AI, Organizations, and Tacit Knowledge

How to cite

Perkowski, Patryk; Kogut, Bruce; Dell'Acqua, Fabrizio. 2019. "AI, Organizations, and Tacit Knowledge." AEA RCT Registry. AEARCTR-0004382. https://doi.org/10.1257/rct.4382-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Patryk Perkowski* , Bruce Kogut , Fabrizio Dell'Acqua

* Primary investigator

Registry ID
AEARCTR-0004382
DOI
https://doi.org/10.1257/rct.4382-1.1
First registered
Last updated
Status
In development
Trial dates
2019-10-16 – 2020-02-04
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/4382

Abstract

There is a growing interest in understanding the future of work. Given recent advancements in artificial intelligence and machine-learning technology, there are increasing concerns that many of today’s jobs will become automated and performed by machines. Much of the research on the future of work has been grounded in a task-based approach introduced in economics, which focuses on tasks’ technical suitability for automation. While the task-based approach represents an important first step, organizations are key drivers of technological change, and we must understand the role they will play in the future of work. In this paper, we propose one way to think about organizations in this debate. Our approach, grounded in the knowledge-based theory of firm, posits that we can theorize of automation as the transfer of knowledge from an individual to a machine. Such transfer is more difficult when knowledge is tacit, so automation will be more difficult to implement for teams whose coordination is based on organizational routines rather than explicit mechanisms. We test these predictions in a coordination game based in the lab.

Intervention

We are running experiments in the Behavioral Research Lab at Columbia Business School. Participants will compete in Super Mario Party mini-games that require coordination and communication.

Experimental design

Our experiment will follow a 2x3 factorial design. First, teams will be randomly assigned to either the
tacit coordination condition or the explicit coordination condition. Second, teams will be randomly
assigned to an organizational change. A third of teams will have a team-member replaced by an AI, a
third will have a player replaced by a new player (the new hire condition), or a control condition. This
treatment arm will allow us to test whether the effects we observe related to tacit-vs-explicit knowledge
are unique to AI’s or also occur when a new human player is introduced to the team.

Design details

Randomization method
Players will be randomly assigned to treatment conditions via randomization done in STATA.
Randomization unit
The randomization unit is the team of four.
Clusters
6 conditions * 10 teams per condition * 4 players per team = 240 subjects
Observations
6 conditions * 10 teams per condition * 4 players per team = 240 subjects
Arms
6 conditions * 10 teams per condition * 4 players per team = 240 subjects
Primary outcomes
Our key primary outcomes are total points scored, total earnings, number of coordination failures, and number of words spoken.
Secondary outcomes
n/a

Cited by

No citation data available for this registration yet.

Keywords