Comprehensive foreign media reports from the official website of Stanford University on February 27，A study by Professor Robb Willer, a professor of sociology at the school, found out，A new generation of artificial intelligence has the ability to generate political persuasive information，This has raised new problems for artificial intelligence supervision。

With the rise of deep learning models trained by oversized large -scale corpus libraries，Artificial intelligence begins to handle many high -level cognitive tasks。Viller and others have inspected the third-generation generative bet365 live casino games pre-training converter (GPT-3) through three experiments.。So far，The views of others are regarded as a unique social behavior of human beings。GPT-3 is one of the most advanced artificial intelligence models，ChatGPT is also developed on the basis of GPT-3。

Researchers asked GPT-3 to generate convincing texts around several controversial political topics，Then mix the generated works with the text created by non -professionals under the same topic，Randomly allocated to 4,836 experimental participants (human) from the United States (human) reading。Participants report the views on related topics before reading the text，so that researchers can measure and compare the persuasiveness of artificial intelligence works and human works。

All experiments are displayed，bet365 Play online games The content generated by GPT-3 has a certain impact on the participants。Although the persuasion is not significant，But in terms of order of order, it is equivalent to human works。This indicates that the persuasion ability of artificial intelligence has been caught up with ordinary people to a certain extent，Personal ability is a key measurement criterion for artificial intelligence people's performance (Human-Like Performance)。Another，Participants generally believe that the work of GPT-3 is more sufficient，Angry and uniqueness is weaker，Human works are more focused on experience and story，Vivid image。In firm confidence、Smart、Moral、Creative、Interesting、Sympathy and other dimensions，There is no obvious difference between the participants' scores of artificial bet365 Play online games intelligence authors and human authors。

Weille and others reminds，This study shows that people should consider supervising the application of artificial intelligence in political activities as soon as possible，Preventive artificial intelligence text generation tools are used in a large number、persuasive information of policy makers and legislators。If the content is inaccurate or misleading，It may cause unreal information to spread widely，Obstructing people's views on politicalization issues and events。"Obvious，The advanced level of artificial intelligence has brought about interest policies and legal issues。Artificial intelligence has the potential to affect political discourse，We should deal with these problems from the beginning。"Willer said。

For the above risks，Weille proposed several preventive bet365 live casino games measures: requiring artificial intelligence chat robots to clearly identify identity during conversation and writing; add identifier to the text generated by artificial intelligence; set up，Make artificial intelligence refuse to produce the existence of position tendency or lack of reason。"Exploring the feasibility and effectiveness of various regulatory measures in future research is very important。"Researchers said。

　　（bet365 live casino games ）