人脸识别技术的算法偏差及其多元治理

Algorithm Bias and Multivariate Governance of Face Recognition Technology

人脸识别技术的快速发展让人们的生活变得越来越便利,但是人脸识别技术中存在着算法偏差,这种算法偏差所引起的技术缺陷、歧视现象、隐私泄露等问题十分隐蔽且不易察觉。其原因主要在于算法设计者对主体观念的继承、数据录入偏差、企业利益诉求以及算法自身的不可预测性。要消除算法偏差所产生的伦理和社会问题需要强化伦理道德建设、成立伦理委员会、建立和完善算法立法、推进算法的优化升级,为人脸识别技术的健康发展创造良好的环境。

The rapid development of face recognition technology makes people’s life more and more convenient,but there are algorithm deviations in face recognition technology,and the technical defects,discrimination,privacy leakage and other problems caused by this algorithm deviation are very hidden and difficult to detect.The main reasons lie in the inheritance of the main idea of the algorithm designer,the deviation of data entry,the interest demands of the enterprise and the unpredictability of the algorithm itself.In order to eliminate the ethical and social problems caused by the algorithm deviation,it is necessary to strengthen the construction of ethics and morality,establish an ethics committee,establish and improve the algorithm legislation,promote the optimization and upgrading of the algorithm,and create a good environment for the healthy development of face recognition technology.