Ghost Coder

Artificial intelligence is increasingly supporting programmers in their daily work. How effective are these tools? What are the dangers? And how can you benefit from Al-assisted development today?

- Tim Schürmann

Developers have been using artificial intelligence for decades. Even text editors such as Kate offer syntax highlighting and auto-completion. Under the hood, these tools are based on predefined rules. If the editor encounters the "for" character string, for example, it highlights the string in green. Modern AI assistants are far smarter, opening up completely new possibilities for developers.

For instance, what happens if you want to convert your existing program code from C to the more modern Rust? No problem for CodeConvert [1]. The cloud service can translate source code between many programming languages (Figure 1). In addition to well-known representatives such as Go and Python, you can also translate code from more exotic languages such as Ocaml, Cobol, and Tcl. Other translators, such as Refraction [2], Codeium [3], and Figstack [4] also help you beam your code to different languages, but these tools can do far more.

The translator tools use plugins to connect to the development environment, monitor the programmer’s work in the background, and then make coding suggestions while the developer is typing. The additions will complete lines started by developers and typically even finish off whole functions. The snippets that GitHub Copilot [5] outputs also stick to the coding conventions of the rest of the document. AI coding thus goes far beyond conventional rule-based autocompletion.