Sixty years, five paradigm shifts — from punch-card regression to foundation models generating £150 billion in annual economic value.
Six decades of inflection points, from mainframes to foundation models.
Each era brought a paradigm shift in capability, accessibility, and the kinds of question data science could answer.
IBM's System/360 (1964) let businesses process thousands of records and run regressions that once took human statisticians months. Early FORTRAN code produced printed tables analysts interpreted over days.
Relational databases, SQL, and data warehouses moved organisations from transaction processing to purpose-built analytics. BI tools let non-technical users build dashboards without code.
Google's MapReduce paper (2004) and Hadoop created distributed-processing paradigms for petabyte-scale data — volume, velocity, and variety that traditional systems couldn't handle.
Deep learning breakthroughs — ImageNet (2012), AlphaGo (2016) — demonstrated neural networks could match or exceed human performance on specific tasks.
GPT-3 (2020) and ChatGPT (late 2022) moved the paradigm from training custom models to orchestrating foundation models. Generative AI doesn't just classify or predict — it creates text, images, code, and multimodal content through simple API calls or prompts.
Mainframe computing arrives; businesses process thousands of records and run regression analyses at scale.
Dedicated statistical software turns analysis from a programming exercise into packaged procedures.
Analysts without programming backgrounds can now perform sophisticated statistical work.
New distributed-processing paradigm ushers in the Big Data era and the rise of Hadoop.
Deep learning demonstrates neural networks can match or exceed human performance on specific tasks.
Reinforcement learning conquers Go, cementing deep learning's move into the mainstream.
Large language models exhibit broad competence without task-specific training datasets.
Generative AI reaches the mainstream — the most fundamental shift since the field's inception.
Key Insight — Capabilities that seem impossibly complex eventually become standard practice, often with surprising speed once key enabling technologies emerge. The professionals who thrive aren't those with deep expertise in legacy systems — they're the ones who recognise paradigm shifts early.