ON JOKES AND BOUNDARIES: NEGOTIATING THE VALUE OF ML AND AI WORK THROUGH MEMES

Abstract

The rise of Large Language Models (LLMs) capable of generating code has unsettled core assumptions about the nature and value of coding-related professions. In this paper, we analyse how memes navigate shifting perceptions of value and worth in this situation, drawing on valuation studies and scholarship on humorous online memes. Our dataset comprises 401 memes tagged with 'machine learning' and 'artificial intelligence' on ProgrammerHumor.io. We purposively sampled 60 image-macro memes containing humour that assign value to AI/ML work. Through thematic analysis of the textual and visual elements of these memes, we demonstrate how they establish symbolic boundaries between various groups of coders, highlight the perceived limitations of AI-generated code and address the uncertainties surrounding the future of coding work. Firstly, memes differentiate between coders based on their relationship with AI, contrasting skilled programmers who have control over code with 'prompt engineers' and 'vibe coders' who rely on LLMs. Secondly, they highlight the perceived shortcomings of AI-generated code, particularly its unreliability and the burden of debugging, thereby asserting the continued superiority of human expertise. Finally, memes articulate automation anxiety by addressing fears of job replacement, while also emphasising AI's dependence on human-generated code. Together, the memes constitute a dynamic cultural arena in which professionals negotiate the impact of AI coding tools, construct hierarchies of worth and process the rapid changes in valuation in their field.