Fairness and Bias
- What It Means: AI learns from old data. If the data is unfair, the AI may be unfair too.
- Real World Example: Imagine a computer that picks players for a school team. If it always chooses the same kids because of old habits, it is not fair.
Job Displacement
Replacing human work with AI poses both risks and opportunities such as;
- Short to long term rising unemployment
- Reskill and upskill costs for training existing labour forces to work with AI or other areas where AI cannot be used to automate work
- Opportunity for more number of people on more important problems such as the environment
Transparency and Accountability
- What It Means: Some AI systems are like “black boxes.” We cannot always see how they decide.
Explainable AI’s
- There are efforts to create explainable AI’s (sometimes called XAI) by making AI models more understandable and interpretable for humans.
- Real World Example: Think of a robot that sorts recycling. If it makes a mistake, it is hard to know why.
- We can make the AI more explainable by;
- Taking note (or log) each time AI makes a sub decision, leading to a bigger decision, it will be easier for us why the recycling robot has made the mistake.
- We can also try to determine which features (for example, color, size, shape of the object) affect the decision making process by the AI and how.
- There are many more techniques that are being used and developed to make AI’s more transparent.
Privacy and Data Use
- What It Means: AI sometimes collects a lot of personal information. This can be a problem if not handled carefully.
- Real World Example: A smart app might know your favorite games and places. This information must be kept safe.
Environmental Impact
- What It Means: Running AI uses a lot of energy. This can hurt our planet.
- Real World Example: Big computer centers that power AI need a lot of electricity. Using too much energy can affect our environment.
Autonomy vs. Human Control
- What It Means: AI can work by itself. But humans must keep an eye on it.
- Real World Example: A self-driving car drives on its own. If it faces a tough decision, a human must be ready to take control.
