Perceptions on AI in the Italian labor market: evidence from a survey experiment

How to cite

Angei, Fabio; Brau, Rinaldo; Balia, Silvia; Sulis, Giovanni. 2025. "Perceptions on AI in the Italian labor market: evidence from a survey experiment." AEA RCT Registry. AEARCTR-0016792. https://doi.org/10.1257/rct.16792-1.0

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

Fabio Angei* , Rinaldo Brau , Silvia Balia , Giovanni Sulis

* Primary investigator

Registry ID
AEARCTR-0016792
DOI
https://doi.org/10.1257/rct.16792-1.0
First registered
Last updated
Status
In development
Trial dates
2025-09-18 – 2026-05-31
Countries
Italy ()
Registry record
http://www.socialscienceregistry.org/trials/16792

Abstract

This research project aims to investigate what workers think about the impact of AI on the labor market and, in particular, how this impact can shape their preferences for welfare-based policies. We will run an online survey experiment to test how different framings of AI – (i) labour substituting, (ii) labour complementing, (iii) skill upgrading necessity can affect the support for various public policies. After exposure to one of the treatments, respondents state their preferences for a set of labor-market policies, including unemployment income support, job retention schemes, minimum income programs, publicly funded training, and hiring subsidies. Our primary outcomes are a Likert-scale favorability score. Moreover, as a behavioral proxy, we measure respondents' willingness to sign a symbolic petition asking policymakers to regulate the impact of AI.

Intervention

We will conduct an online survey with approximately 2.100 employed individuals, resident in Italy, between 18 and 60 years old (excluding self-employed workers). Participants are recruited by IPSOS, a professional survey provider, via quotas for gender, age, macro region, and education to approximate national representatives within the Italian workers population.
The survey collects information within different modules:
• Socio-demographics
• Preferences about and experiences with Artificial Intelligence (AI) – experimental setting
• Working conditions
• Job meaning and work-life balance
• Mental and physical health indicators
The section “Preferences about and experiences with artificial intelligence (AI) is the experimental part of the survey. In this section, we split our participants into three different treatment groups and a control group. We provide information about the AI and its use at work, as in a classical information survey experiment setting.

Experimental design

We evaluate the causal effect of informational framings about Artificial Intelligence (AI) on workers’ policy preferences using a survey experiment with individual-level randomization. We randomize respondents into three information treatments or a pure control group.

Design details

Randomization method
The randomization is done in office by our partner IPSOS through their survey platform.
Randomization unit
Randomization at the individual level.
Clusters
2.100 individuals.
Observations
2.100 individuals.
Arms
T1: 500 individuals T2: 500 individuals T3: 500 individuals T0 (control): 600 individuals
Primary outcomes
• Support for unemployment protection • Support for job retention tools • Support for minimum income policies • Support for publicy-funded worker training • Support for tax incentives for hiring Behavioural outcome: agreement with a symbolic petition asking institutions to govern the impact of AI on the labour market.
Secondary outcomes
We are also very interested in the heterogeneity driven by respondents’ socioeconomic and demographic characteristics—such as income, age, gender, education, and employment status—as well as their initial level of AI knowledge recorded in the experimental-setting section. Moreover, we will also characterize our sample according to the perceived usefulness of public intervention, the perceived family-level financial impact of public intervention, and the perceived risk of job loss.

Cited by

No citation data available for this registration yet.

Keywords