Four myths about artificial intelligence
February 19, 2024 09:58 Source: "China Social Sciences", February 19, 2024, Issue 2834, Issue 2834

2023,Genesis artificial intelligence (AIGC) has set off a wave of global artificial intelligence technology development。big model products represented by "OpenAI) ChatGPT showed AI cross -scene、Multi -purpose、Intelligent universal task processing ability。Relying on massive data、Super strong algorithm、Combination of huge computing power,The big model has been analyzed、Decision、Understanding、Creation and other tasks to achieve performance that can be comparable to humans。Global Technology Corporation has launched their own artificial intelligence models under this boom,Chinese companies have also opened the "Hundred Model War"。"Nature" magazine among the top ten scientific figures in 2023 recently announced,The first non -human ChatGPT is included in it。In addition to these progress booms,The questioning and warning of artificial intelligence never stops。"Father of Deep Learning" Geoffrey Hinton shows the concerns of a series of risks that may generate AI,I believe that a large amount of false information produced by AIGC is full of network,AI will also disrupt the employment market,It may even become a killer robot with autonomous ability。"Simple Human History" author Uvar Herram also pointed out,Since artificial intelligence is mastering language with the ability to exceed the average level of human beings,Therefore, the key to invading the human civilization system,It may even lead to the end of the era of human rule。

In a broad sense,,AI refers to a social technology system created by computers bet365 Play online games and algorithms,The purpose is to imitate、Enhance or replace humans。As the above boom and question coexist,AI has no conclusion on the society。Although AI practitioners, including technology companies and designers, have proposed the slogan of "AI to good",But there are still many opinions about what is "good society"。From the perspective of social science, contribute to a path of this discussion,The principles of promoting the design and governance of AI product development have shifted from "personal -centric" to "society -centered"。Before this,Clarifying people's four myths for AI are useful attempts。

The first myth is AI seemingly neutral,In fact, there are partial。AI technology is increasingly used for decision -making,The result obtained by the algorithm may determine who can get a job,Who can get social welfare assistance,Who is more likely to be arrested。A hypothesis of using AI for decision,Total than people is fair。But,Sociologists have already discovered the "prejudice" in AI design and applications。The "prejudice" of artificial intelligence comes from data and algorithms。The output of machine learning is strongly dependent on input data,These data include existing data from various social institutions,also includes daily data generated in real time。No matter what kind of,The data itself is the texture that deeply embeds social inequality,and when these data are used to train big models again,will strengthen existing inequality。Already found out,Resume pool generated in the background of gender inequality in the labor market,After being learned by the machine,Applicants with output will be significantly not conducive to women。Bet365 lotto review Dagai model trained by data based on the history of arrest,It is also more inclined to define sex for people as high -risk crimes。Even if the data itself has no bias,Design of the algorithm、Interface information presentation, etc., will also exacerbate your own prejudice。Current,The concerns about AI fairness have spawned all kinds of measures to solve such problems,Including AI Ethics Alliance、Open source AI fair tool package、Government's laws and regulations, etc.。These measures are implemented more effective,Undoubtedly more need to combine more research results with social structure and inequality mechanisms in social sciences,Back off with the current shallow crowd classification thinking,Forms a more effective mechanism -dominated intervention framework。

The second myth is AI seems accurate,In fact, there are errors。Generation AI is widely used in generating text、Image、Sound and video and other content。and the content generated by AI participation and even dominated by is increasingly appearing in the media、Business and even academic platforms。Restrictions from the technical itself and the cost of computing power,Make the content of AI generated not as accurate as it looks like,Error information often appears in a very credible form,Confusion about the people who lack the discern,This phenomenon is called AI Hallucination。Related research team found that the hallucination rate of the main generated AI platform on the market is between 3%and 27%。If the content of AI generates is only used as a private purpose,The influence of information error may not be great。But once these contents are used in public domains such bet365 live casino games as court disconnection、Business promotion、During the diagnosis of the disease,will have serious consequences。For example,Some voters are affected by information provided by AI,I believe that candidates have a bribery case,Give up to support it; forgery case library generated by AI misleads the judge's judgment; more false medical research by AI is,It affects the effectiveness of the treatment plan, etc.。Aware that AI may have the impact of "confident nonsense",At present, developers have formed some response plans,For example, use more accurate reminder words、Integrated into the human supervision mechanism、Adopt retrieval enhancement generation method、Make AI generating content, etc.。But,In addition to these technical changes that are designed to strengthen AI accuracy,We also need to deal with the social impact caused by the error of AI information,Improve audience screening information、Verify the ability of information,Establish relevant regulations,Clarifying the responsible subject of social consequences to generate information brings information。

The third myth is AI seemingly all -around,In fact, limited。From decision -making AI to generating formula AI,Machines can not only help humans realize autonomous driving、tasks in professional fields such as face recognition and intelligent decision -making,It also evolved to the stage of being able to be competent for the new content of independent creation。Automation process replaces duplicate work,and in the wave of intelligence, people are worried that the omnipotent AI will fully replace human labor。Related reports found that the use of large language model tools will affect 80 % of the employment bet365 live casino games in the US labor market,Especially high -paying groups。In the discussion of the relationship between AI and human labor,A core controversy is alternative (replacement) or augmentation。On the one hand,AI is indeed good at completing the task that can be converted into a clear parameter,Such as natural language translation、Text writing with a clear style、Data analysis and summary、Modeling user service, but on the other hand,AI's limitations are also very obvious,Tacit Knowledge、Complex reasoning、Emotional perception、The ability in the high professional task of social interaction is still stretched。Existing research show,In the field of judicial,AI's risk assessment tool is not much better than humans。AI while replacing some jobs,will generate other new jobs,Such as prompt engineer、Robot Application Technician、AI Engineering Technician、AI trainer, etc.。Incorporating the skills needed to these emerging occupations into the design of the vocational education framework will be a very useful measure。Therefore,We don’t need to worry about "completely replace by AI",Instead, we should actively plan human cooperation with AI,Clarify the "energy" and "cannot",Guide it to increase energy for human labor,To promote the overall improvement of production efficiency。

The fourth myth is AI seemingly inclusive,In fact, it is not equal。Some people call AI as a general technology innovation,Its promotion of human productivity can even be with steam engines、Electricity and computer comparable。We are rejoicing while we are rejoicing for AI's huge productivity potential,You need to see the possibility of bet365 best casino games the AI ​​continuation or even deepening the existing unequal structure。These inequality is reflected between regions、The labor between the company and the different groups。The development and application of AI require huge capital and technical investment,Therefore, it is often concentrated in some countries and regions,Show the inequality of new technologies in space。other,Whether it is the basic raw material for the predictive task of the big model -data acquisition,Development of algorithms,Still supply of calculations,All companies with the advantages of digital basic platforms are more likely to participate in AI's research and development and application。and this will cause companies to market share、Economic and political influence generates polarization。AI technology application on the labor market is not uniformized,Digital control for workers for low -skilled positions,The rise of AI -related positions with special technologies often occurs synchronously,expanded the existing inequality in the traditional labor market。The above inequality or even existed before the AI ​​technology,Social benchmark facts outside the big model。and social science research helps people understand how these inequality is formed,to promote a more fair AI application。

AI technology was proposed in the 1950s,Now it is quickly widely used in many areas of social life,Gradually and deeply participate in social life。But,AI's impact on society is still a field to be explored。This article sorts out four myths about AI,Calling Social Science's more in -depth research on artificial intelligence。In the era of intelligence,Social scientists should bet365 Play online games actively invest in this debate on new technologies and society,Research on social inequality and social structure changes for a long time,Promote artificial intelligence from "people -centered" to "social -centric" design and service principles.。

bet365 live casino games

Editor in charge: Zhang Jing
QR code icons 2.jpg
Key recommendation
The latest article
Graphics
bet365 live casino games

Friendship link:

Website filing number: Jinggong.com Anmi 11010502030146 Ministry of Industry and Information Technology:

All rights reserved by China Social Sciences Magazine shall not be reprinted and used without permission

General Editor Email: zzszbj@126.com This website contact information: 010-85886809 Address: Building 1, Building 1, No. 15, Guanghua Road, Chaoyang District, Beijing: 100026