Madoń, Ka ol
A icle
The ela ionship be ween a i icial in elligence (AI)
exposu e and e u ns o educa ion
Cen al Eu opean Economic Jou nal (CEEJ)
P o ided in Coope a ion wi h:
Facul y o Economic Sciences, Uni e si y o Wa saw
Sugges ed Ci a ion: Madoń, Ka ol (2024) : The ela ionship be ween a i icial in elligence (AI)
exposu e and e u ns o educa ion, Cen al Eu opean Economic Jou nal (CEEJ), ISSN 2543-6821,
Sciendo, Wa saw, Vol. 11, Iss. 58, pp. 461-474,
h ps://doi.o g/10.2478/ceej-2024-0029
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/324626
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
ISSN: 2543-6821 (online)
Jou nal homepage: h p://ceej.wne.uw.edu.pl
To ci e his a icle
Madoń, K. (2024). The ela ionship be ween A i icial In elligence (AI)
exposu e and e u ns o educa ion. Cen al Eu opean Economic Jou nal,
11(58), 461-474.
DOI: 10.2478/ceej-2024-0029
To link o his a icle: h ps://doi.o g/10.2478/ceej-2024-0029
The ela ionship be ween
A i icial In elligence (AI)
exposu e and e u ns o educa ion
Ka ol Madoń
Open Access. © 2024 K. Madoń, published by Sciendo.
This wo k is licensed unde he C ea i e Commons A ibu ion 4.0 In e na ional License.
Ka ol Madoń
SGH Wa saw School o Economics, Al. Niepodległości 162, 02-554 Wa saw, Poland,
Ins i u e o S uc u al Resea ch (IBS), I ysowa 18c, 02-660 Wa saw, Poland
co esponding au ho : ka [email protected] g.pl
The ela ionship be ween A i icial In elligence (AI)
exposu e and e u ns o educa ion
Abs ac
This pape s udies he ela ionship be ween exposu e o a i icial in elligence (AI) and wo ke s’ wages ac oss Eu opean
coun ies. O e all, a posi i e ela ionship be ween exposu e o AI and wo ke s’ wages is ound, howe e i di e s
conside ably be ween wo ke s and coun ies. High-skilled wo ke s expe ience a highe wage p emiums ela ed
o AI- ela ed skills han middle- and low-skilled wo ke s. Posi i e associa ions a e concen a ed among occupa ions
mode a ely and highly exposed o AI (be ween he 6 h and 9 h decile o he exposu e), and a e weake among he leas
exposed occupa ions. Re u ns o AI- ela ed skills among high-skilled wo ke s a e e en highe in Eas e n Eu opean
Coun ies compa ed o Wes e n Eu opean coun ies. The he e ogenei y likely o igina es om he di e ence in o e all
labou cos s be ween coun y g oups. The esul s p esen ed in his s udy we e ob ained om he es ima ion o
Mince ian wage eg essions on he 2018 elease o he EU S uc u e o Ea ning Su ey.
Keywo ds
a i icial in elligence | wages | echnological change | Eu ope
JEL Codes
E24, J30, O33
1. In oduc ion
O e ecen decades, ad ancemen s in labou -sa ing
echnologies ha e igni ed ea s abou la ge-scale job
displacemen s, o igina ing om i s abili y o au oma e
an inc easing numbe o asks (Acemoglu & Res epo,
2019a, 2020). The p ima y conce ns ega ded
indus ial obo s and au oma ion and solely ouched
on epe i i e and ou ine manual asks. Hence, he
impac s o obo adop ion we e concen a ed mos ly
in he manu ac u ing sec o and in luenced he es
o he economy by second-o de e ec s (D. Au o &
Salomons, 2018). Howe e , he eme gence o new,
IT- ela ed echnologies like machine lea ning (ML),
a i icial in elligence (AI), and Gene a i e P e- ained
T ans o me s (GPTs) may lead o a mo e complica ed
ela ionships. AI, wi h i s gene al-pu pose echnology
cha ac e is ics and abili y o au oma e non- ou ine
cogni i e asks, also has he po en ial o a ec e e y
sec o o he economy (Lane & Sain -Ma in, 2021).
F om a heo e ical poin o iew, an associa ion
be ween AI and wo ke s’ wages is unclea . On he
one hand, AI has he po en ial o be a subs i u e o
labou . Acco ding o economic heo y, i should exe
nega i e p essu e on wo ke s’ wages as he demand
o human wo k dec eases. On he o he hand, AI
may complemen wo ke s aking ca e o epe i i e
and noncomplica ed asks, shi ing humans’ a en ion
owa d mo e challenging ac i i ies. In u n, i could
boos p oduc i i y and esul in highe wages. This
pape aims o desc ibe he ela ionships be ween AI
and employmen and wo ke s’ wages ac oss Eu opean
coun ies and answe whe he AI ends o complemen
o subs i u e human labou .
AI is a gene al-pu pose echnology no connec ed o
a pa icula ype o physical de ice, speci ic applica ion,
o sec o (B ynjol sson & McA ee, 2014). The mul i ude
o po en ial applica ions makes quan i ying po en ial
exposu e o AI a non- i ial ask. The e o e, one
o he li e a u e s ands conce ns he measu emen
and quan i ica ion o he po en ial in luence o new
echnology on human labou . The AI measu es de eloped
so a employ a ask-app oach, ini ially p oposed by
Acemoglu and Au o (2011). They compa e he asks
pe o med by wo ke s in pa icula occupa ions, assess
he impo ance o skills equi ed in hese occupa ions,
CEEJ • 11(58) • 2024 • pp. 461-474 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0029 463
and compa e hem wi h AI po en ial. Webb (2020)
ocuses on wo ke asks and de elops exposu e measu es
o obo s, compu e so wa e, and AI. He employs
na u al language p ocessing o quan i y he o e lap
be ween human asks desc ibed in he occupa ional
O*NET da abase and he capabili ies o he echnology
desc ibed in pa en s. Fel en e al. (2018) employ an
abili y-based app oach. They use he Elec onic
F on ie Founda ion (EFF) AI P og ess Measu emen
da ase o assess he p og ess pace a di e en ca ego ies
o AI ad ancemen s. Then hey map he AI ca ego ies
o he lis o skills ha he O*NET da abase uses o
desc ibe occupa ions. Nex , hey quan i y he po en ial
e ec o he AI on occupa ions aking in o accoun
he impo ance o pa icula skills in pe o ming each
job. F ey and Osbo ne (2017) c ea e a compu e isa ion
p obabili y measu e based on expe judgemen s and an
assignmen algo i hm. B ynjol sson and Mi chell (2017)
and B ynjol sson e al. (2018) de elop a measu e o
sui abili y o di e en wo k ac i i ies o ML. Then hey
calcula e ML exposu e sco e o asks and occupa ions.
The li e a u e conce ning impac s o echnology
ad ancemen s on employmen and wages was g ea ly
in luenced by he wo ks o Acemoglu and Res epo
(2018, 2019). They p o ided a concep ual ask-based
amewo k in which au oma ion ac s as he capi al
and akes o e asks so a pe o med by a labou e . I
ask p oduc ion wi h capi al inpu is cheape han ha
o labou , hen au oma ion will lead o a displacemen
e ec , which in u n pu s nega i e p essu e on
wages and employmen . Howe e , he e a e o he
e ec s o au oma ion, which may coun e balance
displacemen e ec s. Fi s , eins a emen e ec
may lead o ex ension o al eady pe o med asks
o c ea ion o comple ely new, mo e complex asks
in which labou e s ha e a compa a i e ad an age
(Acemoglu & Res epo, 2019a). Au o e al. (2022)
show ha he majo i y o cu en employmen is
in he new jobs, and so a augmen a ion e ec s
o echnological inno a ions p e ail. Howe e , in
ecen decades demand-e oding e ec s o au oma ion
in ensi ied compa ed o demand-inc easing e ec s o
augmen a ion inno a ions. Second, he p oduc i i y
e ec may inc ease he demand o asks ha canno
be au oma ed. Hence, e en i inc easing sha e o asks
is au oma ed, inc easing demand may coun e balance
he displacemen e ec (D. Au o & Salomons, 2018;
Goos e al., 2014). Thi d, new echnologies may
imp o e exis ing ones ins ead o displacing labou .
Thus, i should lead o he p oduc i i y e ec ins ead
o he displacemen e ec .
So a , he as majo i y o e idence on au oma ion
is limi ed o he impac o indus ial obo s on labou
ma ke s. O e all, obo adop ion dec eased employmen
and wages in he US commu ing zones and con ibu ed
o inc eased wage inequali y (Acemoglu & Res epo,
2019b, 2022). The demand o low- and mid-skilled
labou dec eased leading o nega i e wage p essu e
among hese wo ke s (Acemoglu e al., 2023; Acemoglu
& Res epo, 2022; D. H. Au o e al., 2006; DeCanio,
2016) and esul ed in labou ma ke pola isa ion (Goos
e al., 2014, 2014). Howe e , obo adop ion induced
a posi i e p oduc i i y shock and ipple e ec s.
Au oma ion ele a ed o al ac o p oduc i i y (TFP),
eal alue-added, and agg ega e demand (D. Au o &
Salomons, 2018; G ae z & Michaels, 2018; G ego y e
al., 2022). The o al impac s o obo adop ion on he
economy emains posi i e. The o he s eam o he
li e a u e ela es o in o ma ion and communica ion
echnology (ICT) in es men s, e.g., he adop ion o
ICT echnologies led o a dec ease in employmen
and wage bills o olde women and p ime-aged men
in Eu opean coun ies (Albinowski & Lewandowski,
2024). Je bashian (2021) showed ha a dec ease in IT
equipmen p ices inc eased demand o high wage
occupa ions and educed demand o medium wage
occupa ions. Michaels e al. (2014) showed ha ICT
exposu e led o shi in demand o highly educa ed
wo ke s om middle educa ed wo ke s, esul ing in
ma ke pola isa ion.
None heless, indus ial obo s may be employed
in a ela i ely small ange o indus ies, loca ed
mainly in he manu ac u ing sec o (e.g. au omo i e
indus y, elec onics). I is unknown i i s impac s
on he economy may be gene alised o o he sec o s
and echnologies. Ye , he e is li le e idence o AI’s
impac on he economy pe se, as his echnology is
ela i ely new. Acemoglu (2021a, 2021b) claims ha AI
ad ancemen s may be he nex phase o au oma ion and
may u he dep ess wages and labou sha e. Excessi e
in es men in AI may lead o non-ma ke nega i e
ex e nali ies, as i ms aiming o educe labou cos s
do no conside he nega i e impac o au oma ion on
wo ke s, e.g., wo ke s being o ced o ake up lowe -
paid jobs o echnological unemploymen (G owiec,
2023). Howe e , o egoing esea ch on AI impac s
on labou ma ke is op imis ic. Fossen e al. (2022)
ound a posi i e associa ion be ween AI and wo ke s’
wage changes in se ices in he US. The au ho s
claim ha p oduc i i y e ec s and ask enhancemen
coun e balanced he displacemen e ec . Fossen
e al. (2022) ound ha AI ela es o imp o ed job
s abili y and wage g ow h, bu he associa ion is
CEEJ • 11(58) • 2024 • pp. 461-474 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0029 464
he e ogeneous. Engbe g e al. (2023) ound ha wage
g ow h p emia associa ed wi h AI exposu e wi h
con as ing wage g ow h penal ies ela ed o obo
exposu e in Ge many. Among c oss-coun y s udies
e e ing o AI impac s o wo ke -le el ou comes,
Pa eka e al. (2024) in es iga e he ela ionship
be ween wo ke s’ wages and non-mone a y ac o s
and wo ke s’ wellbeing. They ound ha wo ke s
exposed o AI expe ience be e wo king condi ions
han wo ke s exposed o obo s and so wa e ac oss
Eu opean coun ies.
The eme ging li e a u e on he impac s o AI
on he labou ma ke s shows no nega i e impac on
labou demand (G een, 2023). I is es ima ed ha AI
wo ke s cons i u e only 0.3% o OECD employmen
(G een & Lamby, 2023). S udy shows ha employmen
in occupa ions exposed o AI g ew be ween 2012-
2019, bu no causal link was es ablished (Geo gie &
Hyee, 2021).
2. Da a and desc ip i e e idence
In his sec ion, he da a used in his s udy we e
p esen ed and de ails o measu emen o echnology
exposu e we e p o ided.
2.1. Da a and measu emen
The main sou ce o da a is he 2018 elease o he
Eu opean S uc u e o Ea nings Su ey (EU-SES),
which is he mos eliable sou ce o c oss-coun y da a
on wages in he EU. This la ge employe -employee-
ma ched su ey da a p o ide ha monised in o ma ion
on ea nings ac oss Eu opean coun ies. I includes
in o ma ion on wo ke ea nings, as well as indi idual,
job, and i m cha ac e is ics. The da a co e en e p ises
employing mo e han en employees. The s udy co e s
20 Eu opean coun ies, di ided in o wo g oups—
Wes e n Eu opean coun ies: Belgium, Ge many,
G eece, Finland, F ance, I aly, Ne he lands, No way,
Po ugal, Spain, and Sweden; and Eas e n Eu opean
coun ies: Bulga ia, Czech Republic, Es onia, Li huania,
La ia, Poland, Romania, Slo akia, and C oa ia.
To measu e he exposu e o a i icial in elligence,
AI-sco es p oposed by E. Fel en e al. (2021) a e used.
The AI exposu e sco es o occupa ions a e calcula ed
based on he abili ies which a e equi ed o wo k in a
pa icula job (abili ies come om O*NET da abase).
AI exposu e does no di e ac oss ime, only be ween
occupa ions. In his s udy, exposu es o AI a he
Eu opean occupa ional le el (ISCO08 2-digi le el)
a e calcula ed. This measu e akes alues om -1.6 o
1.5 and is s anda dised ac oss occupa ions. The e o e,
posi i e alues cha ac e ize occupa ions ha a e mo e
exposed o AI han a e age occupa ion.
This s udy uses measu es o ade and globalisa ion
a he coun y-indus y le el om he EORA da abase
(Lenzen e al., 2012, 2013). In pa icula , log expo s,
as well as backwa d and o wa d global alue chains
(GVCs), pa icipa ion a e used. To calcula e he
measu es o pa icipa ion in GVC, he me hodology
o Bo in and Mancini (2015, 2019) is applied. The
da a on sec o al ou pu and employmen come om
Eu os a .
2.2. Desc ip i e analysis
Exposu e o a i icial in elligence di e s be ween
socio-economic g oups. Figu e 1 p esen s he AI
exposu e acco ding o age and educa ion le el.
Wo ke s aged 50-59 yea s old a e he leas exposed
g oup, and he a e age AI exposu e is 0.05 (Figu e
1, op panel). The AI exposu e is he highes among
wo ke s ages 30-39 yea s old and o als 0.19. The
a e age AI exposu e among he emaining age g oups
is simila and anges be ween 0.07-0.10.
The AI exposu e acco ding o educa ion le el is
inc easing mono onously— he highe he educa ion,
he highe he exposu e (Figu e 1, bo om panel). The
a e age exposu e among e ia y educa ed wo ke s
(0.89) is signi ican ly highe han among p ima y
educa ed (-0.60) and seconda y educa ed wo ke s
(-0.18). In ac , e ia y educa ed wo ke s pe o m
occupa ions ha a e mo e exposed o AI han he
a e age occupa ion. I means ha be e skilled
wo ke s end o pe o m occupa ions ha a e mo e
exposed o AI han less skilled wo ke s.
The a e age AI exposu e inc eases along wage
dis ibu ion (Figu e 2). The highe he log wage
pe cen ile, he highe he a e age AI exposu e.
Wo ke s ea ning less han he median end o
pe o m occupa ions ha a e less exposed o AI han
he a e age occupa ion. I sugges s ha occupa ions
ha a e mo e exposed o a i icial in elligence pay,
on a e age, highe wages. I is also in line wi h he
inc ease in AI exposu e wi h espec o educa ion.
The e is almos no co ela ion be ween he change
in log wages and AI exposu e, especially in Wes e n
CEEJ • 11(58) • 2024 • pp. 461-474 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0029 465
Eu opean coun ies. The e is some weak nega i e
associa ion (-18%) in Eas e n Eu opean coun ies
(Figu e 3).
3. Me hodology
As a i s model, he ollowing Mince ian wage
eg ession is es ima ed:
(1)
wijsc=
β
0+
β
1AIj+
β
2Eijsc+
β
3ageijsc+
β
4age2
ijsc+
β
5expijsc+
β
6sexijsc+
β
7Msc+
λ
c+
μ
s+
ϵ
ijsc
whe e, wijsc is he log wage o indi idual i in occupa ion
j in sec o s in coun y c. AIj measu es he occupa ional
exposu e o AI desc ibed in he da a sec ion. Eijsc is he
educa ion le el, ageijsc is he age g oup ca ego y, expijsc is
he expe ience in he company, and sexijsc is he gende .
Msc s ands o mac oeconomic con ols (expo s,
alue added, ou pu , and GVC pa icipa ion),
λ
c is o
coun y ixed e ec s, and
μ
s is o sec o ixed e ec s.
To add ess po en ial omi ed a iable bias and con ol
o a e age wage di e ences be ween coun ies
and sec o s, he coun y and sec o ixed e ec s a e
included. The key coe icien o in e es in he wage
eg ession is
β
1, pe aining o he occupa ion-le el AI
exposu e.
Figu e 1. Exposu e o AI by age g oups and educa ion le el
Sou ce: au ho s’ elabo a ion based on EU SES, EORA, E. Fel en e al. (2021) da a
CEEJ • 11(58) • 2024 • pp. 461-474 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0029 466
To accoun o he he e ogenei y o he ela ionship
be ween wo ke s ac oss hei skills, he in e ac ion
be ween AI exposu e and wo ke s’ educa ion is
included. Educa ion le el is a good p oxy o wo ke
skills. In he ollowing sec ions, wo ke s wi h e ia y
educa ion will be e e ed o as high-skilled wo ke s,
wo ke s wi h seconda y educa ion as middle-skilled
wo ke s, and wo ke s wi h p ima y educa ion as
low-skilled wo ke s. To accoun o po en ial non-
linea i ies in associa ion be ween AI exposu e and
wo ke s’ wages, squa ed exposu e and i s in e ac ions
a e also included. Fo mally:
wijsc=
β
0+
β
1AIj ×AIj×Eijsc+
β
2ageijsc+
β
3age2
ijsc+
β
4expijsc+
β
5 sexijsc+
β
6 Msc+
λ
c+
μ
s+
ϵ
ijsc
(2)
whe e × is a Ca esian p oduc o in ol ed a iables.
The key coe icien o in e es is
β
1, which ep esen s
he combina ion o coe icien s o AIj, AIj
2, and i s
in e ac ions wi h educa ion le el dummies. Depending
on model speci ica ion,
β
1 may be in e p e ed as a
wage p emium ela ed o AI skills in a pa icula
educa ional g oup (AI ela ed e u ns o educa ion) o
as e u ns o educa ion along he dis ibu ion o AI
exposu e.
Figu e 2. Exposu e o AI by log wage pe cen ile
Sou ce: au ho s’ elabo a ion based on EU SES, EORA, E. Fel en e al. (2021) da a
Eas e n Eu ope Wes e n Eu ope
Figu e 3. Co ela ion be ween change in log wage be ween 2010-2018 and AI exposu e, in Eas e n and Wes e n
Eu opean coun ies
Sou ce: au ho s’ elabo a ion based on EU SES, EORA, E. Fel en e al. (2021) da a
CEEJ • 11(58) • 2024 • pp. 461-474 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0029 467
A p oposed es ima ion app oach has limi a ions.
None o he p esen ed models can p o ide causal
e idence o AI exposu e on wo ke s’ wages. Fi s ly, AI
exposu es measu e he ulne abili y o occupa ions
o AI bu no he ac ual ea men . I he e o e shows
which occupa ion, and wo ke s, may be a ec ed by AI,
bu i does no mean hey necessa ily a e. Secondly, o
es ima e he causal e ec s o AI on wo ke s’ wages,
one should use i m-le el da a wi h in o ma ion
abou echnology adop ion and ime pe iod o he
da a. Hence, esul s p esen ed in his s udy should be
in e p e ed wi h cau ion and ea ed like a desc ip ion
o inal alloca ion is in equilib ium, a he han causal
ela ionships.
4. Resul s
In his sec ion, es ima ion esul s ega ding
ela ionship be ween AI exposu e and wo ke s’ wages
a e p esen ed.
4.1. Rela ionship be ween AI and
wo ke s’ wages
Posi i e associa ion be ween AI exposu e and wo ke s’
log wages a e ound in all eg ession speci ica ions
(Table 1). Acco ding o a baseline speci ica ion o he
model (Table 1, column 1), he ela ionship be ween
AI exposu e and log wages is 0.082. This means ha
he change in AI exposu e by one sample s anda d
de ia ion will be associa ed wi h an inc ease in log
wages by 0.082 log poin s.
Adding con ols o coun y ixed e ec s inc eases
his associa ion o 0.091 log poin s. I e eals ha he
ela ionship be ween AI exposu e and wo ke s’ wages
may no be linea due o a nega i e and signi ican
coe icien o squa ed exposu e (Table 1, column 2).
Reg ession esul s con i m he associa ions s a ed in
he desc ip i e esul s sec ion. Wo ke s pe o ming
occupa ions ha a e mo e exposed o AI end o ea n
highe wages.
Howe e , he ela ionship be ween AI exposu e
and wo ke s’ wages may di e be ween ypes o
wo k. The e o e, connec ing he AI exposu e wi h
wo ke s’ educa ion le el app oxima es wo ke skills.
Reg ession esul s p esen ed in columns (3) and (4)
o Table 1 sugges ha he ela ionship be ween
AI exposu e and wo ke s’ wages di e s ac oss he
dis ibu ion o wo ke skills. Figu e 4 p esen s he
a e age associa ion be ween he AI exposu e and
wo ke s’ wages among all educa ion g oups. This
coe icien can be in e p e ed as he wage p emium
associa ed wi h AI in di e en wo ke skills g oups.
While he size o he coe icien is no e y di e en
be ween p ima y and seconda y educa ed wo ke s, i
is much highe among he highe educa ed. Resul s o
he model sugges ha AI exposu e is complemen a y
o high-skilled wo ke s, which con ibu es o highe
wage p emia.
Figu e 5 p esen s he es ima ed e u ns o
educa ion ac oss he AI exposu e dis ibu ion o
seconda y and e ia y educa ion (p ima y educa ion
is a e e ence g oup). Among wo ke s wi h seconda y
educa ion, he e u ns o educa ion a y be ween
0.09-0.14, wi h he highes alues in he middle o AI
exposu e dis ibu ion. Re u ns o e ia y educa ion
a e much highe han in he case o seconda y
educa ion and ake alues be ween 0.15-0.43.
Simila ly, he highes e u ns a e also concen a ed
in he middle o AI exposu e dis ibu ion. I sugges s
nonlinea i y in he associa ion be ween wo ke s’
wages and AI exposu e.
Rela ionships es ima ed in his sec ion a e in line
wi h he o egoing li e a u e. O e all, esul s ob ained
in his s udy sugges ha AI exposu e is posi i ely
associa ed wi h wo ke s’ wages (Engbe g e al., 2023;
Fossen e al., 2022; Pa eka e al., 2024). Mo eo e , he
ela ionship concen a es among high-skilled wo ke s
who bene i om AI he mos (Fossen e al., 2022;
Je bashian, 2021).
4.2. Rela ionship be ween AI and
wo ke s’ wages be ween coun y
g oups
Nex , he sample is di ided in o Eas e n and
Wes e n Eu opean coun ies. The o e all di ec ion
o associa ions esembles hose in he ull sample
(Table 2). The poin es ima es o AI exposu e end
o be highe in Eas e n Eu opean coun ies han in
Wes e n Eu opean coun ies. Signi ican coe icien s
o squa ed AI exposu e con i m he nonlinea i y o
associa ion in bo h g oups.
Tu ning o he ma ginal e ec s o AI exposu e a
he educa ion le el e eals subs an ial he e ogenei y
be ween coun ies and wo ke s. In bo h coun y
g oups, he AI- ela ed wage p emium is conside ably
CEEJ • 11(58) • 2024 • pp. 461-474 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0029 468
Table 1. The ela ionship be ween AI exposu e, skill le el (educa ion), and wo ke s’ wages, s anda dised
(1) (2) (3) (4) (5) (6)
AI Exposu e (s d.) 0.028*** 0.063*** 0.025*** 0.084*** 0.122*** 0.017***
(0.003) (0.003) (0.002) (0.004) (0.003) (0.003)
AI Exposu e 2 (s d.) 0.010*** -0.001 0.027*** 0.046*** 0.043*** 0.027***
(0.002) (0.002) (0.001) (0.003) (0.003) (0.002)
AI Exposu e (s d.) X Seconda y
Educa ion dummy
-0.039*** -0.050*** -0.009***
(0.002) (0.002) (0.002)
AI Exposu e (s d.) X Te ia y
Educa ion dummy
-0.080*** -0.080*** 0.036***
(0.003) (0.003) (0.002)
AI Exposu e 2 (s d.) X Seconda y
Educa ion dummy
-0.022*** -0.038*** -0.015***
(0.002) (0.002) (0.001)
AI Exposu e 2 (s d.) X Te ia y
Educa ion dummy
-0.043*** -0.048*** -0.001
(0.003) (0.003) (0.002)
Seconda y Educa ion dummy -0.065*** -0.041*** 0.072*** -0.069*** -0.034*** 0.085***
(0.002) (0.001) (0.001) (0.003) (0.002) (0.002)
Te ia y Educa ion dummy 0.131*** 0.165*** 0.226*** 0.170*** 0.200*** 0.205***
(0.002) (0.002) (0.002) (0.003) (0.003) (0.002)
Occupa ion FE. Yes Yes Yes Yes Yes Yes
Sec o FE. No Yes Yes No Yes Yes
Coun y FE. No No Yes No No Yes
No. o Obse a ions 7.9 M 7.9 M 7.9 M 7.9 M 7.9 M 7.9 M
No e: S anda d e o s in pa en heses. *** p<0.01, ** p<0.05, * p<0.1
Sou ce: au ho s’ elabo a ion based on EU SES, EORA, E. Fel en e al. (2021) da a.
Figu e 4. The es ima ed ela ionship be ween AI exposu e and wo ke s’ wages in p ima y, seconda y, and e ia y
educa ed g oups
Sou ce: au ho s’ elabo a ion based on EU SES, EORA, E. Fel en e al. (2021) da a
