Responsible AI: Social bias, copyright, ownership
Generative AI for Business
Carl Rosseel
Head of Business Intelligence Curriculum at DataCamp
Social Bias in AI
Why is there a risk of bias in AI?
- Lack of diversity or misrepresentation of groups in the training data
- pursuit of narrow goals in the model
- Wrong or malicious application of AI by users
Detection and mitigation techniques
Detection
- Discover or identify the presence or existence of
- Algorithms to calculate fairness metrics
- Human audits
- Continuous evaluation and improvement
Mitigation
- Make (something bad) less severe, serious, or painful.
- Diversifying training data
- Adjusting models
1 Oxford Languages
Intellectual Property and AI content
Generative AI and the legal landscape
- Patchwork of new laws
- Staying informed is vital
- Privacy considerations
- Terms of use
- Sharing implications
- Data handling
Generative AI across industries
- Creative fields resist due to job concerns
- Medical research embraces it for breakthroughs
- Supply chain optimizes and automates processes
Ethical considerations
Generative AI is ever evolving
Clear guidelines and feedback opportunities
- Engaging key stakeholders
- Conducting roundtables
- Partnering with civil society
- Product feedback loop
A complex new domain
- Rich with opportunities
- Ethical considerations and challenges
- Mitigating social bias
- Navigating complex copyright issues
