Matthias Haslberger
@mhaslberger.bsky.social
Postdoc @ Uni St. Gallen. Research on AI, tech. change, employment, wealth inequality, comparative education.
https://matthiashaslberger.github.io
https://matthiashaslberger.github.io
Two inspiring days at the Nuffield Politics of AI workshop!
This was social science at its best, thanks to all involved! @benansell.bsky.social @jburnmurdoch.ft.com @nuffieldcollege.bsky.social
This was social science at its best, thanks to all involved! @benansell.bsky.social @jburnmurdoch.ft.com @nuffieldcollege.bsky.social
November 14, 2025 at 6:36 PM
Two inspiring days at the Nuffield Politics of AI workshop!
This was social science at its best, thanks to all involved! @benansell.bsky.social @jburnmurdoch.ft.com @nuffieldcollege.bsky.social
This was social science at its best, thanks to all involved! @benansell.bsky.social @jburnmurdoch.ft.com @nuffieldcollege.bsky.social
I'm excited to share the schedule for the Politics of AI workshop I have the pleasure of co-hosting together with @benansell.bsky.social at @nuffieldcollege.bsky.social next week.
Link: matthiashaslberger.github.io/ai/
Link: matthiashaslberger.github.io/ai/
November 4, 2025 at 1:10 PM
I'm excited to share the schedule for the Politics of AI workshop I have the pleasure of co-hosting together with @benansell.bsky.social at @nuffieldcollege.bsky.social next week.
Link: matthiashaslberger.github.io/ai/
Link: matthiashaslberger.github.io/ai/
🤔So, people are self-interested, hence they should be less supportive of social policies they don't expect to benefit from themselves, right?
But that's not what we found! Working with AI increased support for social policies, especially for providing training opportunities.
But that's not what we found! Working with AI increased support for social policies, especially for providing training opportunities.
September 11, 2025 at 4:42 PM
🤔So, people are self-interested, hence they should be less supportive of social policies they don't expect to benefit from themselves, right?
But that's not what we found! Working with AI increased support for social policies, especially for providing training opportunities.
But that's not what we found! Working with AI increased support for social policies, especially for providing training opportunities.
💡Working with AI in our tasks did not make people feel like they were at greater risk of losing their jobs. Overall, they became more optimistic about the consequences of AI, for themselves and for society.
September 11, 2025 at 4:42 PM
💡Working with AI in our tasks did not make people feel like they were at greater risk of losing their jobs. Overall, they became more optimistic about the consequences of AI, for themselves and for society.
Crucially, in light of recent arguments that changing skill demands, uncertainty over skill needs, and weakening corporatism might undermine the benefits of dual VET, we test this hypothesis using three different knowledge economy indicators but find no consistent evidence for such a trend.
August 27, 2025 at 10:20 AM
Crucially, in light of recent arguments that changing skill demands, uncertainty over skill needs, and weakening corporatism might undermine the benefits of dual VET, we test this hypothesis using three different knowledge economy indicators but find no consistent evidence for such a trend.
We find a strong and robust association of a high dual VET share with lower wage inequality, concentrated in the lower half of the distribution. The size is substantial: a 1 pp increase in the dual VET share is
associated with a decrease of the 50/10 ratio by about
1.3 to 1.6% of a SD.
associated with a decrease of the 50/10 ratio by about
1.3 to 1.6% of a SD.
August 27, 2025 at 10:20 AM
We find a strong and robust association of a high dual VET share with lower wage inequality, concentrated in the lower half of the distribution. The size is substantial: a 1 pp increase in the dual VET share is
associated with a decrease of the 50/10 ratio by about
1.3 to 1.6% of a SD.
associated with a decrease of the 50/10 ratio by about
1.3 to 1.6% of a SD.
Who shifts most? The tech-risk → VET tilt is stronger for men, right-leaning respondents, higher-income earners, and the tertiary-educated. Politics + socioeconomic position shape how people insure against uncertainty.
August 13, 2025 at 8:32 AM
Who shifts most? The tech-risk → VET tilt is stronger for men, right-leaning respondents, higher-income earners, and the tertiary-educated. Politics + socioeconomic position shape how people insure against uncertainty.
Cross-country pattern: The VET “safe haven” pattern appears in 6 of our 7 European countries—with variation in strength. Context matters for how tech risk translates into education preferences.
August 13, 2025 at 8:32 AM
Cross-country pattern: The VET “safe haven” pattern appears in 6 of our 7 European countries—with variation in strength. Context matters for how tech risk translates into education preferences.
Main result: Higher subjective tech risk → stronger preference for VET.
This is robust to accounting for education, occupation, and other socio-demographics. This is the baseline relationship we test throughout.
This is robust to accounting for education, occupation, and other socio-demographics. This is the baseline relationship we test throughout.
August 13, 2025 at 8:32 AM
Main result: Higher subjective tech risk → stronger preference for VET.
This is robust to accounting for education, occupation, and other socio-demographics. This is the baseline relationship we test throughout.
This is robust to accounting for education, occupation, and other socio-demographics. This is the baseline relationship we test throughout.
🚨 New OA article with Scherwin M. Bajka in @reggovjournal.bsky.social:
“Subjective Technology Risk and Education Preferences: VET as a Safe Haven or Dead End?” (link in last post)
When people fear job loss from new tech, do they prefer vocational training over academic tracks? Short answer: yes.
“Subjective Technology Risk and Education Preferences: VET as a Safe Haven or Dead End?” (link in last post)
When people fear job loss from new tech, do they prefer vocational training over academic tracks? Short answer: yes.
August 13, 2025 at 8:32 AM
🚨 New OA article with Scherwin M. Bajka in @reggovjournal.bsky.social:
“Subjective Technology Risk and Education Preferences: VET as a Safe Haven or Dead End?” (link in last post)
When people fear job loss from new tech, do they prefer vocational training over academic tracks? Short answer: yes.
“Subjective Technology Risk and Education Preferences: VET as a Safe Haven or Dead End?” (link in last post)
When people fear job loss from new tech, do they prefer vocational training over academic tracks? Short answer: yes.
Our article (with @madselk.bsky.social, @jacobnyrup.bsky.social, @benansell.bsky.social, @aslicansunar.bsky.social, and Laure Bokobza) on why it is so hard to counteract wealth inequality through inheritance taxation is out in @worldpolitics.bsky.social!
July 9, 2025 at 8:11 AM
Our article (with @madselk.bsky.social, @jacobnyrup.bsky.social, @benansell.bsky.social, @aslicansunar.bsky.social, and Laure Bokobza) on why it is so hard to counteract wealth inequality through inheritance taxation is out in @worldpolitics.bsky.social!
Last time slot, yet another full house: Stellar end to our series of #EPSA2025 panels on AI and the knowledge economy! Great to see this scholarly community emerging, please get in touch if you want to be involved in future initiatives!
June 28, 2025 at 3:31 PM
Last time slot, yet another full house: Stellar end to our series of #EPSA2025 panels on AI and the knowledge economy! Great to see this scholarly community emerging, please get in touch if you want to be involved in future initiatives!
Another packed #EPSA2025 panel with impressive papers by @hilmal.bsky.social, @canalejoalvaro.bsky.social, Dave Hope, @gaetanoinglese.bsky.social & @matteonebbiai.bsky.social. Great to see so much interest in questions around AI and the knowledge economy! One more to go tomorrow (3pm, room -1.A.03)!
June 27, 2025 at 10:42 AM
Another packed #EPSA2025 panel with impressive papers by @hilmal.bsky.social, @canalejoalvaro.bsky.social, Dave Hope, @gaetanoinglese.bsky.social & @matteonebbiai.bsky.social. Great to see so much interest in questions around AI and the knowledge economy! One more to go tomorrow (3pm, room -1.A.03)!
Great start to #EPSA2025 chairing a packed panel on "Regulation and Governance of AI," with papers by
@paoloagnolin.bsky.social, Alex Kuo, Brady Allardice, and Michael Becher (and @zackgp94.bsky.social discussing). We have three more exciting panels on the politics of AI coming up (see in reply):
@paoloagnolin.bsky.social, Alex Kuo, Brady Allardice, and Michael Becher (and @zackgp94.bsky.social discussing). We have three more exciting panels on the politics of AI coming up (see in reply):
June 26, 2025 at 10:51 AM
Great start to #EPSA2025 chairing a packed panel on "Regulation and Governance of AI," with papers by
@paoloagnolin.bsky.social, Alex Kuo, Brady Allardice, and Michael Becher (and @zackgp94.bsky.social discussing). We have three more exciting panels on the politics of AI coming up (see in reply):
@paoloagnolin.bsky.social, Alex Kuo, Brady Allardice, and Michael Becher (and @zackgp94.bsky.social discussing). We have three more exciting panels on the politics of AI coming up (see in reply):
📢Special issue and workshop on "The Politics of AI: Citizen Perceptions, Preferences, and Priorities" 📢
We aim to bring together cutting-edge research on the political ramifications of AI for a workshop in Oxford and a special issue in a leading political science journal.(1/2)
We aim to bring together cutting-edge research on the political ramifications of AI for a workshop in Oxford and a special issue in a leading political science journal.(1/2)
March 18, 2025 at 2:06 PM
📢Special issue and workshop on "The Politics of AI: Citizen Perceptions, Preferences, and Priorities" 📢
We aim to bring together cutting-edge research on the political ramifications of AI for a workshop in Oxford and a special issue in a leading political science journal.(1/2)
We aim to bring together cutting-edge research on the political ramifications of AI for a workshop in Oxford and a special issue in a leading political science journal.(1/2)
The CfP for our GOVPET PhD Workshop is out! Topic: "Skill mismatches and shortages: Discourse, policies and policies between education, labour markets and migration". If you are a PhD student working on these or related topics, please consider applying! #polisky
Link: ibb.unisg.ch/en/research-...
Link: ibb.unisg.ch/en/research-...
October 17, 2023 at 8:35 AM
The CfP for our GOVPET PhD Workshop is out! Topic: "Skill mismatches and shortages: Discourse, policies and policies between education, labour markets and migration". If you are a PhD student working on these or related topics, please consider applying! #polisky
Link: ibb.unisg.ch/en/research-...
Link: ibb.unisg.ch/en/research-...
But, reassuringly, this does not necessarily translate into increased performance inequality between these groups. While younger workers receive a greater boost than older ones, we find no equivalent pattern along the lines of sex, education, or occupation (6/9)
October 16, 2023 at 4:03 PM
But, reassuringly, this does not necessarily translate into increased performance inequality between these groups. While younger workers receive a greater boost than older ones, we find no equivalent pattern along the lines of sex, education, or occupation (6/9)
Thanks to our (near-representative) sample of the UK working-age population, we can provide a first look at group differences in frequency and benefits of AI use: the usual suspects (younger, male, more educated, professionals) are more likely to use AI at least weekly... (5/9)
October 16, 2023 at 4:02 PM
Thanks to our (near-representative) sample of the UK working-age population, we can provide a first look at group differences in frequency and benefits of AI use: the usual suspects (younger, male, more educated, professionals) are more likely to use AI at least weekly... (5/9)
But what about the equalising effects that several recent studies have found? Overall variation in performance declines in two of our three tasks, but increases in a task that is moderately complex but ambiguous with no clear right solution (3/9)
October 16, 2023 at 4:01 PM
But what about the equalising effects that several recent studies have found? Overall variation in performance declines in two of our three tasks, but increases in a task that is moderately complex but ambiguous with no clear right solution (3/9)
Link to the pre-print: papers.ssrn.com/sol3/papers....
Based on a survey experiment with a general population sample from the UK, we show that people perform a set of text-based tasks faster and between 0.4 SD and 0.9 SD better when exposed to ChatGPT. So far, so intuitive... (2/9)
Based on a survey experiment with a general population sample from the UK, we show that people perform a set of text-based tasks faster and between 0.4 SD and 0.9 SD better when exposed to ChatGPT. So far, so intuitive... (2/9)
October 16, 2023 at 4:01 PM
Link to the pre-print: papers.ssrn.com/sol3/papers....
Based on a survey experiment with a general population sample from the UK, we show that people perform a set of text-based tasks faster and between 0.4 SD and 0.9 SD better when exposed to ChatGPT. So far, so intuitive... (2/9)
Based on a survey experiment with a general population sample from the UK, we show that people perform a set of text-based tasks faster and between 0.4 SD and 0.9 SD better when exposed to ChatGPT. So far, so intuitive... (2/9)
🚨 New working paper 🚨
How do AI tools like ChatGPT affect productivity? Existing studies find equalising effects within occupations, but does this generalize to the labor market at large? My new working paper with Jane Gingrich & Jasmine Bhatia addresses this question: (1/9)
How do AI tools like ChatGPT affect productivity? Existing studies find equalising effects within occupations, but does this generalize to the labor market at large? My new working paper with Jane Gingrich & Jasmine Bhatia addresses this question: (1/9)
October 16, 2023 at 3:57 PM
🚨 New working paper 🚨
How do AI tools like ChatGPT affect productivity? Existing studies find equalising effects within occupations, but does this generalize to the labor market at large? My new working paper with Jane Gingrich & Jasmine Bhatia addresses this question: (1/9)
How do AI tools like ChatGPT affect productivity? Existing studies find equalising effects within occupations, but does this generalize to the labor market at large? My new working paper with Jane Gingrich & Jasmine Bhatia addresses this question: (1/9)