simple list of links after a customer enters a search, SearchGPT tries to organize and understand them.
OpenAI says SearchGPT will be rolled out first among a small group of users and publishers, and then potentially rolled out more widely. The company eventually intends to integrate this search feature into ChatGPT rather than making it a standalone product. According to the report, this could have an impact on Google-led engine searches. https://www.cankaoxiaoxi.com/#/detailsPage/kejiyy/5945e321005d45b0a22e3b134ac605d4/1/2024-07-31%2016:41?childrenAlias=kejiyy
Our looks will become more and more in line with our names
Source: PNAS (2024) The name is a social label for everyone. A new study published in the Proceedings of the United States Academy of Sciences (PNAS) has found that a person's appearance may become more and more consistent with their name. The researchers asked adults and children aged 9 to 10 to match faces and names, and found that both children and adults were significantly more likely to correctly match adult faces to their corresponding names than by chance. But for children's faces and names, these participants couldn't be accurately matched. In addition, the researchers fed a large database of facial images into a machine learning system. It was found that the face representations of adults with the same name were significantly more similar than those of adults with different names, but no similar similarities were found between the faces of children with the same name. Researchers believe that this may be because names give a self-fulfilling prophecy that causes a person's facial appearance to change over time to conform to the social stereotypes associated with names.
https://phys.org/news/2024-07-reveals-people-evolve.html
Avian influenza viruses have been shown to be effective and sustained among mammals
Since 2022, H5N1 avian influenza has caused serious harm to wild birds and poultry around the world. Since April this year, the number of human cases of H5N1 in the United States has continued to rise, and 11 cases have been reported so far. A recent study published in Nature showed that the virus, which originates in birds, is also effective and sustained among mammals.
Using whole-genome sequencing, modeling and epidemiological information, the researchers determined that infected cows in Texas in the United States transmitted the H5N1 virus to healthy cows in the state after moving to the state. Sequencing results showed that the virus also infected cats, raccoons and wild birds near farms. Studies have shown that the virus has a high tendency to infect the mammary glands and that the milk of infected animals contains a large amount of infectious virus. Notably, the H5N1 strain that has been infected in humans in recent cases is the same strain circulating in dairy cows, so researchers suspect that the virus originated from a dairy farm in the same county. It is important to note that although the H5N1 virus is capable of infecting and replicating in humans, it is less efficient at infecting; Whole-genome sequencing also did not reveal any mutations in the virus that cause increased transmission of H5N1 in humans. However, the researchers note that in the future the virus should be wary of adaptation to mammals and efficient transmission in humans, so continuous monitoring of the virus in infected animals and potentially human infected individuals is critical. https://www.eurekalert.org/news-releases/1052083
Specialized optical fibers for quantum technology applications are introduced
The dwarf planet Tronis has rare rings
Source: Cameron · McCarrie/University of Bath, United Kingdom
Physicists at the University of Bath in the United Kingdom have developed a new generation of specialty optical fibers to meet the data transmission challenges that will arise in the future era of quantum computing. Specialty fibers enable quantum computing at the node itself by acting as an entangled single-photon source, quantum wavelength converter, low-loss switch, or quantum memory container. At the same time, special fibers can be directly integrated into the network, which greatly extends the operating distance. This achievement is expected to promote the expansion of large-scale quantum networks. The research results were published in the new issue of Applied Physics Letters · Quantum.
http://digitalpaper.stdaily.com/http_www.kjrb.com/kjrb/html/2024-08/01/content_575490.htm?div=-1
Drinking alcohol in moderation is not beneficial for longevity
A study led by the University of Victoria in Canada found that moderate alcohol consumption did not extend human lifespan. Many previous studies have suggested that moderate drinkers live longer and have a lower risk of heart disease and other chronic diseases than quit drinkers. This has led to a widespread belief that alcohol in moderation can be a health tonic. Here, the researchers identified and included 107 longitudinal studies (covering data from 4838825 participants, including 425564 deaths), setting a series of stringent quality criteria, such as a younger mean age of the study cohort (55 years or younger), long follow-up (more than 55 years), and effective differentiation between former and occasional drinkers and complete abstinent drinkers in the analysis. The results showed that when older people who abstained from alcohol or tapered due to health problems were excluded, the mortality risk of low-volume drinkers was similar to that of those who were completely abstinent, and did not show significant health benefits; When distinguishing between former and abstaining drinkers, there was no significant association between moderate alcohol consumption and longevity. The researchers emphasize that there is no safe level of alcohol consumption, and moderate alcohol consumption may also increase the risk of cancer. This finding has important implications for the global burden of disease estimation and the development of drinking guidelines. The paper was published in the Journal of Studies on Alcohol and Drugs some time ago.
https://mp.weixinbridge.com/mp/wapredirect?url=https%3A%2F%2Fwww.eurekalert.org%2Fnews-releases%2F1051741&action=appmsg_redirect&uin=MTU0MDM5NTEwMg%3D%3D&biz=MzA5NDkzNjIwMg==&mid=2651765384&idx=1&type=0&scene=0
Japanese company develops "power generation glass"
According to a report by Kyodo News on July 26, Japan's Panasonic Holdings Group announced on the 25th that it will realize the practical application of the next-generation technology "perovskite solar cell" that has attracted much attention in 2026, and the battery will be put on the market as a "power generation glass" integrated with glass building materials. The company's previous plan was to put it into service by 2028.
https://www.cankaoxiaoxi.com/#/detailsPage/kejiyy/7469614a2e6f4b22a38b5ff7e63837ed/1/2024-07-29%2016:18?childrenAlias=kejiyy
Engineered parasites can deliver therapeutic proteins
来源：Nature Microbiology
Biomolecules can be used for therapeutics or as tools to study biological processes, but delivery to target cells and tissues is difficult. Toxoplasma gondii is a parasite that naturally travels from the human intestine into the central nervous system. Past studies have shown that it is capable of delivering proteins to host cells. Nature ·Microbiology reports a method to engineer Toxoplasma gondii in a mouse model to cross the blood-brain barrier to deliver therapeutic proteins to host neurons.
Researchers used Toxoplasma gondii's two secretory organelles (specialized structures for intracellular executive functions) – the rhoptry and dense granules – to deliver proteins to host cells. They selected proteins located in the organelles of Toxoplasma gondii and fused them with different proteins known to treat neurological diseases in humans. The results show that proteins can be delivered from both organelles to neurons at the same time. In addition, the study demonstrated that a therapeutic protein, MeCP2, can be delivered to neurons, bind to target DNA, and alter host gene expression within cells, neurons, and brain organoids, with very few Toxoplasma gondii detected outside the target delivery site and no significant inflammation after delivery. The authors conclude that while these findings could provide new approaches for therapeutic protein delivery, further research is needed to understand potential limitations, including efficacy and safety.
https://www.nature.com/articles/s41564-024-01750-6
Training AI with AI may become more and more "outrageous" the more you practice
Source: Nature Recently, a study published in Nature showed that training an AI model with text generated by artificial intelligence (AI) will quickly cause the model to output meaningless content, a phenomenon known as "model collapse". Through mathematical analysis, the research team showed that model crashes may be pervasive and exist in language models of all sizes that use uncurated data, as well as simple image generators and other types of AI. To demonstrate the model crash, the researchers took a pre-trained large language model and fine-tuned it using a dataset based on Wikipedia entries. They then asked the model to generate their own Wikipedia-style articles, and used those articles to train the same pre-trained large language model to get the next generation of the model. To judge the performance of each model, the researchers provide an opening paragraph and ask the model to predict the next few sentences, comparing its output with the output of a model trained on real-world data. It was found that the model may ignore some of the outputs in the training data (such as less common text), causing it to self-train with only a portion of the dataset, weakening the learning ability of the descendant models, and eventually leading to a complete crash. The reason for this is that each model does not learn from reality, but from the predictions of the previous model, and errors are amplified in each iteration. Over time, these errors stack on top of each other until the model essentially learns nothing else.
https://www.nature.com/articles/s41586-024-07566-y
☟
Founded in 1980, Science & Technology Review is the academic journal of the China Association for Science and Technology, which mainly publishes reports on breakthrough achievements in scientific frontiers and technology hotspots, authoritative scientific reviews, and leading high-end reviews, and publishes decision-making and consulting suggestions for promoting economic and social development, improving scientific and technological management, optimizing the scientific research environment, cultivating scientific culture, and promoting scientific and technological innovation and the transformation of scientific and technological achievements. The permanent columns include the preface of academicians, think tank views, science and technology reviews, hot topics, reviews, papers, academic focus, science and humanities, etc.
Founded in 2014, the WeChat public platform of Science and Technology Review mainly publishes the main points of the content of the journal Science and Technology Review, reports hot scientific and technological issues, scientific and technological events, and scientific figures, and creates a new media platform that is closely related to paper journals and has distinctive characteristics. The official account of Science and Technology Review has gathered tens of thousands of future stars and academic experts who are dedicated to academics, adding and editing WeChat, so that excellent you have the opportunity to meet people with the same interests.