In order to find out what happened, the clips of hao tong moon and vagabond were uploaded to an ai-rate testing system of a commonly used thesis, showing that the overall suspicion of ai generation in both works was 62. 8 per cent and 52. 88 per cent, followed by a 100 per cent ai rate in king tung's order and a web-broadcasting: “recommends the abolition of ‘class degrees’ in wang bulgas.”
As a result, a growing number of university students have begun to “speak out”. Qin seong (alias) of the western school of higher education in political and legal sciences said that even his college, class, school number and name were determined to be ai when the ai test was conducted, “i don't know what it means”. A number of interviewees questioned: “ai is a human-created, imitating tool, and now people are being driven by ai's rate
Journalists have learned that not all universities are currently explicitly testing the ai rate, but anxiety is spreading. Journalists viewed the major social media platforms and found that many graduates were concerned about the impact of graduation and job-seeking, and that, under a topic called “damage of dissertations”, the thesis's ai rate stimulated a lively discussion of the negative experiences of graduation, with some teachers and non-graders joining the ranks。
With the growing pressure to “fall down the ai rate”, the “fall down the ai business opportunity” was created. Journalists' surveys have found that advertisements from major platforms that spend “low the ai rate” are everywhere, with prices ranging from tens to hundreds more, and that some traders take money and do not fall to promised figures, and are asked to say “black off without money”。
The journalist, who spoke on a random basis about an "ai-rate drop" service owner, said that a paper with 30,000 words and an ai-rate of 45 per cent could be reduced to a single digits at a cost of $150 by “pure manual adjustment” at a cost of 50 per cent。
Is it necessary to measure the ai rate? What's the current test technique
Professors at a university of political and legal law in the west noted that the integration of artificial intelligence into education was a necessary trend, but that it was important to guard against academic fraud resulting from the misuse of ai and to use ai technology for “guidelines”. Journalists have learned that as artificial intelligence technologies become more “smart” in terms of material integration, writing and so on, they become increasingly involved in learning and education, and in order to avoid “ai dependency” among students and to interfere with the results of the study of graduation papers, a number of universities have begun to regulate the scale of participation of ai tools in academic practice。
For example, the university of jordan has sent a letter detailing the requirements and penalties for students using ai tools, such as the prohibition of the use of ai tools for research programme design, innovative methodology design, etc. The beijing teacher training university requires students to mark the content when they complete their work using the generating ai, and cannot directly generate more than 20 per cent of the full text. The technological university of tianjin provides that, if the undergraduate thesis rate is over 40 per cent, the institute will issue a warning to the students to check themselves and then re-examine the relevant platform for testing。
Journalists have found that, as a result of the technology and standards still being explored, the treatment of the thesis's ai overrated is now being treated with caution, more in the direction of modification than in the direct equivalent of typification and theft。
So, what's the current ai test model? A number of experts pointed out that, first, “ai generation” was not an “ai original”, but rather an “product” of the synthesis of ai, based on human creations and imitations of human thought expression. Inverse identification of the ai paper is essentially a probability measure based on probability and does not conclude that “the paper was produced by ai or assisted by ai”。
At the same time, the existing ai-rate testing platform has different judgement criteria and a deviation from the test method, and the isp, wip-ai-rate testing results are generally lower than for platforms such as papersmart and papersmart. Journalists tested a graduation paper and found that 50 per cent and 70 per cent of ais were available on papersmart and 70 per cent on papersmart and only 5 per cent on the know-how testing system。
“this year, we introduced the aig (contents of artificial intelligence) testing system, which, according to engineering, although agc tests can provide useful advice in terms of the structural integrity of the papers, it also exposes the problem of procedural expressions and objective descriptions that can be miscalculated as agc.” according to a professor at a university of political law in the west。
Ro yangyang, associate professor at the institute of higher education of the university of lanzhou, pointed out that the testing of ai rates and reproduction rates are essentially linguistic tests, not only avoiding tests through language modification, leaving the li gyu unpunished, but also at risk of miscalculation and wrong treatment. Innovation can only be protected and promoted by shifting from a “check-and-check” perspective to an evaluation of outcome content。
"know ai for good," explore "good practices" for ai to empower
Experts pointed out that the ai tool has become an important aid for learning, and that it is important for universities to avoid being “ai-only” and not to allow tools to prevail over human thinking and judgement, while at the same time, there is an urgent need to optimize detection techniques, increase the flexibility of ai-rate testing for participation in academic norms and step-by-step exploring ai-enabling education。
One of the professors interviewed noted that academic papers contained a great deal of professional terminology and logical reasoning and that it was extremely important to test the applicability of models. It is recommended that the relevant departments, institutions of higher learning and scientific research institutions work together to develop models of natural language processing algorithms and machine learning suitable for academic thesis, and to broaden the coverage of the database so as to provide detailed and accurate analysis of the language style, semantic structure, etc. Specific to the text。
According to the professor, the paper measuring the different criteria with a “fixed rule” was too “one-size-fits-all” and suggested that higher education institutions should select the ai-rate testing tool according to their own discipline characteristics, while exploring the setting of an ai-rate threshold range applicable to academic papers, setting different percentages of ai-rate limits for the characteristics of different disciplines and the type of paper, increasing the flexibility of ai-rate testing to participate in academic missions and allowing students to no longer passively “slash fit” down ai rates。
It is also particularly important to raise the “support item” behind the consolidation of ai rates. According to guo shao qing, secretary-general of the group of experts, who was guided by the pilot work of the ministry of education's artificial intelligence-assisting teacher building initiative, the sources of language for the development of the ai-wide mega-model growing are diverse and difficult to discern, and are vulnerable to deviations. To improve the accuracy and ease of reference for the testing of ai rates, in addition to the positive model, it is necessary to monitor and monitor the language base behind it, and to recommend that relevant departments, together with universities, explore the development of basic, common sense and authoritative databases of knowledge and information。
Indeed, the public debate on the “ai rate” is behind the expectation of a new model of higher education in the age of artificial intelligence. Shincheng yu, professor at beijing teacher training university and executive director of the centre for advanced and advanced innovation in future education, pointed out that the direction in which smart technology is applied in teaching and learning is a tool that supports students ' cognitive expansion. It is recommended that the field of higher education explore a progeny model adapted to the information age, with a shift from an individual to a distributed perception of human beings, a multi-state, multi-dimensional and cross-cutting curriculum, and a development-oriented assessment model。
Journalists have learned that some institutions of higher learning have incorporated artificial intelligence into the teaching system, such as the university of jodm, which plans to introduce “ai major courses” to achieve “ai courses covering all students of the institute”. In higher education institutions such as the university of lanzhou, the teaching model of artificial intelligence+ courses is gradually being applied。
Ro yangyang noted that the application of ai technology in the field of higher education in our country has just begun and, while strengthening norms and guidance at the institutional and infrastructure levels, recommends the development of a more pluralistic ai curriculum, along with the upgrading of teacher-student ai and technical ethics education. The test of the ai rate is gender neutral, rational, optimal in its use and perfect in its exploration。
