Risks,Principles and Responsibilities:Research on the Construction of the Ethical Norm Framework of Artificial Intelligence Social Experiment Based on the Experimental Path
The Artificial Intelligence Social Experiment(AISE)investigates the broader social impact of AI technolo-gy.It employs social experimentation as a scientific method of research,with the goal of anticipating AI hazards,seeking solutions in advance,and promoting the benign growth of AI technology.AISEs must be carried out in real-life social cir-cumstances and will surely have deep ethical consequences for both the participants and the social environment within the experimental areas.As a result,there is a pressing need to build a practical and effective ethical norm framework.Al-though some studies have investigated relevant ethical risks and principles,scholars and practitioners remain perplexed and divided about how to make these abstract concepts empirically understandable and useful,that is,to"translate"them into practices to guide the actual conduct of AISE.Meanwhile,the unpredictability of AI technology makes it more diffi-cult to forecast and prevent ethical problems,as well as to assign responsibility in AISEs.In this setting,a more detailed explanation of how to identify ethical risks,apply ethical principles,and define ethical obligations in practice is required.Fortunately,the systermatic settings of AISE equip researchers with the tools and approaches to tackle these difficult topics.By sorting out the basic protocol of a standard AISE,it can be summarized that the typical AISE process consists of seven stages:build experimental scenarios,clarify experimental methods,confirm experimental objects,set observed variables,organize experimental implementation,analyze experimental data,and provide feedback on experimental results.Based on the specific stages of experiments,the ethical risks,principles,and responsibilities may be appropriately as-sessed and integrated,resulting in the construction of an AISE ethical norm system that can not only convey the value con-sensus in general,but also have strong practical operability.With this approach,the study of ethical risks may be merged based on the risks of AI technology and social experi-mentation at each experimental stage,which awaits enrichment and further interpretation based on diverse application sce-narios paired with various AI features.Following the identification of ethical risks,the ethical principles established along the experimental path can directly target the ethical risks inside each stage,maximizing normative efficacy.By organizing the most prevalent ethical principles in AI ethics and social experiment ethics,the overlapping chosen principles of benefi-cence、autonomy、non-maleficence、justice and transparency are shown.Each principle informs different normative criteria and exerts varied normative effects at different phases,such comprehensive refinement works as detailed operating direc-tion for all AISE participants.Furthermore,according to AISE regulations,the key participants in this event can be classi-fied as application subjects,research subjects,technical subjects,and review subjects.Each participant has their own dis-tinct tasks to contribute,which maintains their allocated obligation as well as accountability towards specific elements of the experiments.In this way,a complete comprehension of AISE's ethical norm framework can be acquired and illuminated.The ethi-cal norm system based on the experimental path is not only simple to grasp and reflect the ethical qualities of AISE,but it is also simple to use,which aids in the resolution of practical ethical concerns,the avoidance of ethical harms,the applica-tion of ethical principles,and the definition of ethical obligations.
artificial intelligence social experimentexperimental pathethical risksethical principlesethical respon-sibility
NETL
NSTL
万方数据
