Page 271 - Touhpad Ai

Training Data Bias
Training data bias occurs when the data used to develop AI systems is
unrepresentative, incomplete, or skewed. If the data reflects existing prejudices or
excludes certain groups, the AI system may learn and perpetuate these biases.

AI systems make decisions based on data they are trained on, so it is crucial
to check datasets for bias. One way to do this is to examine whether certain
groups are over- or under-represented in the data. For example, a medical
AI system trained on data from male patients may not perform well for female
patients, leading to misdiagnoses.
Similarly, an AI system used for loan approvals might be biased if the training data primarily includes applicants from
affluent neighbourhoods, disregarding those from poorer areas. Bias can also occur in data labelling.

Algorithmic Bias

Algorithmic bias refers to the bias that may exist in the design, implementation,
and outcomes of algorithms used in AI systems. This bias can result in unfair or
discriminatory outcomes, often reflecting the prejudices or limitations of the data
used to train the algorithm, as well as the assumptions and decisions made by
developers during the algorithm’s creation.
For example, if an AI algorithm used in hiring processes is trained on historical data
that reflects biased hiring decisions, such as favouring one demographic group
over another, the algorithm may perpetuate these biases when making new hiring
recommendations.

Addressing algorithmic bias requires careful consideration of the data used to train algorithms, as well as transparency
and accountability in the design and implementation processes. Techniques such as bias detection, fairness-aware
algorithms, and diverse data collection can help mitigate algorithmic bias and promote more equitable outcomes in
AI systems.

Cognitive Bias
Cognitive bias refers to systematic patterns of deviation from rationality or objectivity
in judgment or decision-making. These biases are often influenced by factors such
as emotions, personal experiences, and social norms, leading individuals to make
judgments or decisions that may not be entirely rational or impartial.

For example, imagine a person who strongly believes that climate change is not
real. When presented with scientific evidence supporting climate change, they might
dismiss it or interpret it in a way that aligns with their preconceived notion, while
ignoring or downplaying evidence to the contrary. This individual’s confirmation bias
prevents them from objectively considering added information and may reinforce
their existing beliefs, even in the face of contradictory evidence.
Cognitive biases can impact various aspects of life, including personal relationships, business decisions, and societal
perceptions. Recognising and understanding these biases is essential for making more informed and rational decisions,
both individually and collectively. Strategies such as critical thinking, mindfulness, and seeking diverse perspectives
can help mitigate the influence of cognitive biases.

Ethical Practices in AI 269