Introduction to AI

In this piece, we explore AI. Specifically, we look at the distinction between conventional systems and AI-based systems i.e.
- design approach
- adaptability
- explainability

AI-Based and Conventional Systems

Conventional systems are typically programmed using imperative languages. Human developers define step-by-step instructions including if then statements and loops. As a result the output is often predictable and transparent thus easier for humans to understand.
On the other hand, most AI-based systems, particularly those leveraging AI do not follow predefined rules. They analyze patterns in data to determine how to respond to new input. Example is an AI-based system trained to identify cats.

Problem solving - AI-based systems rely on probabilistic reasoning, statistical inference and pattern recognition to generate results. Hence can handle more complex forms of uncertainty and ambiguity more effectively. Therefor, outputs are not always predictable.

AI based systems have one key challenge - explainability Many AI models, particularly those using deep learning architectures contain billions of parameters hence internal workings are difficult for humans to interpret. This black-box nature is raising concerns in sensitive domains such as defence, transport, healthcare and finance. As a result, achieving transparency and explainabilty has become a critical focus in AI regulations.

Adaptability Conventional systems are static - Require manual updates to incorporate new knowledge or respond to environment changes. AI based systems on the other hand are self-healing (improve as they encounter new data). Adaptability makes AI based systems powerful in dynamic environments. Adaptability requires continuous monitoring to maintain ongoing alignment with core requirements.