To me, the best summarization of the current dangers of generative AI comes from the fantastic Ted Chiang in his essay in the New Yorker:
The task that generative A.I. has been most successful at is lowering our expectations, both of the things we read and of ourselves when we write anything for others to read. It is a fundamentally dehumanizing technology because it treats us as less than what we are: creators and apprehenders of meaning. It reduces the amount of intention in the world.
I find this true both the software development work that I do as well as any other documents, blog post, e-mails etc. There’s the maintainers of curl who promise to ‘ ban you and ridicule you in public if you waste our time on crap reports’, simply because dealing with the AI slop is too time consuming and threatens the proper maintenance of the project (arstechnica.com post). There’s AI-generated PRs in your own projects that haven’t received the required amount of ‘human touch’, haven’t been fully understood and are now a huge threat to further development.
In writing unrelated to coding, it’s the CVs, LinkedIn messages, meeting notes, and work emails that appear utterly soulless (and are full of em dashes and check marks ✅). It’s easy to compose, looks good enough and probably does the job.
So what is the next year or two going to bring? On the engineering side, I’m not particularly worried for my personal and work projects, but terrified about the wider implications for security, maintainability and career prospects for junior devs. On the side of corresponding with other humans, I’m personally going to contribute to training data and not generate too many completion requests, and I hope that most people that I work with will eventually share a similar mindset.
Daniel Lenz 