What Is AI?

Artificial intelligence is the broad field focused on creating systems that perform tasks usually linked to human judgement, such as recognising images, understanding language, planning and reasoning. Some AI systems follow explicit rules written by people, while others learn from data. The hierarchy described here sits inside this wider space and explains how today’s data-driven methods relate to one another.

Artificial Intelligence At A Glance

Early AI systems often used rule-based logic, search, and knowledge bases. These can still be useful when rules are clear and stable. Data-driven AI learns behaviour from examples instead of fixed rules. This approach scales well when patterns are complex or change over time. In practice, many organisations blend both styles, for example combining rules for policy checks with learned models for text or image understanding.

Machine Learning And Neural Networks

Machine Learning is the subset of AI where models learn patterns from data to make predictions or decisions. In supervised learning, models learn from labelled examples, such as predicting cycle time from past work items. In unsupervised learning, models find structure without labels, such as grouping support tickets by theme. In reinforcement learning, an agent learns by trial and feedback. Neural Networks are a family within Machine Learning that pass numbers through layers of interconnected units, adjusting weights during training to reduce error. They are flexible function approximators that handle complex, non-linear relationships, which is why they are widely used for language, audio and vision tasks.

Deep Learning And Generative AI

Deep Learning uses Neural Networks with many layers, enabling automatic feature extraction from raw data. Convolutional networks power image and video tasks, recurrent and attention-based networks process sequences, and transformers dominate modern language modelling. Generative AI is a subset of Deep Learning that models the distribution of data to produce new content. Large Language Models generate text, while diffusion models and GANs create images, audio or video. For a Scrum Master and Scrum Team, these tools can draft summaries of Scrum Events, suggest test ideas, create refined backlog item descriptions, propose acceptance criteria patterns, and produce stakeholder-ready notes. Useful practice includes keeping prompts clear, checking outputs against team standards, protecting sensitive data, and recording any AI-produced content as part of your team’s working artifacts. Limitations remain, such as bias in training data, incorrect outputs that sound plausible, and gaps in domain knowledge, so human review and context from the team are still essential.