Oy, AI!

Even just a year ago, query engineering for generative AI was something most “serious” computer scientists would scoff at–mostly because, well, most AI understands intuitive and reasonably grammatical language. However, as soon as AI started to

1. compose rather well-written, if sometimes factually inaccurate, text and

2. generate code in various languages,

several lower-level tech industry jobs such as software developer or technical editor (and even sometimes technical writer) have started their decline. With AI becoming more and more specialized and hopefully trained on the right data, I see software development careers switching from generative to editorial, i.e. to tailoring and fixing AI-generated code, and from stovepiped to integrative, i.e. to connecting code AI has generated in different languages.

The consequences for guiding students in computer and data science are clear: Teaching has to include query engineering–which in itself is based on clear, precise, and … wait for it! … grammatically adequate uses of English as the input language. That is because one of the new business KPIs will be employees’ efficient communication with AI. Secondly, students need to engage in comparative approaches between languages; one of the most important new engineering tasks may become building APIs between linguistically different code models. This will, lastly, need to involve a goodly amount of secure coding and thoughtful system engineering in order to minimize attack surfaces and the potential for vulnerabilities.

I didn’t pick the word “thoughtful” in the previous paragraph lightheartedly. As I always tell my students in data mining, critical thinking and an inquiry into the biases and  ramifications for any type of data analysis is crucial to the integrity of their results–and to their reputation as trustworthy engineers. With AI, which is currently not yet  able to distinguish bad from good training data, in the mix, this emphasis on critical thinking becomes even more important. Sadly, still too many students will just cut-and-paste their way out of whatever homework, assuming that the grammatically coherent solution that gets them an A in class will allow them to succeed in a business world whose rules have changed.

(c) Sonja Streuber (this post was not written by AI)

This entry was posted in AI. Bookmark the permalink.

Comments are closed.