The wave of artificial intelligence (AI) is impacting the banks of various industries, and even the scientific community, which has always been regarded as a "high-precision" fortress, is unable to stand alone in this transformation. So, which research positions are at risk of being swallowed up by the tide? The website "Nature" recently published a latest study that surveyed over 40 AI users in academia and industry. Many people admit that the rise of AI has significantly reduced the demand for people who write code and process basic data - jobs that were previously mostly undertaken by graduate students, postdoctoral fellows, or non professionals; Junior positions in fields such as computer modeling are also at risk, as AI's performance in such tasks far exceeds that of novice scientists; The living space for peripheral work such as translating scientific papers is also shrinking. As economist Anton Corinek from the University of Virginia has pointed out, work involving "pure cognitive tasks" will be the first to be affected, and such work closely related to scientific research may be quickly taken over by AI. The academic community generally believes that experimental positions that require hands-on operation and senior scientists who coordinate projects are currently in a "safe zone". But there are also researchers warning that even in these high threshold fields, AI is catching up and gradually approaching. Researchers who have a significant impact on modeling and data processing positions have long been accustomed to using AI to polish papers and organize literature. But the respondents unanimously believe that AI's ability in code generation and data processing has the most severe impact on the scientific job market. For example, some academic laboratories have specifically hired programmers to write research code packages. Regarding this, Brian Haye, a computational biologist at Stanford University in the United States, bluntly stated that with the emergence of AI, such work has become a "thing of the past". The positions that focus on creating simulations and analyzing data can now be filled by AI. The more profound impact is that even before triggering large-scale layoffs, AI has begun to suppress the birth of new jobs. Hannah Steele, a computational biologist at the University of Wisconsin Madison, lamented that if she had set up a laboratory five years ago and hired research programmers, it would have been the right thing to do; But now, AI is capable of handling heavy coding work, and this may no longer be necessary. Lu Nanshu, a materials engineer at the University of Texas at Austin, strongly agrees with this. She said that when recruiting graduate assistants and postdoctoral fellows, the team has become increasingly cautious, considering both financial uncertainty and the ability of AI to share some of the work. However, hidden worries also emerged. Scientists warn that if undergraduate, graduate, and technical personnel cannot receive sufficient training in the laboratory, it may have a long-term negative impact on the research community, as these positions are meant to be a ladder to higher scientific positions. Klaus Wilk, a computational biologist at the University of Texas at Austin, believes that although more output can be obtained at lower costs, the cost may be the rupture of the talent pool. There is evidence to suggest that AI has caused unemployment in certain scientific fields. With the popularity of AI translators, the membership of the Science and Technology Department of the American Translators Association has sharply decreased by 26% in less than two and a half years. Some translators are forced to transform. For example, Jaime Russell from North Carolina, who previously worked translating clinical trial documents, has now transitioned into a medical interpreter responsible for oral interpretation between doctors and patients. When it comes to her former colleagues, some even switched careers as food delivery riders, which made her sigh deeply. Despite this, most researchers believe that AI is still unable to handle the high-level tasks of scientists, such as determining which ideas are worth delving into. Jonathan Oppenheim, a quantum physicist at University College London in the UK, often conducts peer reviews on AI simulations. Although he believes that the criticism provided by AI is highly beneficial, he asserts that AI cannot truly provide novel insights. Even respondents who are optimistic about AI generated creativity believe that humans are indispensable. Computer scientist Kalu Sankarlingam from the University of Wisconsin Madison believes that the best way to conceptualize research directions is through human-machine collaboration, as generating hypotheses requires humans to design prompt words. Moreover, human participation can also prevent the drawbacks of AI "illusion", that is, the system's fictitious output. However, Corinek insists that even high-level research positions, if focused on the cognitive field, are vulnerable to the impact of AI. He predicted that mathematicians would be affected next year, but there are still objections in the mathematical community. Compared to the temporary safety of experimental personnel, laboratory technicians and early researchers engaged in "wet experiments" are currently in a relatively safe situation. AI and robot driven automation laboratories still struggle to complete many intricate tasks, let alone interpret complex results. Oppenheim stated that for a considerable period of time, AI is unlikely to have a significant impact on the work of experimenters. A recent study conducted by scientists at the University of Virginia in the United States is also somewhat reassuring. Despite the surge of AI technology, some positions remain strong. For example, although the AI tool "Alpha Fold 2" can handle many tedious tasks from inferring amino acid sequences to accurately predicting protein structures, research has found that labor-intensive protein structure imaging methods are still being used. Many protein AI still struggle to accurately identify and require manual analysis. This indicates that AI has not made scientists irrelevant, and they can instead solve problems where humans have a "comparative advantage". Researchers emphasize that this ability to adapt flexibly may be the future of science, and those who adapt to change will find vitality in the new era. (New Society)