Intent Programming
A working term for AI-assisted development in which the developer increasingly expresses goals, constraints, and correctness conditions while delegating part of implementation generation to AI tools.
Intent programming is best treated here as a working description of an abstraction shift in AI-assisted software development, not as a fully standardized programming paradigm.
Traditional programming usually asks the developer to encode much of the solution procedure directly. An intent-oriented workflow moves more emphasis toward:
goal + constraints + correctness conditions + verificationCode can be one output of that process, but code is not the intent itself.
Delegating implementation does not remove engineering responsibility. It increases the importance of checking whether generated changes respect system boundaries, performance targets, security requirements, and operational invariants.
In critical systems, “the model generated it” is never an acceptance criterion.
Conceptual discussion: Intent Programming.