UT austin attribution - Beth add
Precision health and artificial intelligence: with privacy, ethics, bias, health equity, best practices, and case studies
This book provides a comprehensive explanation of precision (i.e., personalized) healthcare and explores how it can be advanced through artificial intelligence (AI) and other data-driven technologies.
Manage AI Bias Instead of Trying to Eliminate It
The negative effects of bias in artificial intelligence models’ underlying data has made headlines, and companies need to find ways to address it. But it’s impossible to completely abolish bias in AI data to equitably account for diverse populations — so instead, companies should remediate it to deliberately compensate for unfairness. The author describes a three-step process that can yield positive results for leaders looking to reduce the impact of AI bias.
Empower Your Nonprofit
Your complete guide to AI in the nonprofit sector Empower Your Nonprofit: Simple Ways to Co-Create with AI for Profound Impact is a comprehensive, accessible, and highly practical guide to harnessing the power of emerging AI technologies in the nonprofit sector. This book delivers strategic research, tools, case studies, and advice to help nonprofits advance their missions through AI, with interviews, outlooks, testimonials, and quotes from nonprofit leaders and influencers in the AI industry delivering key insight to all readers regardless of technical expertise.
Bias in generative AI
Generative Artificial Intelligence models are not free from human bias. There is opportunity for bias to infiltrate Gen AI at a variety of levels. Artificial Intelligence models can include both intentional and unintentional biases in their algorithms, training data, Since the training data of LLMs was created by humans, all of whom hold a variety of value sets, the models will reflect and perpetuate existing human biases.
ChatGPT and other generative AI models utilize freely available information as training data for their models. This information is unverified and contained with biases, stereotypes, and hate speech, which can then be replicated and further spread through the use of Gen AI. In addition, as of January 2024, 52% of information available on the internet is in English, which means this bias is built into the system through training data. About 70% of people working in AI are male (World Economic Forum, 2023 Global Gender Gap Report) and the majority are white (Georgetown University, The US AI Workforce: Analyzing Current Supply and Growth, January 2024). As a result, there have been numerous cases of algorithmic bias, which is when algorithms make decisions that systematically disadvantage certain groups, in generative AI systems.
SOURCE: text adapted from University of Texas Austin's AI LibGuide
Scholarly Articles about AI Bias
- Guided By AI: Navigating Trust, Bias, and Data Exploration in AI-Guided Visual AnalyticsHa, S., Monadjemi, S. and Ottley, A. (2024), Guided By AI: Navigating Trust, Bias, and Data Exploration in AI-Guided Visual Analytics. Computer Graphics Forum, 43: e15108. https://doi-org.ezproxy.bgsu.edu/10.1111/cgf.15108
- How the Terminator might affect the car manufacturing industry: Examining the role of pre-announcement bias for AI-based IS adoptionsHow the Terminator might affect the car manufacturing industry: Examining the role of pre-announcement bias for AI-based IS adoptions by Demlehner, Quirin; Laumer, Sven Information & management , 01/2024, Volume 61, Issue 1
- Human bias in AI models? Anchoring effects and mitigation strategies in large language modelsNguyen, J. K. (2024). Human bias in AI models? Anchoring effects and mitigation strategies in large language models. Journal of Behavioral and Experimental Finance, 43, 100971. https://10.1016/j.jbef.2024.100971