Please enable JavaScript.

Coggle requires JavaScript to display documents.

260128 Lecture 7 Artificial Intelligence, This note is a faithful…

- - - - unsupervised
      - supervised
      - reinforced training
  - - - This technology “learns” the patterns and structure of data used during training, and can then generate new output based on the characteristics of this input.
- - - - Discrimination and bias
      - Abuse and misuse
      - Power and concentration
    - - Knowing how to handle technology along this continuum of extremes will be a key skill for the current and future generations of managers
    - - An AI can memorize sensitive data and inadvertently leak or expose data to others, opening the firm to significant business and legal risk.
        
        A related risk is so-called prompt-injection: Compromising generated AI by entering prompts that cause it to behave in unintended ways.
        
        Another concern is data poisoning: Deliberately feeding incorrect data to an AI so that it generates incorrect results.
    - - Facial recognition systems built using data with mostly Caucasian faces have been shown to be weaker in identifying people of color.
      - The AI-based pedestrian detection system used in many self-driving cars has also been shown less likely to detect both children and people of color.
    - - Individuals who buy into the truthfulness of AI answers without fact checking could be making terrible errors.
        
        Shocking recommendations include:
        
        Ant-poison and glue sandwiches
        
         “Bleach-infused rice surprise”
        
         “Methanol bliss”
    - - AI's ability to draft fluent, convincing text and compelling, realistic images has made it a must-have tool in the deceiver's toolkit.
        
        deepfakes: Sophisticated media (audio, image, video) created by AI that attempts to look or sound like a real person or event.
        
        Presenting generative AI outputs as your own work!
    - - Who is at fault if a firm uses results that were scraped from the Internet and inadvertently include copyrighted images?
        
        Can firms protect proprietary assets from use in GenAI?
        
        If firms use a code generator that returns copyrighted code, who is at fault?
    - - Most organizations face an immediate educational need regarding AI in the workforce
      - Most also spend money, time, and other resources on ensuring that employees are aware of AI's risks as well as how AI can be used in their current roles.
      - Beyond that, AI's impact on industries is unknown.