As AI takes on more programming work, one new essay argues that programming languages should stop treating human-friendly syntax as the main design target. The proposed shift is towards stronger guarantees and richer program databases, a provocative idea with practical consequences for the tools people use to build and check software.
Watch Desk analysis
What happened
The essay “Evolving programming languages in the AI era” examines how programming languages and agentic coding tools might evolve as AI writes an increasing share of code. Its central argument is that languages should prioritise stronger runtime and static guarantees over human-centric syntax or token efficiency.
It also questions whether traditional Language Server Protocols will remain the best foundation for development tools. The author instead points towards program databases, which could offer AI coding systems a richer, more structured view of a codebase than the tools designed primarily to help people navigate files and functions.
These are proposals and predictions from the essay, not evidence that the industry has already adopted them. But they identify a real design tension: code may increasingly be produced by systems that do not care whether a language feels elegant to a person, provided its meaning can be checked reliably.
Why it matters
If AI writes more code, the most valuable feature of a programming language may become its ability to constrain mistakes and expose intent clearly to machines. Human-readable syntax will not vanish, but it may no longer be the only organising principle. The humble compiler could find itself promoted from stern proofreader to active member of the development team.
For developers, that could mean tools built less around autocomplete and more around verification, structured context and dependable system-wide reasoning. It could also make language design less about shaving characters from source code and more about making every possible interpretation harder to get wrong.
Our read
The essay is most useful as a design challenge rather than a forecast. AI-generated code does not automatically make guarantees stronger, and a program database is not a magic cupboard containing perfect understanding. Someone still has to decide what the software is allowed to do, test those boundaries and take responsibility when it goes astray.
Still, the direction is persuasive. If coding assistants are becoming collaborators, programming languages may need to communicate with them in ways that go beyond the conventions built for human eyes and keyboards. The interesting future may belong to languages that are slightly less charming to read but considerably harder to misunderstand.
What to watch
- Whether language designers add stronger static and runtime guarantees specifically for AI-generated code.
- Whether program databases begin replacing or supplementing traditional Language Server Protocol tooling.
- How much developers value token efficiency once coding agents routinely generate large sections of software.
- Whether AI-written code becomes easier to verify, rather than merely faster to produce.
Discussion spark: Should programming languages prioritise machine-checkable guarantees over human-friendly syntax as AI writes more code, or would that make software harder for people to understand and control?
Sources and evidence
- Evolving programming languages in the AI era (24 September 2026, 10:39 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.