Skip to Main Content
Rider University Library

Introduction to Gen AI

Challenges and Risks

  • Bias in the training data can affect results. There is a Lack of Diversity in the development of AI systems and training models.It's not fair.
    Example: A hiring recommendation AI too AI prioritizes candidates from specific schools due to historical biases embedded in the training data.
  • Errors may cause harm
    Example: A financial forecasting tool generates incorrect projections, leading to poor investment decisions
  • Private Data might exposed: 
    Example:  A machine learning project inadvertently exposes sensitive user data while testing an AI algorithm.
  • Technology may not address all needs
    Example: Marketing campagain doesn't account for diverse consumer behaviors leading to an ineffective campaign or a tutoring system doesn't adapt to diverser learners, leaving students behind. 
  • Users must trust a complex system 
    Example: An AI diagnostic tool suggests treatments for a biology lab simulation, but students can’t verify the underlying reasoning.
  • Who's liable for AI-driven decisions?
    Example: An AI model predicts optimal land use for a conservation project, but errors lead to unintended ecological damage—who is responsible?
  • Microsoft.2024. Challenges with AI. Retrieved November 20, 2024, from https://learn.microsoft.com/en-us/training/modules/get-started-ai-fundamentals/7-challenges-with-ai
  • OpenAI. (2024, November 20). Conversation with ChatGPT about AI risks and challenges. Retrieved from https://chat.openai.com/