Resolution Criteria
YES resolution will occur if, by December 31, 2024, there is verifiable evidence that a Large Language Model has been the primary developer of a software application that has over one million users or downloads across all platforms and app stores where it is available. The LLM must have handled the majority of the coding and development process without human-led design or significant human intervention in the codebase. No prompts beyond the initial requirements are allowed.
NO resolution will occur if, by December 31, 2024, no Large Language Model has reached the milestone of being the primary developer of a software application with over a million users or downloads. This also includes situations where human developers have played a significant role in the coding or development process beyond providing initial requirements, promoting, or copying/pasting code.
50% resolution will occur if a Large Language Model has written all the code, but while supervised with repeated prompts and human intervention. Human-directed design is allowed, so long as the LLM is the one writing the code.
N/A resolution will apply if there is insufficient data available to verify the download count, or if LLM contribution is too inconclusive to resolve.
The resolution will be based on credible reports, app store data, and statements from the creators or owners of the Large Language Model in question. Independent verification from credible technology news sources may also be used to resolve the market.
Edits
Clarified 50% resolution from "no human written code" to "LLM is the one writing the code". Libraries are allowed and even expected, since that's how most applications are written by humans.
No prompts beyond the initial requirements are allowed
This is unlikely. If nothing else I would "prompt" my AI with incoming reviews. Also the AI is likely to have prompts from the results of compilation and other tools. And it may ask questions about the requirements.
Another way of looking at this: no human-developed game has met these criteria.
A 50% resolution is more possible, but would mean that all libraries included in the game have no human written code, which is unlikely to be worth it by end of year.
@MartinRandall my objective with the 50% resolution was to cover scenarios where LLMs have the coding ability to write useful programs, but the systems haven't been developed to allow them to independently do so. I definitely wouldn't count libraries against this resolution. I'm more imagining a project manager who doesn't know how to code using ChatGPT to write a complete app.