Cognitive Computing

IBM Watson is a cognitive computing system originally developed for the DeepQA project at IBM Research and best known for winning Jeopardy!.

Cognitive Computing is the use of computer systems to simulate human thought processes : learning, reasoning, problem-solving, and language understanding : to solve complex, ambiguous problems, popularised by IBM’s Watson system that won Jeopardy! in February 2011 and largely superseded by large language models (LLMs) and agentic AI as of 2026.

Cognitive computing vs artificial intelligence

The terms cognitive computing and AI overlap substantially but were positioned differently by IBM during the Watson era.

IBM Watson: the term’s origin and trajectory

• 2006-2010: DeepQA project. IBM Research developed Watson to answer natural-language questions in real time, succeeding Deep Blue (defeated Kasparov 1997).
• February 2011: Jeopardy! victory. Watson won against champions Ken Jennings and Brad Rutter, winning $1 million. The landmark public moment for ‘cognitive computing.’
• 2011-2014: Healthcare focus. IBM positioned Watson for cancer diagnosis support with Memorial Sloan Kettering. Mixed clinical results.
• 2015-2018: Watson commercialisation. IBM launched Watson Discovery, Watson Assistant (chatbot), Watson Studio, and Watson for HR.
• 2023: Watsonx repositioning. IBM relaunched its AI platform as watsonx, focused on enterprise AI, foundation models, and AI governance.
• 2024-2026: LLM era. Large language models (Claude, GPT, Gemini, Llama) largely subsume the cognitive computing framing; capabilities are available at lower cost with stronger performance.

Underlying technologies in cognitive computing

• Natural language processing (NLP). Understanding human language in text and speech. Watson’s core capability for Jeopardy!
• Machine learning (ML). Statistical models that learn patterns from data and improve with experience.
• Deep learning. Multi-layered neural networks effective for image, speech, and language tasks.
• Expert systems. Rule-based systems encoding domain expert knowledge for decision support.
• Computer vision. Interpreting visual information from images and video.
• Speech recognition. Converting spoken language to text.
• Knowledge representation. Structuring information so systems can reason over it.
• Reasoning and inference engines. Drawing conclusions from facts and rules.

HR applications of cognitive computing (and modern AI)

• Resume screening and candidate matching. NLP-powered review of resumes against job requirements. Modern implementations use LLMs and AI orchestration (HiredScore, Eightfold, Paradox).
• Employee service chatbots. Natural language interfaces for HR queries on leave policy, benefits, and payroll.
• Sentiment analysis. Analysing employee feedback and survey responses for sentiment patterns.
• Predictive workforce analytics. Predicting turnover, identifying high-potential employees, forecasting capability needs.
• Personalised learning recommendations. Suggesting development content based on role and capability gaps.
• Skills inference and matching. Inferring skills from work history and project content; matching skills to opportunities.
• Compliance monitoring. Identifying potential policy violations and harassment patterns.

The 2024-2026 evolution: from cognitive computing to LLMs

• LLMs as foundation. Claude, GPT, Gemini, and Llama provide language understanding and reasoning that previously required cognitive computing platforms with substantial customisation.
• Reduced engineering overhead. Modern AI needs less manual rule-writing and ontology development.
• Agentic AI emergence. Multi-step AI systems replace rule-based workflows of older cognitive computing.
• Compliance focus shift. EU AI Act (effective August 2026), Colorado AI Act (effective February 2026), and NYC LL144 address transparency, bias audit, and risk classification.
• ‘Augmented intelligence’ framing returning. IBM’s original human-in-the-loop framing has gained renewed relevance in responsible AI discussions.

See also Artificial Intelligence in HR for the broader umbrella, Cloud Computing for infrastructure foundation, Cloud-Based HR Software for applied platforms, and Chatbots for the applied cognitive interface.

Frequently asked questions