Zoea

Zoea have published their first research paper on the arXiv.org preprint server at Cornell University. Entitled "Composable Inductive Programming Without Limits" this paper describes the unique Zoea approach to automation of coding through AI.

Inductive programming which is the automatic generation of software from input-output examples has had a long history of sporadic progress. Until now its widespread adoption has been held back by its inability to cope with larger programs. Zoea incorporates a number of advances that overcome this problem. These include the ability to combine smaller inductive programs to form larger ones as well as the capability to specify any number intermediate values between inputs and outputs.

The paper provides an overview of the Zoea programming language including a number of simple examples of Zoea programs. These show how simple Zoea is compared to any conventional programming language and how it can be learned in a fraction of the time.

The paper also describes the inner workings of the AI behind the Zoea compiler. Zoea utilises a lot of explicit programming knowledge in combination with test case analysis using pattern recognition. The software architecture is organised as a highly parallel variant of the classic blackboard model.