ChatGPT for DevOps: Boost Productivity with AI-Generated Code

ChatGPT, produced by OpenAI, is an excellent tool for artificial intelligence with a Chat-like interface. We can ask ChatGPT any questions. What is unexpected is that the results are quite brilliant when we ask how to write code, too. Since the OpenAI ChatGPT release announcement on 30th November 2022, the ChatGPT tool has been a great hype on the internet. People around the globe tested in the most variant situations. It's impressive the application of the GPT (Generative Pre-trained Transformer) 3.5 large language model with a massive amount of data.

It's a great tool to speed up the DevOps and SysOps workflow. Long story short: it creates a skeleton that we can use as a base for laboratory tests, but nothing is ready for production. It's impressive because it could be improved and probably we could obtain a great result in a few years. After a few tries, I agree with the opinions of Craig Brandt in the "Ansible and ChatGPT: Putting it to the test" article on the Ansible blog.

The good

ChatGPT can understand our request and process it in a way to generate some code. The output code respects the Ansible syntax. It gives us an idea of how automation could be executed in real life.

{{< youtube WTivTbCFK_c >}}

See also: Ansible Lightspeed Review: IBM Watson AI Enhances Development

The bad

The produced code it's far from production ready. Some outputs are definitely better quality than others; it depends on the use case. We can use the Ansible-Lint tool to analyze for syntax checks and test when the Ansible best practices are fulfilled.

Project Wisdom

Project Wisdom announced at AnsibleFest 2022, is a close collaboration between Ansible and IBM Research to create an Ansible AI. The quality will be better because of the specificity of the project. I look forward to testing on my own.

{{< youtube 9Zvuz9r9bc4 >}}

See also: Ansible Lightspeed Review: AI-Powered Automation with IBM Watson

Conclusion

ChatGPT is a great tool to speed up DevOps productivity, but the resulting quality might vary significantly. It's a great way to obtain a laboratory test draft but not a production-ready code.