Why do computer scientists and management scholars study AI so differently?

Raisch and Krakowski (2021) argue that the two disciplines studying AI in organizations are each watching only one half of the automation-augmentation picture, and that this split distorts both bodies of research. "Currently, computer scientists, roboticists, and engineers are the scholars most commonly studying AI," and their "primary objective is to automate as far as possible," because, quoting Cummings (2014: 62), they often regard humans as "a mere disturbance in the system that can and should be designed out." Management scholars, by contrast, gravitate to augmentation, because managerial tasks keep humans "in the loop" working closely with machines — which is why, the authors note, "the three business books devote the bulk of their attention to augmentation." Raisch and Krakowski's conclusion is symmetrical: "management research limited to augmentation may be as biased as computer science's traditional focus on automation."

The reasoning follows from how each field works, not just what it studies. Computer scientists "tend to use laboratory settings for their research, which reduce the inherent variability that characterizes human behavior," and their methodologies are built to maximize algorithmic performance while disregarding "the wider organizational and societal implications." Management scholars run the opposite risk: well-placed to study real-world human-machine interaction, but, per the authors, not positioned to see the automation side computer science owns. The authors' proposed fix is disciplinary integration, not better separation: citing Orlikowski (2007), they argue "the technological and the social worlds are merging," so "computer scientists and management scholars no longer study separate phenomena" and need to jointly build a shared research foundation — plus, per their conclusion, a multi-level agenda spanning micro (managers' changing roles and status), meso, and macro (institutions, standard-setting, and the networks of "activists, companies, governments, international organizations, and public institutions" that act on AI's societal implications).

This claim about who studies what sits one level above the paradox itself — see the sibling note Are automation and augmentation really a simple trade-off? — and names a plausible cause of why the automation/augmentation trade-off framing persisted uncorrected in the three books the authors reviewed: a literature split by discipline rather than integrated by phenomenon. Two notes elsewhere in the vault read as instances of the kind of research Raisch and Krakowski call for rather than against: Do humans and AI divide labor differently in knowledge work? is a management-adjacent field study of real-world human-AI collaboration at Boston Consulting Group, the exact terrain the authors say management scholars are "particularly well-equipped" to study; Does AI augmentation protect workers from skill erosion? is itself an interdisciplinary risk-mapping study that treats augmentation as worth the same scrutiny automation gets, closing precisely the gap Raisch and Krakowski describe.

The excerpt makes this disciplinary-split claim as an argument, not as a measured finding: it offers no bibliometric count of how AI-in-management papers divide by discipline or by automation/augmentation focus, and its evidence for computer science's bias is a single secondary quotation (Cummings, 2014) rather than a survey of that literature. The claim should be read as the authors' diagnosis and research-agenda pitch, not as an established account of the field's composition — though it does supply a specific, checkable prediction: that interdisciplinary work bridging computer science and management on AI should be rarer than single-discipline work on either automation or augmentation alone.