Why AI is making old programming languages stronger

This article summarizes an episode of Ryan Peterman’s video series featuring Anders Hejlsberg, creator of TypeScript and C#.

Anders Hejlsberg, creator of TypeScript and C# / Photo credit: Ryan Peterman

AI coding tools will not replace traditional programming languages. Instead, they will make older systems even stronger.

Anders Hejlsberg, creator of TypeScript and C#, says AI tools work best when they are used in systems with lots of existing documentation.

Incumbent languages dominate AI

This dynamic forces developer tools to prioritize compatibility over reinvention:

• Training volume: AI requires vast data to study, favoring frameworks like JavaScript and Python.
• Error correction: Mainstream platforms feature type annotations that flag model mistakes.
• Gradual adoption: Upgrades allow teams to test features on isolated project modules before committing to integration.

As Hejlsberg puts it, “incumbent languages are actually going to flourish because of AI.”

Porting safeguards system reliability

Upgrading performance demands a strict approach to backward compatibility:

• Identify performance bottlenecks: Pinpoint processing issues before changing a programming environment.
• Match design requirements: Select software environments like Go that offer speed and automatic memory management.
• Enable parallel processing: Deploy multiple computer processors to handle workloads simultaneously without forcing a redesign.

When migrating TypeScript infrastructure, the engineering team refused to build from scratch because “only by porting can we preserve the semantics, the algorithms, and the exact behavior of our existing compiler, which everyone depends on for backward compatibility.”

AI demands deterministic testing

Just as porting preserves predictability during system upgrades, deploying AI requires teams to isolate unpredictable outputs into standalone translation programs.