Limitations of AI and Key Terminology

Covered in this video are the core limitations of AI that people professionals need to understand before deploying or evaluating AI tools in a workplace setting. The video addresses why AI does not genuinely think, explaining how tools such as ChatGPT produce outputs based on statistical pattern recognition rather than comprehension or reasoning. It examines the dependency of AI systems on data quality and training data, and why incomplete or biased training data directly affects the fairness of outputs — including in recruitment contexts where historical hiring patterns may disadvantage particular groups of candidates.

The video introduces the key terminology that underpins informed conversations about AI in HR and people practice. Algorithms, data training, prompts and models are all covered, alongside the distinctions between automation as a rule-based process and AI as a system capable of learning and prediction. Within that, the video addresses generative AI, predictive analytics and natural language processing (NLP) as the types of AI most relevant to HR work, and looks at workplace applications including HR chatbots, CV screening tools and automated onboarding workflows.

Assessment Criterion 1.1 of FREE08 asks learners to explain what is meant by artificial intelligence and automation in a workplace context, and this video addresses that directly by drawing clear boundaries between what automation does and what AI does, while also setting out where AI falls short. The contextual limitations covered include AI's inability to apply common sense or ethical reasoning, its failure to recognise when changed circumstances make historical patterns unreliable, and the risks this creates for people decisions. The video treats human judgement not as a supplement to AI but as a necessary check on it throughout.