A month ago, the artificial intelligence company DeepMind (a subsidiary of Google/Alphabet) announced to the world the creation of AlphaCode, an AI capable of performing like an average developer when faced with programming problems.
This was, of course, well received in the technology industry, as it opened up a whole series of possibilities when working as an assistant to human users (not necessarily programmers). However, the problem with AI models like AlphaCode is, as a group of researchers from Univ. Carnegie Mellon explains, that
In fact, a 2020 study by the startup AI21 Labs put the cost of training a code generator model with 1.5 billion parameters (that is, roughly half the complex of PolyCoder) at $80,000-1.6 million. . For its part, solutions like GitHub Copilot have 12,000 million parameters.
The programmer career in 2017 and in the future (with Javier Santana)
PolyCoder was created to democratize research on programming AIs
It is capable of generating code in C, C#, C++, Go, Java, JavaScript, PHP, Python, Ruby, Rust, Scala and TypeScript, although its own creators point out that it particularly excels at writing code in C. In fact, it is capable of coding in C more accurately than any other known model, including Codex (the AI model behind GitHub's CoPilot feature).
A particularity of PolyCoder is that it was not only trained with code files, but also with natural language information extracted from Stack Overflow, the question and answer website for developers:
For example, the datasets used to train Codex have not been made publicly available and its API output follows a 'black box' model, thus preventing researchers from tuning the AI model or studying certain aspects of it. , such as its interpretability.
…[but] our model has already pushed the limit of what can be trained on a single server: any larger model already requires a cluster of servers, dramatically increasing cost."
