Google’s bard learns programming. The generative artificial intelligence should be able to create, debug and explain code, but also improve existing code and make it faster. However, everything is still running under the guise of experiment and is accordingly available via a waiting list, which is currently only open to people from the USA and Great Britain.
Bard is said to be able to use more than 20 programming languages, including C++, Go, Java, Javascript, Python and Typescript, Google lists in the blog post on. Python code can also be exported directly to Google Colab – an interactive environment in which code can be written and executed. Bard should also be helpful when using Google Sheets. Individual sections of code can be explained by Bard. This should help beginners in particular. An example shows that both connections and individual terms are explained. Of course, everything is done in natural language. The same applies to debugging. All you have to do is tell Bard that the wizard is designed to find bugs and fix them.
(Image: Google blog post)
Google also explains in the blog that Bard will make mistakes – the AI is in an early experimental phase. In this way, Bard can output code that leads to a result, but not to the desired result. “Always double check Bard’s answers, test and review code for errors, bugs and vulnerabilities before relying on it.” Bard will take code from open source projects. Google writes explicitly that the source is given as soon as longer parts are taken over. Conversely, this also means that no source is given when copying short passages.
Generative AI is moving in everywhere
The ability to help with programming is said to have been requested particularly frequently in the feedback on Bard, writes Google. Other services have been able to do this for a long time. Github’s copilot, for example, has served these purposes for years. Copilot X has now been announced: With the availability of OpenAIs GPT-4, the intelligent tool should exploit the extended multimodal capabilities of the large language model and become a universal development assistant for programmers.
Generative AI has also already moved into Google Workspace. There she helps, for example, when writing an e-mail. Microsoft also integrates AI into its Office products, as with Github, the assistant is called Copilot there.
also read
(emw)
