CO3722 Data Science
CO3722 Lecture 1 - Introduction To Data Science
Lecture Documents¶
Written Notes¶
Lecture Learning Objectives¶
- Evaluate the principles of AI ethics & their importance in research design
- Discuss ethics & apply to emerging systems.
Key Concerns on AI¶
- Impact on Jobs
- Privacy
- Bias & Discrimination
- Accountability
Ethics¶
Why ethics is important in emerging technologies?
Preliminary study on the ethics of AI
Why is Ethics Relevant¶
- Data is the foundation for machine learning algorithms
- Human beings & cognitive bias
- Data security & integrity
- Impact research design
- Legal exposure & cost penalties
Principles on Ethics¶
Belmont Report - Guide to Ethics within Experimental Research & Algorithm Design
Three Main Principles of Ethics¶
- Respect for Persons - All individuals have a variety of personal circumstances which pose a risk during experiment design
- Benefit - Do No Harm. Intention to do good. When Designing AI - Consider bais/political leanings, race and gender.
- Justice - Fairness & Equality
Establishing AI Ethics¶
- Governance - Companies can leverage existing infrastructure to help manage ethical Ai, if a company is collecting data (e.g. chatgpt) than it is likely to have already established some form of governance system to facilitate ethical sanitisation for data standardisation and quality assurance. (i.e. Follows Laws)
- Explainability - Machine-Learning models (specifically deep learning models) are frequently called
Black Box Models.
Explainability seeks to eliminated the ambiguity by providing AI responses that are clear and transparent to the user. (IBM, AI Ethics)
Poisoning The Well (Cyber-Security Challenges)¶
There are numerous challenges in the field of cyber-security in relation to AI such as:
- Shared Resources
- Open Source Sets
- Pre-Trained Models
- Machine Learning Libraries