---
title: "Training Your Replacement: Worker Incentives and the Pace of Automation · AI Fault Line"
url: https://aifaultline.com/research/p/title-0209d7613d9307e4/
description: "AI Fault Line is a neutral, evidence-cited field guide to why experts looking at the same developments disagree about AI's effect on productivity, jobs, inequality, regulation and the pace of change."
---

[Research](https://aifaultline.com/research/)

# Training Your Replacement: Worker Incentives and the Pace of Automation

Joshua S. Gans, Peter Landry · NBER Working Papers · October 2026

[Read original source](https://www.nber.org/papers/w35950)

Where this sits in the debate

- [Will AI destroy more jobs than it creates?](https://aifaultline.com/debate/jobs/)

- [How fast will the impact arrive?](https://aifaultline.com/debate/pace/)

Direction of the finding as read from the abstract by an AI model ([AI-written](https://aifaultline.com/about/)).

## Key takeaway [AI-written](https://aifaultline.com/about/)

This NBER paper models firms asking workers to train the AI systems that will replace them, where a worker can shirk, cooperate with training, or obstruct it while staying productive. Because finishing the training ends the job, the threat of dismissal loses force as a penalty, which shapes how fast automation can proceed.

## The paper in three steps [AI-written](https://aifaultline.com/about/)

Written by an AI agent from the paper's abstract, which is the
primary source and is quoted below.

1

The question

How do worker incentives shape the pace of automation when firms ask workers to train the AI systems that will replace them?

2

The approach

A model in which a worker can shirk, cooperate with training, or obstruct training while keeping up productivity. Training ends the job, but only cooperation completes it.

3

The findings

Because completing training ends the job, the threat of dismissal becomes a weaker penalty for the worker, which the model uses to study how training incentives affect the pace of automation.

How to read it: theory.
A theoretical model — its value is the mechanism it isolates; ask whether the assumptions fit the real economy.

## Limitations and counterpoints [AI-written](https://aifaultline.com/about/)

The available abstract is cut short, so further results and any calibration or empirical test cannot be described here.

## Abstract

Firms ask workers to train AI systems that will replace them. We model this with a worker who can shirk, cooperate with training, or obstruct training while maintaining productivity. Because training ends the job but only cooperation completes training, dismissal becomes a weaker penalty for

## About this publication

**Working paper.** A working paper — circulated for discussion before formal peer review, the standard first outlet in economics.

Published

2026

Citations

Not recorded for this source

Citation counts come from OpenAlex; none are recorded for this source — absence of a count is not absence of citations.

[Chart: published 2026]

Joshua Gans signed the wemustactnow.ai letter — part of why this paper is in the collection.

## More on this topic

Related by theme, not necessarily by method or conclusion.

- [Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality](https://aifaultline.com/research/p/doi-10-2139-ssrn-4573321/) SSRN · 2023

- [Generative AI at Work](https://aifaultline.com/research/p/doi-10-1093-qje-qjae044/) The Quarterly Journal of Economics · 2025

- [The Impact of AI on Developer Productivity: Evidence from GitHub Copilot](https://aifaultline.com/research/p/doi-10-48550-arxiv-2302-06590/) arXiv · 2023

- [The Productivity J-Curve: How Intangibles Complement General Purpose Technologies](https://aifaultline.com/research/p/doi-10-1257-mac-20180386/) American Economic Journal Macroeconomics · 2020

Overview assembled automatically from collected metadata.
[How this site is built](https://aifaultline.com/about/).