– Robert L. Read, May 9th 2026
I am forced to write this against my will. I love humanity and human intelligence. Public Invention advocates for human intelligence to both guide our own destiny and to help us all live in abundance and peace. Large language models threaten me emotionally, but it has now become a force that must be reckoned with rather than resisted. These models are built on the output of the open-source software movement and the free human scientific corpus.
It is clear that Public Invention, and indeed the entire open-source movement, needs a policy for best utilizing AI because:
- We owe it to our volunteers to allow them to exploit AI to be as productive as possible.
- Although we are not an educational organization, we attempt to advance the career of our volunteers. It is now the case that an educated person must understand how to use AI.
Artificial Intelligence is not free. Public Invention prefers to use free tools, both in the sense of “free speech” and “free beer”. Nonetheless, our policy is that our volunteers will have No Out-of Pocket Expenses (NOOPE). Therefore, artificial intelligence, like soldering irons, oscilloscopes, and 3D printed parts, is a tool that we would prefer to provide to our volunteers if at all possible. We would accept an earmarked donation of money or “tokens” for this purpose, if a generous donor feels it fitting to make such a donation.
AI is also capable of making errors and creating an intellectual dependence that we must guard against. In my personal experience, it has the pernicious effect of closing off technical possibilities by presenting the conventional wisdom as an absolute, which is anathema to our goal of true invention and innovation.
I therefore suggest the following policy:
- AI should not be used for writing prose. If you are not an English speaker, you may write in your native language and use AI for translation into English. However, you must read over the English to check that it is what you intended to say (and symmetrically, for translations out of English).
- We encourage the use of AI for programming and coding. If you can find a free-of-charge AI that is useful, please use it. However, it is incumbent upon you to READ EVERY LINE of the produced code before using it.
- If code generated by an AI is to be committed to one of our GitHub repos and released under an open-source license (we generally use the GNU Affero GPL), then you must CLEARLY mark that the code is generated by an AI, which AI, and on what date. It is best to place your own name in the content as the “AI orchestrator”.
- If you want to have a paid subscription to an AI for your volunteer work, please contact me. We will attempt to accomplish this subject under guidance and budgetary constraint.
- If you generate an image with an AI, carefully check it and clearly mark it as generated by the AI with your orchestration.
- One of the best uses of AI is to find technical papers or other resources as part of our research. The fundamental rule is the same for citing a paper in an academic paper—YOU MUST READ the paper. (“Reading” an academic paper often means “scanning” or reading lightly. Some papers can be scanned in 5 minutes and may take 5 months to fully understand. The key is that you must understand the paper enough that what you are asserting in writing about the paper is true).
- Do not fully trust AIs. They are programmed to stroke your ego, and they do not present uncertainty in their results. In particular, do not let them tell you that something cannot be done, or that your idea won’t work. It will tend to present conventional wisdom as absolute truth. This is not necessarily unuseful—but it must be understood as a danger to true creativity and invention.