ARTICLE
Human-made news s AI-gene a ed news: a
compa ison o Po uguese and Spanish jou nalism
s uden s’e alua ions
João Ped o Bap is a1,2✉, Rubén Ri as-de-Roca 3, Anabela G adim4& Concha Pé ez-Cu iel5
This s udy examines how uni e si y s uden s in Po ugal and Spain pe cei e he quali y o
jou nalis ic news p oduced by gene a i e a ificial in elligence (AI). As AI—pa icula ly
gene a i e models like Cha GPT-3—becomes inc easingly in eg a ed in o news dis ibu ion
and news oom ope a ions, i o e s he p omise o g ea e e ficiency and he educ ion o
non- alue-added asks. Howe e , i s ise also aises conce ns abou po en ial isks, e hical
implica ions, and he e ol ing ela ionship be ween human jou nalis s and AI-d i en con en
c ea ion. The s udy ocuses on he pe cep ions o jou nalism s uden s, add essing key
esea ch ques ions ela ed o hei abili y o disce n AI-gene a ed news, compa ing he
quali y be ween AI-gene a ed news and news w i en by jou nalis s, and he influence o
news opics in assessing quali y. We applied wo su eys o 444 pa icipan s, e alua ing
dimensions o news quali y, such as eadabili y, in o ma i eness and s yle. The findings
indica e ha s uden s gene ally a ed news gene a ed by Cha GPT-3 as being o highe
quali y han news w i en by jou nalis s. S uden s in Spain ga e highe a ings o AI-
gene a ed news. The findings sugges a po en ial shi in pe cep ions o news quali y, p o-
mo ing discussions abou he e ol ing ole o AI in jou nalism educa ion and i s b oade
implica ions o he indus y.
h ps://doi.o g/10.1057/s41599-025-04872-2 OPEN
1Depa men o Le e s, A s and Communica ion, Uni e si y o T ás-os-Mon es and Al o Dou o, 5000 Vila Real, Po ugal. 2CICANT—The Cen e o
Resea ch in Applied Communica ion, Cul u e, and New Technologies, Lusó ona Uni e si y, 1749 Lisbon, Po ugal. 3Depa men o Communica ion Sciences,
Facul y o Communica ion Sciences, Uni e si y o San iago de Compos ela, 15782 San iago de Compos ela, Spain. 4Labcom, Uni e si y o Bei a In e io ,
6200 Co ilhã, Po ugal. 5Depa men o Jou nalism II, Facul y o Communica ion, Uni e si y o Se ille, 41092 Se ille, Spain. ✉email: [email p o ec ed]
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2 1
1234567890():,;
In oduc ion
Algo i hms play a c ucial ole in he con empo a y media
ecosys em, significan ly influencing he dis ibu ion o
news con en . The use o a ificial in elligence (AI) is one
o he main echnological inno a ions in jou nalism, being
unde s ood as he peak o au oma ion in in o ma ion
(Diakopoulos 2020) and he la es example o he Fou h
Indus ial Re olu ion (Six o-Ga cía e al. 2024). While jou nalis s
adi ionally held exclusi e au ho i y in news c ea ion, gene a i e
a ificial in elligence (AI) could now play a pi o al ole no only
in dis ibu ion bu also in he day- o-day ope a ions o news-
ooms. Humans a e no longe solely esponsible o selec ing and
w i ing news; he digi al e olu ion has gi en ise o obo jou -
nalis s (Lemelsh ich La a 2018).
The use o AI causes ensions in he news indus y as he e a e
se e al nuances in scope and use (Mo an and Shaikh 2022). Some
news ooms a e skep ic o his echnology, bu a he same ime AI
ools a e en e ing in o he wo k ou ines o di e en news media
ou le s wo ldwide (Gu ié ez-López e al. 2023; Muno iya wa
e al. 2023). In he field o jou nalism, au oma ion is no new as
di e en ini ia i es ha e eso ed o his echnology o assis
news oom p ac ices (Cle wall 2014). Besides ha , AI is a e m
ha in ol es di e en echnologies such as a machine lea ning o
na u al language p ocessing, which ha e a long his o y in shaping
a hi- ech jou nalism (Ande son 2018).
Howe e , he launch o Cha GPT a he end o 2022 has
opened a new e a o AI, making i possible o au oma e some
jou nalis ic p ocesses (Gu ié ez-Caneda e al. 2023). Cha GPT is
a gene a i e AI ool de eloped by OpenAI company. I is based
on a simple use in e ace, which is a ailable in di e en lan-
guages. As gene a i e AI is a so o AI ha can p oduce se e al
ypes o con en , i s possibili ies in jou nalism and educa ion a e
as (Pa lik 2023). I should be no ed ha au oma ion implies o
se up p e-defined ules, whe eas AI makes i s own decisions
h ough machine-lea ning models.
Al hough he use o obo s has been expe imen al and limi ed
o some ac o s o he news media sec o (de-Lima-San os and
Ce on 2021), ea ly s udies e eal ha AI imp o es jou nalis ic
co e age by educing epo e s’non- alue-added asks, especially
in he a eas o Spo and Finance (Cana ilhas 2022). Po ai s o
AI in media ou le s in ecen yea s poin o a news co e age ha
no malizes his kind o echnology, mixing op imis ic and pes-
simis ic app oaches (Cools e al. 2024). Ne e heless, he li e a-
u e also de ec ed how gene a i e AI is li le used o c ea e ex s,
oge he wi h a lack o AI expe s in he news ooms (Quian and
Six o-Ga cía 2024).
The analyses o his hi- ech jou nalism a e usually ma ked by
discussions on hei isks, e en in he news media. The e a e c i-
icisms o he inabili y o dis inguish ue and alse con en due o
he p oli e a ion o AI-enabled disin o ma ion p oduc s (Why e
2020) and he lack o in o ma ion abou sou ces (González-A ias
and López-Ga cía 2023), oge he wi h e hical issues, since AI
b ings a new ela ionship be ween humans and machines. Mo e-
o e , i is impo an o acknowledge he di e ence be ween ‘ ake
news’, as s o ies w i en o con ey alse o misleading in o ma ion,
and he inciden al hallucina ions in oduced by gene a i e AI. Bo h
phenomena cause p oblems in news c edibili y.
As a esul , he deba e su ounding AI in jou nalism has
spa ked a ying opinions in li e a u e. Some a gue ha AI se es
as a aluable ool o news o ganiza ions, eeing jou nalis s om
epe i i e asks and enabling he c ea ion o mo e complex
epo s o inc eased in es men in in es iga i e jou nalism.
O he s exp ess pessimism owa ds he u u e o he sec o . Aca-
demia and he p o ession i sel end o emphasize ins umen al
app oaches o AI, ueling compu e -assis ed epo ing (Deuze
and Becke 2022).
Along wi h hese deba es, Jou nalism educa ion has also been
a ec ed by he ise o AI. Acco ding o Wes lund and Lewis
(2017), s uden s need o lea n algo i hm managemen wi h AI in
highe educa ion. These echnological ans o ma ions in oduce
changes in news p oduc ion, based on news ou ines and quan-
i a i e con en s ha may equi e ins uc ion in p omp engi-
nee ing. Jou nalism p o esso s unde line he ole o obo s as a
helping hand in he news oom (Cal o-Rubio and U a e-Ruiz
2020), bu p io schola ship has paid li le a en ion o how less
human edi ing is being aken ad an age o p omo e disin o ma-
ion messages and low-quali y news. Ou esea ch examines he
pe cei ed quali y o news gene a ed by Cha GTP-3 among
jou nalism s uden s in Po ugal and Spain. This esea ch ollows
up on ano he s udy ocused exclusi ely on Po ugal (Bap is a
and G adim 2024).
AI and he aining o jou nalis s in Spain and Po ugal
Algo i hmic jou nalism p o ides audience-o ien ed con en , since
he pa e ns o he public a e seen as quan ifiable h ough big da a
(Ande son 2013). The use o AI o assis epo ing is ano he
example o he quan i a i e u n ha is now p e alen in p o-
essional jou nalism (Codding on 2015). The massi e use o his
echnology equi es a p ope aining on algo i hms and jou -
nalism, wi h a special ocus on echnical and e hical compe encies
(Lopezosa e al. 2023). Ne e heless, some e idence poin s ou he
scan p esence o AI in he cu icula o communica ion Bachelo
and Mas e deg ees in Spain (Gómez-Diago 2022).
Ou s udy add esses Spain and Po ugal as bo h coun ies belong
o he same media sys em, which explains some p e ious com-
pa a i e analysis o he Ibe ian news landscape in e ms o disin-
o ma ion (Bap is a e al. 2023). The li e a u e has iden ified some
success ul AI ini ia i es in hese coun ies, using case s udies and
su eys as me hods. These p ac ices on au oma ed jou nalism a e
s ill explo a o y bu seem o be mo e de eloped in Spain.
Acco ding o esea ch ca ied ou by Cal o-Rubio and U a e-
Ruiz (2020) wi h heads o news media, jou nalis s, uni e si y
p o esso s, and s uden s o Communica ion S udies in Spain,
hese s akeholde s a e con inced AI will no ha e a nega i e
impac on he jou nalis ic sec o . These findings con as wi h he
p e iously iden ified lack o knowledge abou AI among Spanish
jou nalis s (Túñez-López e al. 2018) and highligh he need o
solid aining in au oma ed jou nalism (Cal o-Rubio and U a e-
Ruiz 2020).
The conce n o e limi ed AI knowledge is also e iden in
Po ugal, whe e su eys o spo s jou nalis s indica e ha eco-
nomic cons ain s p e en i s adop ion in news ooms (Cana ilhas
2022). This lack o financial esou ces is a pa e n in o he
Po uguese-speaking coun ies such as B azil, bu he B azilian
media eso mo e o au oma ed jou nalism, pa icula ly in he
field o spo s (Cana ilhas and Giacomelli 2023). Unsu p isingly,
AI acili a es e ficien wo k when epo ing ou ine ac i i ies like
spo s sco es.
Beyond specific a eas, he applica ion o ques ionnai es o
jou nalis s o he main daily newspape s in Po ugal (Co eio da
Manhã, Diá io de No ícias, Jo nal de No ícias, and Público)
highligh ed no only he poo knowledge o AI and he equi ed
aining on he subjec , bu also ha Po uguese jou nalis s a e
open o using his echnology (Gonçal es and Melo 2022).
Compa a i e s udies on he use o AI in he Ibe ian Peninsula a e
ew, bu al eady show au oma ed jou nalism is mo e a ailable
wi hin Spanish news ooms; meanwhile, in Po ugal he highe
educa ion ins i u ions o Google’s Digi al News Inno a ion Fund
play a key ole in os e ing compu e -assis ed epo ing (Pé ez-
Seijo e al. 2023).
ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2
2HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2
Ano he key di e ence be ween Po ugal and Spain is he use
o AI-d i en ools in Spain o p oduce pe sonalized con en o
local communi ies (Noain-Sánchez, 2022). Despi e globaliza ion,
people s ill seek in o ma ion abou hei neighbo hoods and il-
lages. In his con ex , he au oma ic gene a ion o news om
da ase s can help mee in o ma ion needs ha adi ional news-
ooms ha e o e looked, as happened wi h he p ojec o au o-
ma ed news co e age by RTVE (Spanish public b oadcas ing
se ice) o he 2023 municipal elec ions (A ambu ú Moncada
e al. 2023).
Algo i hmic jou nalism encompasses a ious AI echniques
ha enable compu a ional news disco e y (Diakopoulos 2020).
These AI-d i en jou nalis ic insigh s ha e implica ions o con-
cep s such as newswo hiness and quali y, unde sco ing he need
o p ope aining in AI managemen . In Spain, jou nalis s began
o ecei e aining on how o apply au oma ion, bu he
ins uc ion on his ma e is s ill a ea ly s ages (U a e e al.
2020a). As a solu ion, hey p opose he inclusion o gene a i e AI
on Jou nalism deg ees cu icula.
E en hough au oma ed jou nalism has gained g ound, he
educa ional aining in AI a Spanish uni e si ies is insu ficien ,
wi h only ew p oposals ha ackle he echnological and e hical
compe ences equi ed o u u e jou nalis s (U a e e al. 2020b).
Teaching ini ia i es o inco po a e AI in o uni e si y cou ses
emain limi ed bu ollow wo main app oaches: a c i ical pe -
spec i e, which examines he e hical and social consequences o
au oma ion, and an applied pe spec i e, which ocuses on da a
p ocessing and con en c ea ion (Gómez-Diago 2022), in which
ins uc ion in p omp enginee ing could also be needed.
The li e a u e has no conside ed he absence o AI aining in
bachelo ’s and mas e ’s deg ees in communica ion in Po ugal,
whe e he ocus is much mo e o ien ed o media li e acy, com-
bining he ac ion o p ima y and seconda y educa o s and jou -
nalis s o figh disin o ma ion a he local le el. In hese
ini ia i es, jou nalis s ac as ainees and aine s (Foà e al. 2023),
o ins ance sha ing e ifica ion p o ocols. In imes o pla o -
miza ion, con inuous aining is essen ial o bo h jou nalis s and
s uden s.
Media li e acy equi es s eng hening knowledge o bo h adi-
ional and digi al echnologies (Tomé e al. 2023). Howe e , as AI is
inc easingly used o gene a e alse con en (Ka inshak and Jin 2023),
i s ole in disin o ma ion is eshaping he jou nalism indus y.
Despi e his, a esea ch gap emains in add essing AI wi hin uni-
e si y educa ion, pa icula ly ega ding s uden s’pe cep ions.
Acco dingly, ou s udy poses he ollowing esea ch ques ions:
RQ1 - How well can jou nalism s uden s dis inguish AI-
gene a ed news om human-w i en news?
RQ2 - Does he news subjec influence he quali y a ibu ed o
bo h news?
Ma e ials and me hods
This s udy compa es how uni e si y s uden s in Po ugal and
Spain e alua e he ex ual quali y o news p oduced by Cha GPT-
3 o c ea ed by jou nalis s. Two class oom su eys we e dis-
ibu ed sepa a ely in he espec i e coun ies and a di e en
pe iods du ing he fi s semes e o he academic yea 2023–2024.
All pa icipan s esponded olun a ily and anonymously o he
su eys. The explo a o y sample consis ed o a o al o 444 s u-
den s, 99 in Po ugal and 345 in Spain, conside ing ha he
Spanish uni e si y sys em has much mo e s uden s han he
Po uguese. The subjec s a e aged be ween 18 and 65 yea s old,
57.9% a e emale and 41.9% a e male; whe eas he esponse a e
was 100% since he ac i i y ook place in he class ooms and all
he a ge s uden s pa icipa ed.
In he Ibe ian coun ies a ound 4000 jou nalism s uden s
g adua e each yea (Minis e io de Ciencia, Inno ación y
Uni e sidades 2024; Cana ilhas and Di Fá ima 2025). Ou
esea ch ocuses on his sample, and no a wide popula ion, since
hese s uden s a e u u e jou nalis s and hei pe cep ions on
gene a i e AI may ha e implica ions in jou nalis ic p ac ices,
including key concep s such as e hics o news quali y. Spain and
Po ugal we e selec ed because hey a e coun ies wi h a cul u al
p oximi y and belong o he same media sys em (Hallin and
Mancini 2004), e en hough he Spanish poli ical sphe e is much
mo e pola ized, which led us o expec some di e ences. P io
compa a i e s udies ha e shown how some Po uguese news
media a e mo e likely o adop echnological ools (Ri as-de-Roca
2022), in which AI migh be included.
The s udy is based on h ee p ocedu es. Fi s , he e was a
selec ion and collec ion o news w i en by jou nalis s. Second, a
se o p omp s we e ex ac ed om he news o Cha GPT-3 o
c ea e he same news. The p omp s consis ed o six keywo ds
ex ac ed om he o iginal news, ollowing he answe o he
adi ional jou nalis ic Lead ques ions, he six W’s (who, wha ,
when, whe e, why, how). Thi d, wo su eys we e c ea ed wi h
di e en languages, in Po uguese and Spanish.
P ocedu es. 1) Selec ion and collec ion o news
We chose o selec news om wo newspape s wi h na ional
co e age in bo h coun ies. In Po ugal, he sample was collec ed
om he websi e o he newspape ‘Co eio da Manhã’(h ps://
www.cmjo nal.p ) and, in Spain, om he digi al newspape
‘OKdia io’(h ps://okdia io.com). We chose hese wo news-
pape s because hey emphasize he same news alues, such as
celeb i ies, c imes, acciden s and cul u e. Pa icula ly, hey we e
selec ed due o hei sensa ionalis app oach, based on he
assump ion ha biased in o ma ion and clickbai p ac ices could
be easily eplaced by AI.
Bo h media ou le s a e cha ac e ized by a sensa ionalis agenda,
bu Co eio da Manhã is a long-es ablished newspape ( ounded
in 1979) ha has a p in e sion, while OKdia io is a digi al na i e
c ea ed in 2015 by Edua do Inda, a well-known con o e sial
jou nalis in Spain. Despi e some ailed a emp s, he Spanish
media landscape does no ha e p in ed abloid p ess, which is no
he case in mos Eu opean coun ies.
P in news ela ed o celeb i ies, c ime, acciden s and cul u e
we e selec ed and hen eplica ed on Cha GPT-3. These a e opics
associa ed wi h so news and a e o en less complex and less
subjec o opinion, c i icism and judgmen . Topics such as c ime
and acciden s esul in mo e ac ual news epo ing and we e
hough mo e sui able o c ea e p omp s o Cha GPT-3.
Fo he ques ionnai e, we andomly chose one news i em, in
Po uguese and Spanish, o each opic. The p omp s we e c ea ed
based on keywo ds closely ela ed o he w i ing echnique o news
leads. The p omp keywo ds answe ed o he six classic jou nalis ic
ques ions (who, wha , whe e, why, how, when). Figu e 1is a
sc eensho ha exemplifies how he p omp s we e p esen ed o
Cha GPT-3, on a news i em in he ‘acciden ’ca ego y.
We asked he cha he ollowing: “can you w i e a news s o y
wi h he ollowing keywo ds: man, 40 yea s old [who?], San a
Ma ia da Fei a [whe e?], wo k acciden [why], o kli [how?] and
F iday [when?]. We eplica ed he same p ocedu e o he
emaining news. Cha GPT c ea ed a news s o y abou a 40-yea -
old man who su e ed an acciden a wo k in San a Ma ia
da Fei a, a own in Po ugal, gene a ing he ollowing headline:
“40-yea -old man su e s an acciden a wo k in a co k indus y
company in San a Ma ia da Fei a”. In Table 1we can find he
headlines o he news w i en by jou nalis s and gene a ed by AI
and he espec i e online media.
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2 ARTICLE
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2 3
2) Su ey
Bo h su eys ha e h ee sec ions. Fi s , pa icipan s we e asked
o answe sociodemog aphic ques ions. Secondly, he su ey
p esen ed ques ions o unde s and s uden s’news consump ion
habi s and p ac ices. Thi d, pa icipan s we e exposed o ou
news s o ies c ea ed by jou nalis s and he same ou news s o ies
gene a ed by Cha GPT-3. Each news i em co esponds o a
ca ego y: acciden , c ime, celeb i y and cul u e (su ey can be
consul ed in he Supplemen a y Ma e ials).
S uden s did no ecei e any ale abou he exis ence o news
gene a ed by AI, o ensu e he neu ali y o hei assessmen , as
o he s udies ha e done when measu ing, o example, s uden s’
abili y o dis inguish ake news om eal news (Bap is a e al.
2022). A he s a o he su ey, pa icipan s we e in o med ha
he s udy aimed o assess how audiences e alua e he quali y o
online news. Fu he mo e, bo h news c ea ed by jou nalis s and
AI we e p esen ed in an iden ical empla e, designed exclusi ely
o he esea ch (Fig. 2).
As Fig. 2shows, he same e en was p esen ed in wo di e en
news s o ies, one c ea ed by Cha GPT and he o he c ea ed by
jou nalis s. In he su ey, dis ibu ed in Po ugal, he news
gene a ed by Cha GPT we e p esen ed as ha ing been published
by he fic i ious newspape called “No iciá io”. News w i en by
jou nalis s was a ibu ed o he fic i ious newspape called
“No ícias”. In Spain, Okdia io news was p esen ed by he fic ional
newspape ‘Pe iódico’and Cha GPT news was a ibu ed o he
fic ional newspape ‘In o ma i o’. To ensu e a ex ual e alua ion
o he news and ha no o he ac o s in e e ed wi h he s uden s’
e alua ion, he news was no only p esen ed in he same empla e,
bu no au ho was a ibu ed o he ex , as i is well known ha
he name o he au ho can influence he c edibili y o he news
(Ru o and Seme a o 2022). The same image was also always used
o all news on he same subjec . The image ha illus a es each
news i em, om each ca ego y, was aken om he news
published by he newspape s Co eio da Manhã and Okdia io.
Rega ding he assessmen o he ex ual quali y o news, we
used a quali y indica o , adap ed om A apakis e al. (2016),
composed o h ee dimensions: eadabili y, in o ma i eness and
s yle. As we in ended o s uden s o e alua e he news based
exclusi ely on he a icle i sel , wi hou conside ing o he con ex s
o ac o s, he su ey ques ions we e designed acco ding o hese
c i e ia. A 7-poin Like scale was used (1—S ongly Disag ee;
Fig. 1 Sc eensho : example o p omp s pos ed o Cha GPT-3. Sou ce: Own elabo a ion.
Table 1 News headlines w i en by jou nalis s and news headlines gene a ed by Cha GPT by each newspape .
Newspape News headline Cha GPT headline
Co eio
da Manhã
Joana Dua e became a mo he o he fi s ime. Ac ess Joana Dua e gi es bi h o he fi s ime a Hospi al da Luz
and p ese es he a he ’s anonymi y
Man se iously inju ed in wo k acciden in San a Ma ia da Fei a 40-yea -old man su e s an acciden a wo k in a co k indus y
company in San a Ma ia da Fei a
Man kills woman wi h ea -naked choke in Lisbon Woman is mu de ed in he home by he pa ne wi h a ea -naked
choke in Lisbon
Pain ing “Banana Woman”by Paula Rego sold a auc ion o
266 housand eu os
Pain ing ‘Banana Woman’by Paula Rego sells o 266 housand eu os a
mode n and con empo a y a auc ion
OKdia io Wha is Miley Cy us’new Mulle hai cu like. Miley Cy us causes a s i on social ne wo ks by debu ing a da ing
mulle on Ins ag am
T agic acciden a a pa y in Is ael: he floo o a swimming pool
gi es way and ‘swallows’ wo people
T agic acciden in Is ael du ing summe pa y lea es one man dead and
ano he inju ed in pool collapse
P ison o he de ainee o killing he boy Mohamed in Ceu a o
c imes o mu de and sexual assaul
A man a es ed in Ceu a accused o mu de and sexual assaul agains
a six-yea -old boy
A iana G ande ecei ed his la ge amoun o money o he
‘Coachella’conce
A iana G ande ecei ed $8 million o he spec acula pe o mance a
he Coachella es i al
Sou ce: own elabo a ion.
ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2
4HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2
7—S ongly Ag ee), whe e a highe sco e always means highe
quali y. We conside ed ha he eadabili y o a ex is associa ed,
abo e all, wi h he ease o eading a ex . This may be ela ed o
he leng h, posi ioning and o ganiza ion o sen ences, which
implies he fluency and conciseness o he ex (Zamanian and
Hayda i 2012; A apakis e al. 2016). Conside ing his dimension,
s uden s e alua ed he news acco ding o he ollowing pa a-
me e s: (a) easy o ead and in e p e ; (b) consis en and accu a e
p esen a ion o ac s.
In he in o ma i eness dimension, A apakis e al. (2016)
in eg a es se e al undamen al aspec s, pa icula ly a he le el o
desc ip ion. The deg ee o desc ip ion, o example, o a i le in
line wi h he con en o he news body is a quali y pa ame e . Fo
example, clickbai i les a e no e y desc ip i e and do no
co espond o he expec a ions c ea ed by use s, o e ing li le
in o ma ion (Ga cía O osa e al. 2017). In addi ion o desc ip-
i eness, A apakis e al. (2016) also conside comple eness and
e e encing as pa ame e s o he in o ma i eness dimension. On
he one hand, news a icles mus co e a opic in de ail and wi h
enough in o ma ion o suppo he allega ions. On he o he
hand, he jou nalis ic ex mus con ain sou ces and quo a ions
ha a e easily unde s ood by eade s.
Thus, in e ms o in o ma i eness, uni e si y s uden s classified
he i le o each news i em, namely on how well i eflec s i s main
con en . The way in which a ce ain subjec o opic is epo ed
was also e alua ed. We ques ioned s uden s abou whe he he
news p o ides enough in o ma ion abou he s o y. When
e alua ing in o ma i eness, s uden s also exp essed hei opinion
on he use o news sou ces in an app op ia e and balanced way.
As o he hi d dimension o ex ual quali y, s yle, as A apakis
e al. (2016) we conside aspec s such as he use o punc ua ion,
jou nalis ic w i ing guidelines and he use o g amma ical ules.
Resul s
Ou esul s show ha Ibe ian uni e si y s uden s conside ed, on
a e age, news gene a ed by Cha GPT o be o highe quali y
(M =4.85, SD =0.82) han news w i en by jou nalis s
(M =4.30, SD =0.84). The di e ences be ween he wo g oups
a e s a is ically significan [ (444) =11.919, p< 0.001, d =0.566].
E en di iding he sample by coun y o pa icipan s, we ound
ha bo h g oups gene ally conside ed he news c ea ed by
Cha GPT-3 o be o highe quali y. Howe e , Spanish uni e si y
s uden s ga e a highe a ing o Cha GPT-3 news (M =4.92,
SD =0.80) compa ed o he a e age a ing gi en by Po uguese
s uden s (M =4.59, SD.84). Rega ding he pe cei ed quali y o
news w i en by jou nalis s, s uden s om Spain ga e on a e age
he lowes a ing (M =4.33, SD =0.79). The uni a ia e analysis
o a iance (ANOVA) e ealed s a is ically significan di e ences
only in ela ion o he pe cei ed quali y o Cha GPT news, when
compa ing he wo coun ies (F(1,444) =12.238, p< 0.001).
Ac oss he en i e sample, ou findings indica e ha Cha GPT-
3-gene a ed news ecei ed, on a e age, highe pe cei ed quali y
a ings ac oss all news ca ego ies: celeb i y, c ime, acciden and
cul u e (Table 2).
As shown in Table 2, he news gene a ed by Cha GPT ha
ob ained he highes sco e we e abou cul u e (M =4.92,
SD =1.01). This may be ela ed o he ac ha he piece abou
cul u e w i en by jou nalis s ecei ed one o he lowes a ings
(M =4.21, SD =1.06). Howe e , i was c ime news ha ecei ed
he lowes sco e. In ac , we can see ha Cha GPT news ecei ed
significan ly highe alues in all ca ego ies.
Analyzing he esul s by coun y (Table 3), we was ound ha ,
in Po ugal, he news ha ob ained he highes sco e we e also
c ea ed by Cha GPT and belong o he “acciden ”ca ego y
(M =4.76, SD =1.09). The news w i en by jou nalis s ha
Fig. 2 Example o news empla es used in su eys. Sou ce: own elabo a ion.
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2 ARTICLE
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2 5
ecei ed he bes a ing a e consis en ly below he Cha GPT news
a ings.
In Spain, he a e age Cha GPT news a ings a e highe han
he Cha GPT news a ings in Po ugal. The Cha GPT news s o y
abou c ime sco ed he highes (M =4.95, SD=0.99). In e es -
ingly, c ime news was he one wi h he lowes a ing among news
w i en by jou nalis s (M =4.10, SD =1.18).
Conside ing news quali y indica o s, we ound ha , o e all,
jou nalis s pe o med poo ly in p esen ing sou ces o e e ences
and in hei use o punc ua ion in he news hey w o e (Table 4).
These we e he quali y indica o s ha ob ained he lowes
sco es, namely he use o a co ec and adequa e punc ua ion in
he news (M =4.03, SD =1.11). In his indica o , Cha GPT news
p esen ed a significan ly highe quali y (M =4.73, SD =0.99)
[ (444)=–10.899, p< 0.001, d =–0.517]. News w i en by jou -
nalis s ecei ed i s highes a ing o use o clea and p ecise
language (M =4.64, SD =1.16), bu his was s ill significan ly
lowe han Cha GPT news (M =4.83, SD =1.20), [ (444) =–
3.611, p< 0.001,d =–0.171].
In gene al, Cha GPT news has a highe pe cei ed quali y in all
indica o s analyzed. Cha GPT news ecei ed he highes sco e in
e ms o eadabili y, p esen a ion o ac s, c ea ion o i les, ci a-
ion o sou ces, use o punc ua ion and unde s andable language
(Table 4).
Focusing on each coun y, we no e ha he e a e no significan
di e ences ha s and ou when compa ing he quali y o news
indica o s and Cha GPT news. Howe e , he e a e s a is ically
significan di e ences in ela ion o eadabili y when e alua ing
Cha GPT news [F(1,444) =13.717, p< 0.001], wi h s uden s om
Spain gi ing a highe a ing.
The e a e also s a is ically significan di e ences in ela ion o
he pe cei ed quali y o he Cha GPT news language [F(1,444) =
26.294, p< 0.001], in which s uden s om Spain ecognize highe
quali y. E en so, compa ing he esponses om he wo coun ies,
mos indica o s do no show s a is ically significan di e ences,
which e eals ha he e is no significan influence o he coun y
o esidence and ha s uden s e eal simila abili ies.
Associa ion be ween a iables. Th ough a mul i a ia e analysis o
a iance (MANOVA) we no iced ha a iables such as sex and yea
o schooling do no ha e any significan e ec on he assessmen o
news quali y [Fsex(2,444) =0.743, p=0.476 x Fyea (6.444) =
1.900, p=0.079]andin hee alua iono Cha GPTnews
[Fsex(2,444) =0.813, p=0.444 x Fyea (6,444) =0.743, p=0.476].
To measu e he e ec o (1) us in news, (2) in e es in news
and (3) equency o access o news, we fi s ecoded he a iables
in o ca ego ies (1=li le, 2=somewha , 3=a lo ). Se e al MAN-
OVAS did no show s a is ically significan di e ences in ela ion
o he e ec o a iables on quali y assessmen .
Discussion and conclusions
Ou s udy in es iga ed he pe cei ed quali y o news gene a ed by
Cha GPT-3 compa ed o news w i en by human jou nalis s
among uni e si y s uden s in he Ibe ian Peninsula. The esul s
e ealed a significan p e e ence o Cha GPT-gene a ed news,
wi h highe a ings o eadabili y, in o ma i eness, and ex ual
s uc u e ac oss all news ca ego ies. No ably, Spanish s uden s
ended o a e Cha GPT news e en mo e a o ably han hei
Po uguese coun e pa s, which may e eal ha his ool ollows a
w i ing s uc u e o news ha is mo e like he pa e n ha is
augh in Spanish acul ies.
The findings also indica ed ha he quali y indica o s assessed,
such as p esen a ion o sou ces, use o punc ua ion, and o e all
language quali y, consis en ly a o ed Cha GPT-gene a ed news
o e human-w i en news. In e es ingly, he pe cei ed quali y o
news c ea ed by jou nalis s was lowe , pa icula ly in e ms o
p esen ing sou ces and e e ences, indica ing po en ial conce ns
abou jou nalis ic p ac ices. Thus, jou nalism s uden s s uggle o
disce n newsw i ing gene a ed by AI om newsw i ing c ea ed by
Table 2 Pe cei ed quali y o news by ca ego y.
Celeb i y (N=444) Acciden (N=444) C ime (N=444) Cul u e (N=444)
News. GPT News. GPT News. GPT News. GPT
4.39 ± 1.14 4.77 ± 1.00 4.47 ± 1.09 4.83 ± 1.07 4.10 ± 1.15 4.86 ± 1.01 4.21 ± 1.06 4.92 ± 1.01
p<.001 p <.001 p <.001 p <.001
=-5.303 =-5.189 =-10.266 =-10.986
Sou ce: own elabo a ion.
Table 3 Analyzing by coun y: pe cei ed quali y o news by ca ego y.
Coun y Celeb i y Acciden C ime Cul u e
News. GPT News. GPT News. GPT News. GPT
Spain (N=345) 4.36 ± 1.18 4.85 ± 1.00 4.53 ± 1.10 4.85 ± 1.06 4.10 ± 1.18 4.95 ± .99 4.14 ± 1.11 5.02 ± 1.01
Po ugal (N=99) 4.51 ± 1.02 4.50 ± .092 4.25 ± 1.02 4.76 ± 1.07 4.11 ± 1.05 4.55 ± 1.02 4.44 ± 0.85 4.57 ± 0.95
p=0.262 p =0.002 p =0.021 p =0.434 p =0.940 p <0.001 p =0.015 p < 0.001
F=1.252 F =9.749 F =5.360 F=0.613 F=0.006 F=12.068 F =5.984 F =15.638
Sou ce: own elabo a ion.
Table 4 Sco e assigned by quali y indica o s.
Indica o s News (N=444)
News. GPT p
Readabili y 4.77 ± 1.02 5.13 ± 1.04 <0.001 –9.046
Fac ual 4.16 ± 1.09 4.87 ± .98 <0.001 –11.600
Headline 4.30 ± 1.04 4.95 ± 1.01 <0.001 –11.553
Co e age 4.27 ± 1.07 4.83 ± .97 <0.001 –9.848
Sou ces 4.08 ± 1.04 4.59 ± .95 <0.001 –8.780
Punc ua ion 4.03 ± 1.11 4.73 ± .99 <0.001 –10.899
Language 4.64 ± 1.16 4.83 ± 1.20 <0.001 –3.611
Sou ce: own elabo a ion.
ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2
6HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2
human jou nalis s (RQ1), challenging he con en ional no ion o
human jou nalis s as he gold s anda d in news c ea ion.
Ou esea ch is a fi s app oach o he pe cei ed quali y o
news gene a ed by Cha GPT-3, a gene a i e a ificial in elligence
language model, in he con ex o he Ibe ian Peninsula. Ou
findings showed, om a gene al pe spec i e, ha news c ea ed by
Cha GPT-3 we e a ed as ha ing be e quali y han news w i en
by human jou nalis s. In none o he quali y dimensions assessed
did uni e si y s uden s a e jou nalis -c ea ed news as supe io ,
highligh ing implica ions o he ole o humans in he con ex o
AI use. This end was consis en ly obse ed among s uden s
om bo h Spain and Po ugal.
These esul s seem o be in line wi h o he s udies ha analyzed
he capaci y and quali y o au oma ed jou nalism and ound ha
eade s end no o be able o dis inguish be ween ex s gene a ed
by AI o humans (Jia 2020; Cle wall 2014; Wölke and Powell
2021; Van de Kaa and K hame 2014). S ill, ou s udy clea ly
shows ha s uden s conside ed he a icles c ea ed by Cha GPT
o be be e w i en, e en hough hey we e p esen ed in di e en
languages. Simila o ou s udy, Cle wall (2014) also examined he
e alua ion o uni e si y media and communica ion s uden s, who
a ibu ed g ea e c edibili y o he au oma ed e sion in ela ion
o he e sion w i en by jou nalis s. Howe e , unlike ou s udy,
in Cle wall (2014) s uden s a ibu ed g ea e eadabili y o a i-
cles w i en by humans.
Likewise, G ae e e al. (2018) and Haim and G ae e (2017)
ound ha a icles w i en by humans ecei ed a highe ead-
abili y a ing. Gi en hese da a, we can discuss wo hypo heses.
On he one hand, he ex ual quali y o w i en news may ha e
los quali y o e he yea s. The p essu e o publish, he accele -
a ed digi aliza ion o jou nalism, he a en ion economy, social
ne wo ks and g owing sensa ionalism may ha e had a nega i e
e ec on he ex ual quali y o news. On he o he hand, he
compu e ’s abili y o w i e news has imp o ed, pa icula ly a he
le el o na u al human language (Caswell and Do 2018), which
jus ifies he significan ly highe sco e in e ms o eadabili y o
Cha GPT-3 in ela ion o news w i en by humans.
In ac , he abili y and language model o Cha GPT-3 has been
ecognized o he quali y o he ex s i gene a es (Lei e e al.
2024), al hough he same does no apply o he w i ing o ex s o
a scien ific na u e (Gao e al. 2022). In some s udies, au oma ed
jou nalism appea s mo e c edible o specific ca ego ies, such as
Spo s (Wölke and Powell 2021). In ou s udy, news c ea ed by
Cha GPT had highe quali y sco es ac oss all opics: acciden ,
celeb i y, cul u e and c ime. This means ha he subjec o he
news influence he pe cei ed quali y (RQ2). In e es ingly, he
quali y indica o o news c ea ed by humans ha ob ained he
lowes a e age sco e was he indica o ela ed o he p esen a ion
o sou ces and e e ences in he ex , and he use o co ec and
app op ia e punc ua ion.
This s udy highligh s a new issue in he ongoing deba e abou
AI’s u u e and i s g owing impac on jou nalism—specifically,
he quali y o human-w i en jou nalism a he le el o key c i-
e ia. S uden s assigned lowe sco es o he p esen a ion o
sou ces and e e ences, indica o s ha ex end beyond he ex ual
quali y o he news.
In addi ion, in ou discussion we also conside he cha ac e -
is ics o he pa icipan s who e alua ed he news. I is well known
ha young people oday ead li le news and s uggle wi h
a en ion defici s. Mo eo e , hey gene ally show low in e es in
news, e en among hose s udying jou nalism. Resea ch also
indica es ha young people o en ha e di ficul y dis inguishing
eal news om ake news o om ad e isemen s.
Ou s udy p omp s conside a ions ega ding he po en ial
impac o AI on jou nalism, aising ques ions abou he cu en
s a e o human-w i en news quali y. The highe a ings o
Cha GPT-3, e en ac oss di e en languages, sugges a possible
decline in he ex ual quali y o human-w i en news o e he
yea s. Howe e , i is impo an o no e ha he s udy ocused on
uni e si y s uden s, who may ha e specific eading habi s and
p e e ences, including a po en ial human bias owa ds AI-
gene a ed con en . Ano he limi a ion is ha headlines o cu -
en jou nalism could no include all he basic ques ions due o
edi o ial o economic cons ain s, bu we app aise ha he six
classic jou nalism ques ions u he ou unde s anding on a good
jou nalis ic p ac ice.
Also, as men ioned, Co eio da Manhã in Po ugal and
OKdia io in Spain a e bo h sensa ionalis abloid news media ha
ope a e wi h p o essional ques ionable s anda ds. Ou decision o
e alua e he pe cei ed quali y le el h ough hese newspape s
es ed on he assump ion ha low-quali y jou nalism may be
easily eplaced by AI-gene a ed news. This poin s o he possi-
bili y o using AI o non-se ious opics, bu he selec ion o he
sample is also a limi a ion as i migh ha e shaped pe cep ions in
a nega i e way.
Since hese cons ain s o edi o ial guidelines may esul in he
s o ies being less clea o each a la ge audience, a enue o
u u e esea ch is o app oxima e hese guidelines and equi e
Cha GPT o ollow hem o see i he di e ences emain. In he
con ex o he ongoing deba e abou he u u e o AI in jou -
nalism, ou findings highligh no only he capabili ies o
Cha GPT-3 bu also po en ial conce ns abou he quali y o
human-w i en news ha can be conside ed, especially in aspec s
beyond pu e ex quali y. As he field o AI con inues o ad ance,
u he esea ch and discussions should explo e he implica ions
o jou nalism and how i may shape he e ol ing landscape o
news consump ion.
Da a a ailabili y
Da a om he su eys used o ca y ou his esea ch a e a ailable
in he Supplemen a y Da a.
Recei ed: 12 Sep embe 2024; Accep ed: 3 Ap il 2025;
Re e ences
Ande son CW (2013) Towa ds a sociology o compu a ional and algo i hmic
jou nalism. N. Media Soc 15(7):1005–1021. h ps://doi.o g/10.1177/
1461444812465137
Ande son CW (2018) Apos les o Ce ain y: Da a Jou nalism and he Poli ics o
Doub . New Yo k: Ox o d Uni e si y P ess
A ambu ú Moncada LG, López Redondo I, López Hidalgo A (2023) In eligencia
a ificial en RTVE al se icio de la España acía. P oyec o de cobe u a
in o ma i a con edacción au oma izada pa a las elecciones municipales de
2023. Re La de Comunicación Soc 81:1–16. h ps://doi.o g/10.4185/RLCS-
2023-1550
A apakis I, Peleja F, Be kan B, Magalhaes J (2016) Linguis ic benchma ks o online
news a icle quali y. In P oceedings o he 54 h Annual Mee ing o he
Associa ion o Compu a ional Linguis ics (Long Pape s) (pp. 1893–1902)
Bap is a JP, G adim A (2024) A e jou nalis s no longe needed? Compa a i e
analysis o he pe cei ed quali y o eal news and Cha GPT news. In:
P oceedings TEEM 2023: Technological Ecosys ems o Enhancing
Mul icul u ali y
Bap is a JP, G adim A, Co eia E (2022) A elação en e a c ença nas ake news e as
es a égias pa a p ocu a in o mação dos jo ens po ugueses. Obse a ó io
(OBS*), 16(3). h ps://doi.o g/10.15847/obsOBS16320222082
Bap is a JP, Ri as-de-Roca R, G adim A, Lou ei o M (2023) The Disin o ma ion
Reac ion o he Russia–Uk aine Wa : An Analysis h ough he Lens o
Ibe ian Fac - Checking. KOME 11(2):27–48. h ps://doi.o g/10.17646/
KOME.o .2
Cal o-Rubio L, U a e-Ruiz M (2020) Pe cepción de docen es uni e si a ios,
es udian es, esponsables de inno ación y pe iodis as sob e el uso de in eli-
gencia a ificial en pe iodismo. P o esional de la In ón 29(1):e290109. h ps://
doi.o g/10.3145/epi.2020.ene.09
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2 ARTICLE
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2 7
Cana ilhas J (2022) A ificial In elligence and Jou nalism: Cu en Si ua ion and
Expec a ions in he Po uguese Spo s Media. Jou nalism Media
3(3):510–520. h ps://doi.o g/10.3390/jou nalmedia3030035
Cana ilhas J, Di Fá ima B (2025) Repo e os del u u o: consumo de no icias en e
es udian es de pe iodismo en Po ugal. Doxa Comunicación Re In e -
disciplina De Es udios De Comunicación Y Cienc Soc 40:305–319. h ps://
doi.o g/10.31921/doxacom.n40a2392
Cana ilhas J, Giacomelli F (2023) In eligencia a ificial en el pe iodismo depo i o:
es udio en B asil y Po ugal. Re de Comunicación 22(1):53–69. h ps://doi.
o g/10.26441/RC22.1-2023-3005
Caswell D, Dö K (2018) Au oma ed Jou nalism 2.0: E en -d i en na a i es:
F om simple desc ip ions o eal s o ies. Jou nalism P 12(4):477–496.
h ps://doi.o g/10.1080/17512786.2017.1320773
Cle wall C (2014) En e he obo jou nalis . Jou nalism P 8(5):519531. h ps://doi.
o g/10.1080/17512786.2014.883116
Codding on M (2015) Cla i ying jou nalism’s quan i a i e u n: A ypology o
e alua ing da a jou nalism, compu a ional jou nalism, and compu e -assis ed
epo ing. Digi al Jou nalism 3(3):331–348. h ps://doi.o g/10.1080/
21670811.2014.976400
Cools H, Van Go p B, Opgenha en M (2024) Whe e exac ly be ween u opia and
dys opia? A aming analysis o AI and au oma ion in US newspape s.
Jou nalism 25(1):3–21. h ps://doi.o g/10.1177/146488492211226
de-Lima-San os MF, Ce on W (2021) A ificial In elligence in News Media: Cu -
en Pe cep ions and Fu u e Ou look. Jou nalism Media 3(1):13–26. h ps://
doi.o g/10.3390/jou nalmedia3010002
Deuze M, Becke C (2022) Imagina ion, algo i hms and news: de eloping AI
li e acy o jou nalism. Digi al Jou nalism 10(10):1913–1918. h ps://doi.o g/
10.1080/21670811.2022.2119152
Diakopoulos N (2020) Compu a ional news disco e y: owa ds design conside a-
ions o edi o ial o ien a ion algo i hms in jou nalism. Digi al Jou nalism
8(7):945–967. h ps://doi.o g/10.1080/21670811.2020.1736946
Foà C, Tomé V, Ma ga o D, Paisana M, C espo M, Ca doso G (2023) Roles o
jou nalis s in media li e acy ini ia i es: ainees and aine s. Con inui y,
collabo a ion, and sus ainabili y o media li e acy ainings o mi iga e dis-
in o ma ion in Po ugal. P o esional de la In ón 32(6):e320621. h ps://doi.
o g/10.3145/epi.2023.no .21
Gao CA, Howa d FM, Ma ko NS, Dye EC, Ramesh S, Luo Y, Pea son AT (2022)
Compa ing scien ific abs ac s gene a ed by Cha GPT o o iginal abs ac s
using an a ificial in elligence ou pu de ec o , plagia ism de ec o , and
blinded human e iewe s. BioRxi . h ps://doi.o g/10.1101/2022.12.23.
521610
Ga cía O osa B, Gallu San o um S, López Ga cía X (2017) El uso del clickbai en
cibe medios de los 28 países de la Unión Eu opea. Re La de Comunicación
Soc 72:1261–1277. h ps://doi.o g/10.4185/RLCS-2017-1218
Gómez-Diago G (2022) Pe spec i as pa a abo da la in eligencia a ificial en la
enseñanza de Pe iodismo. Una e isión de expe iencias in es igado as y
docen es. Re La de Comunicación Soc 80:29–46. h ps://doi.o g/10.4185/
RLCS-2022-1542
Gonçal es A, Melo PV (2022) In eligencia a ificial y pe iodismo: una ap oxima-
ción al con ex o po ugués. Fonseca-J Commun 25:23–34. h ps://doi.o g/10.
14201/ jc.29682
González-A ias C, López-Ga cía X (2023) Cha GPT: S eam o opinion in fi e
newspape s in he fi s 100 days since i s launch. P o esional de la In ón
32(5):e320524. h ps://doi.o g/10.3145/epi.2023.sep.24
G ae e A, Haim M, Haa mann B, B osius HB (2018) Reade s’pe cep ion o
compu e -gene a ed news: C edibili y, expe ise, and eadabili y. Jou nalism
19(5):595–610. h ps://doi.o g/10.1177/1464884916641269
Gu ié ez López M, Po lezza C, Coope G, Mak i S, MacFa lane A, Missaoui S
(2023) A Ques ion o Design: S a egies o Embedding AI-D i en Tools in o
Jou nalis ic Wo k Rou ines. Digi al Jou nalism 11(3):484–503. h ps://doi.
o g/10.1080/21670811.2022.2043759
Gu ié ez-Caneda B, Vázquez-He e o J, López-Ga cía X (2023) AI applica ion
in jou nalism: Cha GPT and he uses and isks o an eme gen echnol-
ogy. P o esional de la In ón 32(5):e320514. h ps://doi.o g/10.3145/epi.
2023.sep.14
Haim M, G ae e A (2017) Au oma ed news: Be e han expec ed? Digi al Jou -
nalism 5(8):1044–1059. h ps://doi.o g/10.1080/21670811.2017.1345643
Hallin DC, Mancini P (2004) Compa ing Media Sys ems: Th ee Models o Media
and Poli ics. New Yo k: Camb idge Uni e si y P ess
Jia C (2020) Chinese au oma ed jou nalism: A compa ison be ween expec a ions
and pe cei ed quali y. In J Commun 14:22
an de Kaa HA, K ahme EJ (2014) Jou nalis e sus news consume : The pe -
cei ed c edibili y o machine w i en news. In Compu a ion+Jou nalism
Symposium 2014
Ka inshak E, Jin Y (2023) AI-d i en disin o ma ion: a amewo k o o ganiza-
ional p epa a ion and esponse. J Commun Manag 27(4):539–562. h ps://
doi.o g/10.1108/JCOM-09-2022-0113
Lei e C, Zhang R, Chen Y, Belouadi J, La iono D, F esen V, Ege S (2024)
Cha gp : A me a-analysis a e 2.5 mon hs. Mach Lea n Appl 16:100541.
h ps://doi.o g/10.1016/j.mlwa.2024.100541
Lemelsh ich La a N (2018) Robo jou nalism: Can human jou nalism su i e?
New Yo k: Wo ld Scien ific Publishing
Lopezosa C, Codina L, Pon -So ibes C, Vállez M (2023) Use o gene a i e
a ificial in elligence in he aining o jou nalis s: challenges, uses and
aining p oposal. P o esional de la In ón 32:e320408. h ps://doi.o g/10.
3145/epi.2023.jul.08
Minis e io de Ciencia, Inno ación y Uni e sidades (2024) Da os y ci as del Sis ema
Uni e si a io Español, publicación 2023-2024. Mad id: Adminis ación Gen-
e al del Es ado. h ps://www.uni e sidades.gob.es/wp-con en /uploads/2024/
05/SIIU_Da osCi as2024.pd
Mo an RE, Shaikh SJ (2022) Robo s in he News and News ooms: Unpacking
Me a-Jou nalis ic Discou se on he Use o A ificial In elligence in Jou nal-
ism. Digi al Jou nalism 10(10):1756–1774. h ps://doi.o g/10.1080/21670811.
2022.2085129
Muno iya wa A, Chiumbu S, Mo saa hebe G (2023) A ificial In elligence P ac ices
in E e yday News P oduc ion: The Case o Sou h A ica’s Mains eam
News ooms. Jou nalism P 17(7):1374–1392. h ps://doi.o g/10.1080/
17512786.2021.1984976
Noain-Sánchez A (2022) Add essing he Impac o A ificial In elligence on
Jou nalism: he pe cep ion o expe s, jou nalis s and academics. Commun
Soc 35(3):105–121. h ps://doi.o g/10.15581/003.35.3.105-121
Pa lik JV (2023) Collabo a ing Wi h Cha GPT: Conside ing he Implica ions o
Gene a i e A ificial In elligence o Jou nalism and Media Educa ion.
Jou nalism Mass Commun Educa o 78(1):84–93. h ps://doi.o g/10.1177/
10776958221149577
Pé ez-Seijo S, Ba bosa S, Vicen e PN (2023) A ificial In elligence in Jou -
nalism: Case S udy o he Spanish, Po uguese and B azilian News Media
Sys ems. In: Neg ei a-Rey MC, Vázquez-He e o J, Six o-Ga cía J, López-
Ga cía X (eds). Blu ing Bounda ies o Jou nalism in Digi al Media.
S udies in Big Da a,140.Cham:Sp inge .h ps://doi.o g/10.1007/978-3-
031-43926-1_18
Quian A, Six o-Ga cía J (2024) In eligencia a ificial en la p ensa: es udio com-
pa a i o y explo ación de no icias con Cha GPT en un medio adicional y
o o na i o digi al. Re de Comunicación 23(1):457–483. h ps://doi.o g/10.
26441/RC23.1-2024-3374
Ri as-de-Roca R (2022) Calidad y modelos de negocio en los medios de p ox-
imidad: Es udio de casos en Alemania, España y Po ugal. Es udos em
Comun ção 34:8196
Ru o G, Seme a o A (2022) FakeNewsLab: Expe imen al s udy on biases and
pi alls p e en ing us om dis inguishing ue om alse news. Fu u e
in e ne 14(10):283
Six o-Ga cía J, Quian A, Rod íguez-Vázquez A-I, Sil a-Rod íguez A, Soengas-Pé ez
X (2024) Jou nalism, Digi al Media and he Fou h Indus ial Re olu ion.
Cham: Palg a e Macmillan
Tomé V, C espo M, B anco S, Ne y I (2023) Po uguese jou nalis s aining ea-
che s in he new media educa ion landscape du ing Co id-19. In: F iesem Y,
Raman U, Kanizaj I, Choi GY (eds.). The Rou ledge handbook o media
educa ion u u es pos -pandemic. London: Rou ledge
Túñez-López J-M, Tou al-B an C, Cachei o-Requeijo S (2018) Uso de bo s y
algo i mos pa a au oma iza la edacción de no icias: pe cepción y ac i udes
de los pe iodis as en España. P o esional de la In ón 27(4):750–758. h ps://
doi.o g/10.3145/epi.2018.jul.04
U a e MJ, Cal o LM, Mu cia FJ (2020a) Las ecnologías hi- ech en los g ados en
Pe iodismo. Planes de es udios, o mación de los pe iodis as y p opues as de
inse ción cu icula . adComunica 20:43–66. h ps://doi.o g/10.6035/2174-
0992.2020.20.3
U a e MJ, Fiei as C, Túñez M (2020b) La enseñanza-ap endizaje del pe iodismo
au oma izado en ins i uciones públicas: es udios, p opues as de iabilidad y
pe spec i as de impac o de la IA. àlisi: Quad de Comunicaciói Cul ,
62:131–146. h ps://doi.o g/10.5565/ e /analisi.3289
Wes lund O, Lewis SL (2017) Reconside ing News P oduc ion: How Unde -
s anding he In e play o Ac o s, Ac an s, and Audiences Can Imp o e
Jou nalism Educa ion (409-428). In: Goodman, R. S., & S eyn, E. (eds.).
Global Jou nalism Educa ion in he 21s Cen u y: Challenges and Inno a ions
Wo ld Jou nalism Educa ion Council, pp. 409-428. Aus in: Knigh Cen e o
Jou nalism in he Ame icas, Uni e si y o Texas a Aus in
Why e C (2020) Deep ake news: AI-enabled disin o ma ion as a mul i-le el public
policy challenge. J Cybe Policy 5(2):199–217. h ps://doi.o g/10.1080/
23738871.2020.1797135
Wölke A, Powell TE (2021) Algo i hms in he news oom? News eade s’pe cei ed
c edibili y and selec ion o au oma ed jou nalism. Jou nalism 22(1):86–103.
h ps://doi.o g/10.1177/1464884918757072
Zamanian M, Heyda i P (2012) Readabili y o Tex s: S a e o he A . Theo y &
P ac ice in Language S udies, 2(1)
ARTICLE HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2
8HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2
Au ho con ibu ions
All au ho s we e equally in ol ed in he applica ion o he esea ch me hodology. The co -
esponding au ho , João Ped o Bap is a, and he au ho Rúben Ri as-de-Roca ca e ully w o e
he a icle, unde he supe ision o he au ho s Anabela G adim and Concha Pé ez-Cu iel.
Compe ing in e es s
The au ho s decla e no compe ing in e es s.
E hical s a emen s
This s udy adhe ed o he e hical guidelines and esea ch policies o he Uni e si y o
Bei a In e io (UBI), which align wi h he highes in e na ional s anda ds o academic
in eg i y and esponsible esea ch. A UBI, hese e hical p inciples a e no me ely
ecommenda ions bu manda o y equi emen s o all esea ch in ol ing human pa i-
cipan s. The s udy was conduc ed using he p inciples o he Decla a ion o Helsinki,
which ensu es he p o ec ion o pa icipan s’ igh s, digni y, and confiden iali y. Among
he egula ions ollowed, we s ic ly adhe ed o he Academic In eg i y Code o UBI,
issued in 2018 (h ps://www.ubi.p /Fichei os/En idades/91363/codigo_in eg idade.pd ),
and he Regula ion o Pe o ming Resea ch Su eys, issued in 2022 (h ps://www.ubi.p /
Fichei os/En idades/91051/Despacho%202022-RT-18.pd ). Ou esea ch was ca ied ou
in a class oom, whe e pa icipa ion is olun a y, non-in asi e and does no ep esen a
isk o pa icipan s. The e o e, ou s udy did no equi e o mal e hical app o al
acco ding o hese ins i u ional guidelines.
In o med consen
All pa icipan s we e ully in o med abou he objec i es, me hodology and con-
fiden iali y measu es o he s udy be o e hei olun a y pa icipa ion. The in o med
consen p ocess ook place on 08-05-2023, be o e da a collec ion, and was documen ed in
he su ey i sel , which was sen as suppo ing ma e ial.
Addi ional in o ma ion
Supplemen a y in o ma ion The online e sion con ains supplemen a y ma e ial
a ailable a h ps://doi.o g/10.1057/s41599-025-04872-2.
Co espondence and eques s o ma e ials should be add essed o João Ped o Bap is a.
Rep in s and pe mission in o ma ion is a ailable a h p://www.na u e.com/ ep in s
Publishe ’s no e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in
published maps and ins i u ional a filia ions.
Open Access This a icle is licensed unde a C ea i e Commons
A ibu ion-NonComme cial-NoDe i a i es 4.0 In e na ional License,
which pe mi s any non-comme cial use, sha ing, dis ibu ion and ep oduc ion in any
medium o o ma , as long as you gi e app op ia e c edi o he o iginal au ho (s) and
he sou ce, p o ide a link o he C ea i e Commons licence, and indica e i you modified
he licensed ma e ial. You do no ha e pe mission unde his licence o sha e adap ed
ma e ial de i ed om his a icle o pa s o i . The images o o he hi d pa y ma e ial
in his a icle a e included in he a icle’s C ea i e Commons licence, unless indica ed
o he wise in a c edi line o he ma e ial. I ma e ial is no included in he a icle’s
C ea i e Commons licence and you in ended use is no pe mi ed by s a u o y
egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om
he copy igh holde . To iew a copy o his licence, isi h p://c ea i ecommons.o g/
licenses/by-nc-nd/4.0/.
© The Au ho (s) 2025
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | h ps://doi.o g/10.1057/s41599-025-04872-2 ARTICLE
HUMANITIES AND SOCIAL SCIENCES COMMUNICATIONS | (2025) 12:567 | h ps://doi.o g/10.1057/s41599-025-04872-2 9
