This site is a work in progress and has not been widely shared. Content may contain errors. Feedback is welcome.
This site is undergoing review. Some annotations were human-generated, some AI-generated — all are being verified.
Back to papers

The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise

Dell’Acqua, Ayoubi, Lifshitz, Sadun, Mollick, Mollick, Han, Goldman, Nair, Taub, Lakhani

2025NBER Working Paper24 citations
Experimental evidenceInterdisciplinaryCausal
LLM / Generative AIHuman-AI collaborationCollective intelligence / teamsDecision-making
View RepositoryDOI: 10.3386/w33641
Summary

Dell'Acqua et al. conduct a pre-registered field experiment with 776 Procter & Gamble professionals randomly assigned to work individually or in teams, with or without AI assistance, on real new product development challenges to examine how generative AI affects performance, expertise sharing, and social engagement in collaborative work

Main Finding

Individuals using AI achieved performance improvements of 0.37 standard deviations and matched the quality of two-person teams without AI, while AI eliminated functional silos between R&D and Commercial professionals and increased positive emotions compared to working alone

Primary Datasets

Dell'Acqua P&G Cybernetic Teammate

AI-focused

Procter & Gamble field experiment data (May-July 2024); AI interaction logs (prompts and responses); expert quality evaluations; pre- and post-task surveys

Secondary Datasets

Microsoft Teams collaboration transcripts

Key Methods
Pre-registered 2x2 randomized field experiment with 776 professionals; random assignment to individual/team and with/without AI conditions; regression analysis with multiple specifications including fixed effects and controls
Sample Period
2024
Geographic Coverage
Europe and Americas (Procter & Gamble operations)
Sample Size
776 professionals (154 individuals without AI, 155 individuals with AI, 115 teams without AI comprising 230 people, 126 teams with AI comprising 252 people); 550 solutions evaluated
Level of Analysis
Individual, Firm, Task
Occupation Classification
None
Industry Classification
Consumer packaged goods (specific business units: Baby Care, Feminine Care, Grooming, Oral Care)
Notes
Harvard Business School Working Paper No. 25-043 [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations. [Claude classification]: Harvard Business School Working Paper No. 25-043. Pre-registered at AEA RCT Registry (AEARCTR-0013603). Experiment conducted May-July 2024 as part of organizational upskilling program. Solutions evaluated by 22 expert evaluators (1,595 evaluations total). Cross-over design where control groups later received AI training. AI retention analysis and semantic similarity analysis using NLP. Some robustness checks used AI-generated evaluations.