Get started
Work

Artificial Intelligence, Bias, and Ethics

Year: 2023
Type: conference-paper
Abstract: Although ChatGPT attempts to mitigate bias, when instructed to translate the gender-neutral Turkish sentences “O bir doktor. O bir hemşire” to English, the outcome is biased: “He is a doctor. She is a nurse.” In 2016, we have demonstrated that language representations trained via unsupervised learning automatically embed implicit biases documented in social cognition through the statistical regularities in language corpora. Evaluating embedding associations in language, vision, and multi-modal language-vision models reveals that large-scale sociocultural data is a source of implicit human biases regarding gender, race or ethnicity, skin color, ability, age, sexuality, religion, social class, and intersectional associations. The study of gender bias in language, vision, language-vision, and generative AI has highlighted the sexualization of women and girls in AI, while easily accessible generative AI models such as text-to-image generators amplify bias at scale. As AI increasingly automates tasks that determine life’s outcomes and opportunities, the ethics of AI bias has significant implications for human cognition, society, justice, and the future of AI. Thus, it is necessary to advance our understanding of the depth, prevalence, and complexities of bias in AI to mitigate it both in machines and society.
DOI: 10.24963/ijcai.2023/799
Language: English
Subfield: Safety Research
Open Access status: bronze
Award: EXPLOITING ALTERNATE COMPUTING TECHNOLOGIES II THEISS RESEARCH PURPOSE: THIS PROPOSAL SEEKS TO MAKE ADVANCES IN TWO SEPARATE AREAS OF EMERGING COMPUTING TECHNOLOGY: QUANTUM INFORMATION SCIENCE AND NEUROMORPHIC COMPUTING. FOR THE FORMER, WORK WILL FOCUS ON THE DEVELOPMENT AND ASSESSMENT OF QUANTUM COMMUNICATION SYSTEMS AND APPLICATIONS. FOR THE LATTER, THE GOAL IS TO IDENTIFY PROMISING APPLICATIONS AND DEVELOP TRAINING METHODS FOR SPIKING NEURAL NETWORKS, AN ALTERNATIVE COMPUTATIONAL SYSTEM LOOSELY INSPIRED BY OUR UNDERSTANDING OF HOW THE HUMAN BRAIN WORKS ACTIVITIES TO BE PERFORMED: COLLABORATION WITH THE EUROPEAN TELECOMMUNICATIONS STANDARDS INSTITUTE WILL BE UNDERTAKEN TO HELP CLARIFY THE VITAL LINKAGE BETWEEN SECURITY PROOFS FOR IDEALIZED QUANTUM CRYPTOGRAPHIC KEY DISTRIBUTION (QKD) SYSTEMS AND THE CHALLENGES OF REAL SYSTEMS BUILD FROM IMPERFECT COMPONENTS. THE PI WILL ALSO APPLY HIS VAST EXPERIENCE IN DEVELOPING QKD HARDWARE TO THE DEVELOPMENT OF AN EXPERIMENTAL QUANTUM NETWORK IN COLLABORATION WITH NIST SCIENTISTS. FINALLY, IMPROVED METHODS TO TRAIN ARTIFICIAL NEURAL NETWORKS (NN) WILL BE STUDIED, AND THE IMPLEMENTATION OF NN APPLICATIONS AND BENCHMARKS THAT MAP WELL TO SCALABLE, LOW ENERGY NN PLATFORMS SUCH AS THE SUPERCONDUCTING OPTOELECTRONICS NETWORK (SOEN) UNDER STUDY AT NIST WILL BE DEVELOPED. EXPECTED OUTCOMES: THE FUNDAMENTAL RESEARCH SUPPORTED BY THIS GRANT WILL CONTRIBUTE TO THE BODY OF KNOWLEDGE NEEDED FOR THE DEVELOPMENT OF PRACTICAL AND RELIABLE SYSTEMS AND APPLICATIONS OF QUANTUM NETWORKS AND SPIKING NEURAL NETWORKS. THE RESULTS OF THIS WORK WILL BE PUBLISHED IN THE OPEN SCIENTIFIC LITERATURE. INTENDED BENEFICIARIES: THE WORK PROPOSED HERE WILL PROVIDE THE NIST WITH A DEEPER UNDERSTANDING OF THE STATE-OF-THE-ART IN POTENTIAL NEW COMPUTING TECHNOLOGIES, ALLOWING IT TO ANTICIPATE NEEDED MEASUREMENT TECHNIQUES AND TOOLS TO ENABLE DOWNSTREAM COMMERCIALIZATION. WHEN FULLY REALIZED, QUANTUM NETWORKS COULD PROVIDE NEW MEANS OF SECURE INFORMATION EXCHANGE, A WAY TO SCALE CURRENT TODAY?S SMALL QUANTUM COMPONENTS INTO A LARGE, DISTRIBUTED COMPUTING SYSTEM, AS WELL AS TO DEVELOP NEW HIGHLY SENSITIVE SENSING CAPABILITIES. NEUROMORPHIC DEVICES, SUCH AS THE ARTIFICIAL NEURAL NETWORKS TO BE STUDIED HERE, COULD PROVIDE VERY HIGH-SPEED PROCESSING FOR CERTAIN APPLICATIONS WITH VERY LOW POWER REQUIREMENTS, WHICH IS A LIMITING FACTOR IN SCALING UP CONVENTIONAL COMPUTING DEVICES. SUBRECIPIENT ACTIVITIES: THERE ARE NO PLANNED SUBAWARDS.
OpenAlex ID: W4385764026
FWCI: 5.34
Cites: 22
Cited by: 25
Related to: 10