Introduction
In 1987 Computerworld convened a roundtable to explore how artificial intelligence might intersect with database systems. Chaired by Esther Dyson, the discussion featured three distinct voices: Tom Kehler (Intellicorp), John Landry (Cullinet), and Larry Ellison, then president and CEO of Oracle.
The Participants and Their Views
Kehler represented the expert‑systems movement, advocating for AI that could replicate human decision‑making in tasks such as underwriting. Landry pushed a broader vision, suggesting AI could become the architectural foundation for a new generation of cooperative applications.
Ellison, however, took a contrarian stance. He dismissed AI as a flashy end‑user feature and instead framed it as an internal tool to improve how systems are built.
Ellison’s Contrarian View
Ellison argued that AI should only be applied where it changes the economics or usability of system development. He coined the idea of “fifth‑generation tools” – not new programming languages, but higher‑level, declarative systems that eliminate procedural complexity.
- Focus on intent over instruction.
- Move application logic to servers, letting browsers serve as the primary interface.
- Use AI as an implementation detail, valuable only when it reduces complexity or increases leverage.
Legacy and Modern Relevance
The philosophical foundation Ellison laid in 1987 echoes in today’s cloud platforms and low‑code/no‑code environments. Modern declarative tooling, AI‑assisted code generation, and database‑centric development all reflect his belief that intelligence belongs where it serves a larger architectural goal.
Conclusion
The 1987 roundtable captured a timeless tension: how much intelligence should systems possess versus how much complexity users are willing to tolerate. Ellison’s early insight—that AI is a means to simplify, not a destination—remains a guiding principle for contemporary software engineering.