Just read some fascinating data from GitHub’s 2025 Octoverse report: TypeScript surged 66% to become the #1 programming language, displacing JavaScript (source).
The explanation? AI coding assistants.
GitHub researchers identified what they call a “convenience loop”: developers are choosing TypeScript because AI tools work better with static typing. Types provide essential guardrails for LLMs, reducing hallucinations and improving code generation quality.
This is wild to think about from a product strategy perspective. We’re selecting technologies based on how well they work with AI, not just how well they work for humans.
Some questions this raises:
Are we optimizing for AI or for developers? If a language is harder for humans but easier for AI, is that a net win?
Does this change make sense? Static typing arguably makes code more maintainable regardless of AI. Maybe AI is just accelerating a shift that was already happening?
What other tools will AI reshape? If language choice is being driven by AI effectiveness, what about frameworks, libraries, architectural patterns?
Business implications: Should we factor “AI compatibility” into technology decisions now? Is it a competitive advantage to choose tools that work well with AI?
From a product lens, this feels like a fundamental shift. We used to choose technologies based on team expertise, ecosystem maturity, and performance characteristics. Now we might need to add “how well does AI understand this” as a first-class decision criterion.
Curious what engineering leaders think about this. Are you seeing teams push for language/framework changes “because AI works better with it”?
This makes total sense from a developer experience perspective, and honestly I think the shift is beneficial beyond just AI.
Types are documentation for AI the same way they’re documentation for humans.
When I’m working in TypeScript with Cursor, the AI has context about what parameters a function expects, what it returns, and what the data shapes are throughout the system. This reduces hallucinations because there are fewer ways to be wrong.
But here’s the thing: this helps human developers too! When I’m reviewing AI-generated code in TypeScript, I can immediately see type mismatches. The compiler catches a whole class of errors that would be subtle bugs in JavaScript.
Are we adding complexity for AI’s benefit?
I don’t think so. Static typing adds some upfront complexity (writing type definitions) but reduces downstream complexity (fewer runtime errors, better refactoring, clearer interfaces). AI just makes the tradeoff even more favorable because it can write a lot of the type boilerplate for you.
The “convenience loop” works both ways: AI works better with types, and AI makes working with types easier. That’s a virtuous cycle, not a compromise.
We’re absolutely seeing this pressure from teams. In Q4 alone, I’ve had three different teams propose TypeScript migrations with “AI works better with it” as a primary justification.
Here’s my challenge with this: Is this a valid technical reason or cargo cult thinking?
Maya makes a great point that types help both AI and humans. I agree with that. But we need to be rigorous about whether the migration cost is justified by the actual productivity gains.
What we’re evaluating:
Does TypeScript + AI actually improve velocity including verification time?
Will the migration disrupt delivery for 2-3 sprints?
Do we have the TypeScript expertise to maintain it long-term?
Is this solving a real problem or following a trend?
Valid reason: “Our dynamic JavaScript code has high bug rates, and TypeScript’s static analysis would catch these errors earlier. AI makes the migration faster because it can write type definitions.”
Cargo cult: “Everyone’s using TypeScript now because of AI, so we should too.”
The data suggesting TypeScript reduces AI hallucinations is compelling. But each team needs to evaluate whether the specific benefits apply to their codebase and whether the transition cost is worth it.
Not every technical trend is right for every team. We need to be thoughtful about which AI-driven shifts genuinely improve our outcomes versus which are hype.
Luis raises the critical question about cargo cult vs legitimate technical decision. This is something we’re navigating at the organizational level.
The trend is real and accelerating. We’re seeing AI reshape not just language choice but hiring requirements, developer education, and technology strategy.
Here’s what I’m observing across our 80-person engineering org:
Hiring impact:
Job descriptions now list “experience with TypeScript” as non-negotiable
Candidates who can work effectively with AI-assisted development are prioritized
Static typing proficiency is becoming a baseline skill expectation
Skill development shift:
Onboarding now includes “working effectively with AI tools” training
We’re investing in TypeScript education for JavaScript developers
Focus shifting from syntax memorization to architectural thinking
Strategic technology decisions:
“AI compatibility” is now a factor in framework selection
We evaluate: How well do AI tools understand this technology?
Not the only factor, but a legitimate consideration alongside performance, ecosystem, team expertise
I think Maya’s framing is right: types help both AI and humans, so this isn’t purely optimizing for AI. But Luis’s caution is also valid: we need evidence-based decisions, not trend-following.
The question for leadership: Is this a temporary adjustment to AI capabilities, or a permanent shift in how we think about technology choices?