From b4abf250fb6aa859b2a3e9ac0fdfe197b3fec882 Mon Sep 17 00:00:00 2001 From: Austin Parker Date: Fri, 8 Nov 2024 09:16:30 -0500 Subject: [PATCH] Add Generative AI/Low Effort Contribution Policy (#2417) * add genai policy; closes #2400 * address feedback * spellcheck * add toc to top-level guide readme * be more explicit about CoC * address review comments --------- Co-authored-by: Morgan McLean --- .cspell.yaml | 4 +- guides/README.md | 7 ++- guides/contributor/genai.md | 97 +++++++++++++++++++++++++++++++++++++ 3 files changed, 106 insertions(+), 2 deletions(-) create mode 100644 guides/contributor/genai.md diff --git a/.cspell.yaml b/.cspell.yaml index 5ca6d010c..3acaa0baf 100644 --- a/.cspell.yaml +++ b/.cspell.yaml @@ -1,5 +1,6 @@ version: 0.2 -ignorePaths: ["elections/**/*", "tools/github/*", "scripts/**/*.py", "scripts/**/*.sh"] +ignorePaths: + ["elections/**/*", "tools/github/*", "scripts/**/*.py", "scripts/**/*.sh"] # Ignore certain patterns in the sigs.yml file patterns: - name: Slack Channel ID @@ -24,6 +25,7 @@ words: - Coralogix - DASD - Docu + - Dosu - datadog - devex - devstats diff --git a/guides/README.md b/guides/README.md index 13b18b381..6e2ab3447 100644 --- a/guides/README.md +++ b/guides/README.md @@ -7,4 +7,9 @@ the OpenTelemetry community. These should be considered a living resource, and everyone is welcome to contribute to them! - [Contributor Guide](./contributor/README.md) -- [Maintainer Guide](./maintainer/README.md) \ No newline at end of file + - [Policy on Generative AI Contributions](./contributor/genai.md) + - [Donations of pre-existing code](./contributor/donations.md) + - [Becoming a member of the OpenTelemetry project](./contributor/membership.md) + - [The Contribution Lifecycle](./contributor/processes.md) +- [Maintainer Guide](./maintainer/README.md) + - [Resolving technical conflicts](./maintainer/conflict-resolution.md) diff --git a/guides/contributor/genai.md b/guides/contributor/genai.md new file mode 100644 index 000000000..e9c7f4565 --- /dev/null +++ b/guides/contributor/genai.md @@ -0,0 +1,97 @@ +# Generative AI Contribution Policy + +This policy provides general guidance for contributors and maintainers relating +to the use of Generative AI in OpenTelemetry projects. This guidance supersedes +and extends the policy defined in the [Linux Foundation Generative AI +Policy](https://www.linuxfoundation.org/legal/generative-ai). + +## The Short Version + +While we welcome contributions from anyone, maintainers of individual +projects may -- at their discretion -- hide or close issues, pull requests, or +other contributions that are made totally or in part through generative AI +tooling. + +## The Long Version + +Increasingly, we have observed a trend of contributors who are utilizing LLMs +and other generative tools to participate in issues and create pull requests. +Regurgitating the output of an LLM is unlikely to be particularly helpful, or +valuable, to other contributors, maintainers, and end-users for a couple of reasons. + +First, time is a very scarce resource for maintainers and approvers. Thoughtful, +high-quality code reviews both essential to the success of OpenTelemetry, and +require a significant time commitment. There is not enough time to give proper +responses to low-effort pull requests without compromising our responsiveness to +high-effort pull requests. Second, OpenTelemetry is a complex, fast-moving +project -- LLMs will often have stale data in their training sets, and are prone +to offering output that is not relevant to the current state of the project. +While LLMs can be incredibly powerful coding assistants they are not a +substitute for human judgement and knowledge. + +With proper usage, Generative AI can be a valuable tool for writing code, +documentation, tests, and more. This level of usage requires enough +understanding of the project to evaluate the LLM output, and to know when to +accept or reject it. Therefore, we ask that contributors do not rely on LLM +output as the sole basis for their contributions. + +Examples of this include: + +- Copying and pasting LLM output into issues or pull requests without any +additional context or explanation. +- Reviewing existing pull requests solely via +LLMs, or using LLMs to respond to issues without any additional context or +explanation. + +## Frequently Asked Questions - Contributors + +_Q: Can I use LLMs to help me write code, documentation, or tests?_ + +Yes, this policy does not prohibit the use of LLMs to assist in writing code, +documentation, or tests. However, we ask that you do not rely on LLM output as +the sole basis for your contributions. + +_Q: Can I use LLMs to help me review pull requests, issues, or understand the code base?_ + +Yes, this is also allowed -- and a good idea! You should use LLMs as a tool to +assist in your understanding, but not as a replacement for your own judgement +and ability. + +_Q: How do I know the difference between allowed and disallowed usages of LLMs?_ + +"If you have to ask, you already know the answer." This policy is not a broad +ban of LLMs, it is a request that you -- as an individual -- use them in a way +that adds value to the project and respects the time of other contributors and +maintainers. If you are using LLMs to help you write code, that is fine; You +should be clear about this in pull requests and reviews. If you are using LLMs +to understand code so that you can participate in issues or reviews, that is +also fine -- but you should be clear about this as well. What is not fine is +copying and pasting a GitHub issue into an LLM prompt and asking it to write the +PR for you, then blindly submitting that response. You must be an active and +willing participant in the process of contributing to OpenTelemetry. + +## Frequently Asked Questions - Maintainers + +_Q: Can I close or hide issues or pull requests that are made through LLMs?_ + +Yes, as your discretion you may close or hide issues or pull requests that are +made through LLMs. We ask that you provide a clear explanation for why you are +doing so, and -- if possible -- provide guidance on how the contributor can +improve their contribution. + +_Q: How do I address contributors who are making consistent, low-effort contributions via LLMs?_ + +If an individual contributor continues to engage in low-effort PRs or issues, +_and_ you have exhausted other avenues of communication, please escalate the +situation to the OpenTelemetry Governance Committee. Per the [Code of Conduct](https://github.com/cncf/foundation/blob/main/code-of-conduct.md), +contributors are expected to help maintain a positive environment, which would +include following guidance and published policy. + +_Q: Can I use LLM or Generative AI tooling to assist in my own work as a maintainer?_ + +In general, you should evaluate the output of LLMs -- regardless of how you use +them -- in the same way you'd evaluate the output of a human contributor or +non-AI tool. For example, tools like [Dosu](https://dosu.dev/) are being used in +certain repositories to aid in code review and issue management. Remember that +these tools can make mistakes, and use your best judgement when evaluating their +output.