Quick introduction
In this blogpost I will only talk about gen AI language models (LLMs) such as ChatGPT, GPT-4, Claude, Mistral, etc... I may refer to those as "AI" or "gen AI" in the rest of the post for brevity.
AI is here to stay (until the bubble bursts, then it will have to be re-evaluated). It is up to us to use it responsibly to augment our capabilities, not replace them.
Fair use of gen AI
What I consider fair and "healthy" use of gen AI is for:
- Non-creative tasks, for example brainstorming, sounding board, summarization, trying to find specific term from a vague description, etc...
- Free or Open Source non-critical projects, where the generated content is not used commercially.
My use of AI
In my daily work, I do use gen AI, although I use it sparingly, and try not to rely on it.
I use it for various tasks, including the ones mentioned above, as well as:
- Making logos for my open source apps (although I generally tweak them afterwards) when I feel particularly uninspired.
- An improved debugging duck: I explain my problem, the solution I'm planning to implement, and have the AI to point out potential flaws in my reasoning or stuff I may have overlooked or overthought.
- A general helper for formatting community stuff. I manage a community of 60k+ members. And while I'm not really fond of emojis, I understand that they are used by the majority of the people. So I use gen AI to help me format announcements, rules, and other community-related posts to make them more appealing and easier to read for the community members.
- Internationalization help: While I do most of the translations myself (for French and English), the process of moving all the hardcoded strings to i18n files can be tedious. So I use gen AI to help me extract those strings and format them properly for i18n files.
- Tab Completions: I don't mind using gen AI for tab completions. Most of the code developpers write is stuff that has been types many times before. Tab completions are just a way to speed up the writing of tedious and trivial parts of an application.
Gen AI as a professional tool
In my opinion, using gen AI in other uses than stated above should be considered fair only if used in the scope of the profession.
Meaning that in a professional setting, image generation tools is fine for a graphic designer, but not for a lawyer or a marketing team. Github Copilot is fine for a developer, but not for a content writer or a journalist.
Why such restrictions, and the dangers of Gen AI over-use?
The reason I think such restrictions are necessary is that gen AI, when used extensively (for example "Vibe Coding"), can lead to a number of issues, including:
- Over-reliance on AI: Users may become overly dependent on AI tools, leading to a decline in their own skills and creativity.
- Quality concerns: The quality of AI-generated content can vary significantly, and relying on it without proper oversight can result in subpar work.
- Ethical implications: There are ethical concerns around the use of AI, particularly in fields like journalism and law, where accuracy and accountability are paramount.
- Job displacement: People (tech and/or recruiters alike) may start to see gen AI as a replacement for human workers, especially junior roles, leading to job losses and economic disruption.
All of those may cause long-term issues, which may be as basic as unsolvable bugs in software, to more complex societal issues such as misinformation, loss of critical thinking skills, and economic inequality.
For these reasons, I believe it is important to establish clear guidelines and boundaries around the use of gen AI in professional settings.