The Evolution: From Drafting Tables to AI

How We Got Here

Digital engineering didn't appear overnight. It's the result of decades of evolution in how engineers work — each era building on the last, each shift making new capabilities possible.

The Pattern of Each Shift

Every transition follows the same pattern:

1. New capability becomes available

A technology matures enough to be practical: computers, 3D modeling, internet connectivity, machine learning.

2. Early adopters apply it to existing workflows

Engineers use the new technology to do what they already do, but faster. CAD replaces drafting — but the process of "draw, review, revise" stays the same. Spreadsheets replace hand calculations — but the process of "calculate, document, file" stays the same.

3. Workflows transform around the capability

Eventually, practitioners realize the new technology enables fundamentally different ways of working. 3D CAD didn't just replace 2D drawings — it enabled parametric design, virtual prototyping, and design reuse. PLM didn't just store files — it enabled configuration management, change tracking, and supply chain integration.

4. The old way becomes unthinkable

Once the transformation takes hold, the previous practice seems absurd. No engineer today would design a jet engine on a drafting board. In 10 years, no engineer will design one without AI-assisted analysis and optimization.

We Are in Transition 3 → 4

Right now, most engineering organizations are somewhere between stages 2 and 3 of the AI shift:

• Stage 2 organizations: Using AI tools occasionally — running ML models on some data, using ChatGPT for documentation, experimenting with generative design on isolated projects. AI is a novelty, not a practice.
• Stage 3 organizations: Restructuring workflows around AI capabilities — automated design exploration, continuous verification, AI-graded assessments, intelligent pipelines. AI is embedded.
• Stage 4: Not yet reality for most. But it's coming: engineering practices where AI is as assumed as electricity.

What Drove Each Shift

The 1960s → 1980s shift: Compute power

Mainframes and minicomputers gave engineers access to computation. But compute was expensive, scarce, and required specialists to operate.

The 1980s → 2000s shift: Software maturity and networking

Engineering software became capable enough for routine use. Networks enabled collaboration. The internet connected supply chains.

The 2000s → 2018 shift: Data volume and system complexity

Systems became too complex for document-based management. The sheer volume of data from sensors, simulations, and global supply chains demanded structured approaches.

The 2018 → Now shift: AI capability and data availability

Machine learning matured. Engineering data became structured enough for AI to process. Large language models demonstrated that AI could understand technical language and reasoning. For the first time, forming associations across information streams became computationally practical at scale.

The Lesson from History

Every shift met resistance. Engineers argued that CAD would never replace drafting. That simulation couldn't be trusted like physical testing. That models would never replace documents. Every argument was valid at the time — the early tools were genuinely inferior to mature practices. But the tools improved, and the practices transformed.

The same pattern is playing out with AI. Today's objections — that AI makes mistakes, that it can't be trusted for safety-critical work, that it doesn't understand physics — are valid today. They won't be valid forever. The organizations building AI-native engineering practices now will be positioned when the technology catches up to the ambition.

Assessment