This is an open access
a icle dis ibu ed unde
he e ms o he C ea i e
Commons A ibu ion
License (CC BY 4.0).
Eu opean Science Edi ing
/
ese
Ci a ion
Ng JY, Bha sa D, Dhan an h y N, e al. A i udes and pe cep ions owa ds he
use o a i icial in elligence cha bo s in medical jou nal pee e iew: A p o ocol
o a la ge-scale, in e na ional c oss-sec ional su ey. Eu Sci Ed. 2025;51:e159921.
h ps://doi.o g/10.3897/ese.2025.e159921
Recei ed 26 May 2025
Accep ed: 8 Aug 2025
Published: 10 Oc 2025
O iginal A icle
A i udes and pe cep ions owa ds A i udes and pe cep ions owa ds
he use o a i icial in elligence he use o a i icial in elligence
cha bo s in medical jou nal pee cha bo s in medical jou nal pee
e iew: A p o ocol o a la ge-scale, e iew: A p o ocol o a la ge-scale,
in e na ional c oss-sec ional su eyin e na ional c oss-sec ional su ey
Je emyY. Ng1,2,3,4, Dai a Bha sa 1,2, Neha Dhan an h y1,2, Lex Bou e 5,6,
Te esa Chan7, Anne e Flanagin8, Al onso Io io3,9, Cyn hia Lokke 3,
He é Maisonneu e10,11, Ana Ma ušić12, Da id Mohe 13,14,15, Holge C ame 1,2
1Ins i u e o Gene al P ac ice and In e p o essional Ca e, Uni e si y Hospi al
Tübingen, Tübingen, Ge many
ngjy2@mcmas e .cao je emy[email p o ec ed]
o cid.o g/0000-0003-0031-5873
o cid.o g/0000-0003-3682-918X
o cid.o g/0009-0002-9002-151X
o cid.o g/0000-0002-3640-8046
2Robe Bosch Cen e o In eg a i e Medicine and Heal h, Bosch Heal h Campus,
S u ga , Ge many
3Depa men o Heal h Resea ch Me hods, E idence, and Impac , Facul y o Heal h
Sciences, McMas e Uni e si y, Hamil on, Canada
o cid.o g/0000-0002-3331-8766
o cid.o g/0000-0003-2436-4290
4School o Public Heal h, Facul y o Heal h, Uni e si y o Technology Sydney,
Sydney, Aus alia
5Depa men o Epidemiology and Da a Science, Ams e dam Uni e si ies Medical
Cen e , Ams e dam, The Ne he lands
o cid.o g/0000-0002-2659-5482
Ng e al.
A i udes and pe cep ions owa ds he use o a i icial in elligence
X
XX
doi.o g/10.3897/ese.2025.e159921
Da a A ailabili y S a emen
The da a ha suppo he
indings o his s udy a e
a ailable on eques om he
co esponding au ho .
Au ho con ibu ions
Concep – J.Y.N.; Design – J.Y.N.,
L.B., T.C., A.F., A.I., C.L., H.M.,
A.M., D.M., H.C.; Supe ision–
J.Y.N., H.C.; Resou ce H.C.;
Ma e ials – H.C.; Li e a u e
Sea ch – J.Y.N., D.B., N.D.;
W i ing – J.Y.N., D.B., N.D.;
C i ical Re iews – J.Y.N., D.B.,
N.D., L.B., T.C., A.F., A.I., C.L.,
H.M., A.M., D.M., H.C.
Decla a ion o In e es s
He é Maisonneu e, Ana
Ma ušić, and Je emy Y. Ng a e
membe s o he in e na ional
ad iso y boa d o
Eu opean
Science Edi ing
bu had no ole
in he edi o ial decision-making
p ocess o his manusc ip . All
o he au ho s decla e ha hey
ha e no compe ing in e es s.
Funding
The au ho s decla ed ha his
s udy ecei ed no inancial
suppo .
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 2 / 12
Eu opean Science Edi ing
/
ese
6Depa men o Philosophy, V ije Uni e si ei Ams e dam, Ams e dam,
The Ne he lands
7School o Medicine, To on o Me opoli an Uni e si y, To on o, On a io, Canada
o cid.o g/0000-0001-6104-462X
8JAMA and he JAMA Ne wo k, Chicago, Illinois, USA
o cid.o g/0000-0001-9114-6221
9Depa men o Medicine, McMas e Uni e si y, Hamil on, On a io, Canada
10Consul an , Lyon, F ance
o cid.o g/0000-0001-8365-7558
11Scien i ic Commi ee, Ins i u e o Resea ch and Ac ion on F aud and Plagia ism
in Academia, Gene a, Swi ze land
12Depa men o Resea ch in Biomedicine and Heal h and Cen e o
E idence-based Medicine, Uni e si y o Spli School o Medicine, Spli , C oa ia
o cid.o g/0000-0001-6272-0917
13Cen e o Jou nalology, Me hodological and Implemen a ion Resea ch P og am,
O awa Hospi al Resea ch Ins i u e. O awa, Canada
o cid.o g/0000-0003-2434-4206
14School o Epidemiology and Public Heal h, Facul y o Medicine, Uni e si y o
O awa, O awa, Canada
15Ins i u e o Heal h Policy, Managemen & E alua ion, Dalla Lana School o Public
Heal h, Uni e si y o To on o
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 3 / 12
Eu opean Science Edi ing
/
ese
Keywo ds: a i icial in elligence, a i udes, cha bo s, gene a i e a i icial
in elligence, pee e iew, e iewe s, su ey
Abs ac
Backg ound: A i icial in elligence (AI) cha bo s a e ad anced con e sa ional p o-
g ammes capable o pe o ming asks such as iden i ying me hodological laws, e i-
ying e e ences, and imp o ing language cla i y in manusc ip s. Thei use in pee
e iew has he po en ial o enhance e iciency, educe e iewe wo kload, and add ess
inconsis encies in e iew quali y. Howe e , conce ns emain ega ding hei eliabil-
i y, e hical implica ions, and anspa ency in decision-making, and li le is known
abou how pee e iewe s pe cei e hese ools.
Objec i es: To assess pee e iewe s’ a i udes and pe cep ions owa ds he use o AI
cha bo s in he pee e iew p ocess, including hei amilia i y wi h AI, pe cei ed
bene i s and challenges, e hical conside a ions, and expec a ions o u u e oles.
Me hods: An in e na ional c oss-sec ional su ey will be conduc ed among academic
pee e iewe s. The su ey will collec da a on pa icipan s’ p io expe ience wi h AI,
pe cep ions o he u ili y o cha bo s in suppo ing pee e iew, conce ns ela ed o
e hics and anspa ency, and an icipa ed u u e applica ions.
Resul s: This s udy will epo desc ip i e and compa a i e analyses o e iewe s’
esponses, highligh ing pa e ns in a i udes and pe cep ions by demog aphic and
p o essional cha ac e is ics.
Conclusions: The indings may o e e idence o in o m he de elopmen o u u e
policies and bes p ac ices o he e hical and e ec i e in eg a ion o AI cha bo s in
pee e iew, wi h he goal o imp o ing e iew quali y while add essing po en ial
isks.
A i udes and pe cep ions owa ds he use o a i icial in elligence
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 4 / 12
Backg ound
A i icial in elligence (AI) b oadly e e s o he
capabili y o compu e sys ems o compu e -
con olled obo s o pe o m asks ypically
associa ed wi h human in elligence, such as
easoning, p oblem-sol ing, gene alizing,
and lea ning om expe ience.1,2 Al hough
cu en AI p og ammes lack he e sa ili y o
human in elligence, specialized applica ions
ha e pe mea ed nume ous ields, including
sel -d i ing ca s, speech ansc ip ion, heal h-
ca e, and educa ion.2 In many domains, AI
has demons a ed bene i s such as inc eased
p oduc i i y, ewe e o s, and cos sa ings.
Fo example, in medicine, AI sys ems can
imp o e diagnos ic accu acy, op imize ea -
men plans, and educe heal hca e cos s o
many heal hca e sys ems.3,4 These hypo he i-
cal bene i s ha e spa ked in e es in applying
AI o schola ly publishing, including he pee
e iew p ocess, whe e e iciency and quali y
imp o emen s a e e y much needed.5
AI cha bo s, a subse o AI p og ammes, a e
gene a i e AI ools designed o simula e
human con e sa ion h ough ex o speech.6
AI cha bo s ha e demons a ed e sa ili y in
applica ions anging om cus ome se ice
o educa ion.6,7 Wi hin schola ly publishing,
AI cha bo s hold p omise o asks such as
imp o ing language cla i y, iden i ying me h-
odological laws, e i ying e e ences, and
s anda dising e iew quali y.8 The in eg a ion
o cha bo s in o pee e iew wo k lows could
lowe e iewe wo kloads, s eamline p o-
cesses, and add ess long-s anding issues such
as bias in and inconsis ency be ween e iew
epo s.8 Mo eo e , AI cha bo s may add ess
e iewe a igue and help pee e iewe s cope
wi h he inc easing olume o eques s, ensu -
ing mo e imely and consis en e alua ions.9,10
Despi e hese po en ial ad an ages, he use o
AI cha bo s in pee e iew aces challenges.
Majo conce ns include he eliabili y o AI
cha bo s in e alua ing complex scien i ic con-
en , he isk o o e looking c i ical nuances,
he po en ial o manipula ion o he pee
e iew p ocess (such as by injec ing p omp s),
and e hical conside a ions ela ed o he use
o AI.11 Fo example, well-documen ed limi a-
ions o Cha GPT include gene a ing plausible
bu inco ec o misleading con en , inaccu-
a e ci a ions, and e e ences o non-exis en
sou ces.11-13 Addi ionally, as AI cha bo s ely
on aining da a se s ha may lack cu ency
o inclusi i y, he e is a isk o pe pe ua ing
ou da ed in o ma ion and biases.14
E hical conce ns also ex end o issues o pla-
gia ism, con iden iali y, in ellec ual p ope y
igh s, anspa ency, esea ch in eg i y, and
accoun abili y.8 While he use o AI cha bo s
in pee e iew may be accep able unde open
pee e iew models, whe e he au ho ’s con-
sen o b oade sha ing o manusc ip con en
is ypically implied, his p ac ice aises e hical
and p ocedu al conce ns o he pee e iew
o manusc ip s ecei ed by jou nals ha
ollow closed (o con iden ial) pee e iew
models. In such cases, sha ing unpublished
manusc ip s wi h AI cha bo s, pa icula ly
hose hos ed on p op ie a y pla o ms ha
s o e use inpu , could comp omise e iewe
con iden iali y and iola e jou nal policies.15
Fu he mo e, o ce ain medical a icles,
e iewe s may ha e access o sensi i e pa ien
in o ma ion ha is no in ended o publica-
ion. Sha ing his in o ma ion wi h AI sys ems
may cons i u e a b each o pa ien p i acy
and, in some ju isdic ions such as he Uni ed
S a es, could po en ially iola e he Heal h
Insu ance Po abili y and Accoun abili y
Ac .16,17
Fu he mo e, ques ions emain abou he
le els o esponsibili y pee e iewe s should
be allowed o delega e o AI cha bo s and
whe he such ools could inad e en ly
A i udes and pe cep ions owa ds he use o a i icial in elligence
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 5 / 12
in luence decision-making o pe pe ua e
sys emic biases.12 Addi ionally, o he pee
e iew wo k delega ed o AI cha bo s, i is
unclea o wha ex en e iewe s can ely on
AI cha bo s o be accu a e, as AI cha bo s may
misin e p e in o ma ion and p o ide inco -
ec conclusions.8,12
The schola ly publishing communi y has
s a ed o add ess hese issues h ough
policies o edi o s and pee e iewe s.
Academic publishe s, such as Sp inge Na u e,
JAMA Ne wo k, and PLOS, ha e inco po-
a ed guidelines on he use o AI cha bo s
in hei edi o ial p ocesses, eques ing ha
pee e iewe s e ain om uploading any
manusc ip in o ma ion o AI cha bo s and
ha any use o AI o pee e iew ( ha is,
e alua ion o he claims made in he manu-
sc ip ) mus be disclosed in e iew o ms.18-20
Sage has also manda ed ha e iewe s should
no use AI cha bo s o c ea e e iew epo s
because doing so may esul in b eaches o
con iden iali y and copy igh conce ns.21
Simila ly, Science has s a ed ha AI cha bo s
canno be used o e iew pu poses, because
e iewe s a e equi ed o w i e hei e iew
epo s independen ly and should no seek
ex e nal inpu s wi hou he pe mission o he
edi o .22 O he o ganiza ions, including he
Wo ld Associa ion o Medical Edi o s and
In e na ional Commi ee o Medical Jou nal
Edi o s, ha e ecommended ha he use o
AI cha bo s should be p ohibi ed in cases
whe e con iden iali y canno be gua an eed,
and i e iewe s do use AI cha bo s, hey
mus ha e pe mission om he jou nal wi h
clea disclosu e o how he cha bo s we e
used.23,24 Howe e , codi ying and egula ing
he esponsible use o he lack he eo by pee
e iewe s may become inc easingly di icul
wi h he apidly e ol ing oles and popula i y
o AI cha bo s.
Fu he mo e, despi e g owing in e es in he
ole o AI cha bo s in pee e iew, he e is
li le esea ch on he a i udes and pe cep-
ions o pee e iewe s owa ds he use o
such ools.25 Fo example, a ecen 6-ques ion
poll o 5229 manusc ip au ho s conduc ed
by Na u e in Ma ch 2025 inqui ed abou hei
gene al a i udes owa ds he use o AI in he
publica ion p ocess, including pee e iew:
he majo i y (60% o esponden s) s a ed ha
i was ‘no app op ia e’ o use AI o ini ial
pee e iew epo s.25 Pee e iew is essen ial
o main aining he in eg i y o published
esea ch, and as in e es in using AI wi hin
schola ly publishing con inues o ise, i is
c ucial o unde s and how pee e iewe s pe -
cei e and engage wi h AI cha bo s, along wi h
he o eseeable impac s on he pee e iew
p ocess and publishing o e all. This unde -
s anding will help le e age he po en ial o AI
cha bo s while mi iga ing hei limi a ions,
ul ima ely enhancing he quali y, anspa -
ency, and ai ness o he pee e iew p ocess.
To add ess his gap, his a icle p oposes
an in e na ional c oss-sec ional su ey o
assess pee e iewe s’ expe ience wi h using
AI cha bo s, pe cei ed bene i s and chal-
lenges, e hical conce ns, and an icipa ed oles
o AI cha bo s in he pee e iew p ocess.
By p o iding insigh s in o pee e iewe s’
pe spec i es, such a s udy can in o m he
de elopmen o e hical guidelines, p ac ical
ecommenda ions, and e alua ions needed
o he e ec i e in eg a ion o AI cha bo s in
pee e iew.
Me hods
Open science s a emen
A comple e s udy p o ocol, wi h he da a
analysis plan, has been egis e ed on he
Open Science F amewo k (OSF) (h ps://
doi.o g/ 10.17605 /OSF.IO/ FHC2M) be o e
A i udes and pe cep ions owa ds he use o a i icial in elligence
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 6 / 12
ec ui ing pa icipan s o he p oposed
su ey. The s udy ma e ials and da a will be
made a ailable ia OSF as hey become a ail-
able, and he inal manusc ip will be pos ed
as a p ep in p io o submission o a pee
e iewed jou nal.
Resea ch e hics app o al
E hics app o al was ob ained om he
Uni e si y Hospi al Tübingen Resea ch E hics
Boa d (REB Numbe : 080/2025BO2) o con-
duc his s udy.
S udy design
We will conduc an anonymous, c oss-
sec ional, and closed su ey o pee e iew-
e s o medical jou nals o in es iga e hei
a i udes and pe cep ions owa ds he use o
AI cha bo s in he pee e iew p ocess. The
su ey will be adminis e ed online using
Su eyMonkey,26 a secu e web-based su ey
ool. The su ey will include bo h closed-
ended ques ions ( o example, mul iple
choice, yes/no) and open-ended ques ions
( o example, ee- ex esponses), add essing
he ollowing opics.
• Demog aphic in o ma ion: Age, sex, coun-
y o employmen , le el o educa ion, p i-
ma y a ea o expe ise, publica ion eco d,
and yea s o expe ience as a pee e iewe .
• Expe ience wi h AI cha bo s: Familia i y
wi h AI cha bo s, p io use o AI cha bo s in
hei p o essional o academic wo k, and
likelihood o pe mi ed AI cha bo use in
he pee e iew p ocess in he u u e.
• Role o AI cha bo s in pee e iew:
Pe cep ions ega ding he po en ial oles o
AI cha bo s in pee e iew, such as aiding in
iden i ying me hodological laws, de ec ing
plagia ism, e i ying e e ences, ansla ing
esea ch ma e ials, o assessing he quali y
o w i ing.
• Pe cei ed bene i s o AI cha bo s in pee
e iew: Views on po en ial bene i s, such as
educing wo kload, imp o ing e iciency,
s anda dizing e iew quali y, add essing
biases, and ensu ing g ea e consis ency in
decision-making.
• Pe cei ed challenges o AI cha bo s in
pee e iew: Conce ns abou he eliabili y
and accu acy o AI cha bo s, issues wi h
au ho ship o pee e iew epo s, isk o
ampli ying biases, lack o anspa ency, and
po en ial o e - eliance on AI.
• E hical conside a ions: Pe cep ions o
he e hical implica ions o in eg a ing AI
cha bo s in pee e iew, including conce ns
abou accoun abili y, con iden iali y, da a
p i acy, in ellec ual p ope y igh s, and
hei po en ial impac on he in eg i y o
he pee e iew p ocess.
• Addi ional commen s and eedback:
Th ough an open-ended ques ion, pa -
icipan s will be gi en he oppo uni y o
p o ide addi ional commen s and eedback
on he use o AI cha bo s in pee e iew,
sha e opinions on u u e in eg a ion, and
sugges po en ial a eas o imp o emen o
guidelines o e hical use.
The su ey will be pilo ed wi h a small g oup
o pee e iewe s, who will be excluded om
he inal su ey, o ensu e cla i y, ele ance,
and comp ehensi eness o he ques ions.
Sugges ions o po en ial pee e iewe s will
be solici ed om he co-au ho s o his s udy.
Feedback om he pilo will be inco po a ed
in o he inal su ey design.
Sampling amewo k
A comp ehensi e lis o all jou nals indexed
in MEDLINE (app oxima ely 5300 as o
No embe 2024) will be compiled along
wi h hei unique iden i ie s assigned by
he US Na ional Lib a y o Medicine (NLM
IDs).27 A sea ch s a egy using hese NLM
IDs will be o mula ed in O id MEDLINE
(a pla o m o sea ch biomedical li e a u e),
limi ing he sea ch o eco ds indexed wi hin
A i udes and pe cep ions owa ds he use o a i icial in elligence
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 7 / 12
wo mon hs p eceding he sea ch. This ime
ame was selec ed because e iewe s who
ha e submi ed a pee e iew epo du -
ing his pe iod a e likely o be s ill ac i ely
engaged in esea ch and a ailable o espond
o he in i a ion o pa icipa e in he su ey.
The ‘co esponding au ho s’ o all ypes o
esea ch a icles will be conside ed o inclu-
sion. Any duplica e eco ds will be emo ed
be o e he ec ui men p ocess. All PMID
numbe s – O id MEDLINE uses PMID, sho
o PubMed ID, numbe s o uniquely iden i y
ci a ions wi hin i s da abase – co esponding
o he iden i ied a icles will be expo ed om
O id as a .cs ile, and his ile will be p o-
cessed using an R sc ip (based on he easy-
PubMed package) o ex ac au ho names,
a ilia ed ins i u ions, and email add esses.28
Addi ionally, he ‘Find Full Tex ’ unc ion in
EndNo e will be used o e ie e he a icles
in PDF. These iles in PDF will hen be p o-
cessed wi h ano he R sc ip o ex ecogni-
ion o ex ac email add esses. All esul ing
da a will be compiled in o a mas e lis , which
will be ho oughly checked o e o s o
duplica es be o e dis ibu ing he su ey. The
sea ch s a egy can be ound he e: h ps:// os .
io/n caj .
Inclusion c i e ia
To be eligible o pa icipa ion, pa icipan s
mus ha e p e iously se ed as pee e iew-
e s o esea ch a icles submi ed o medical
jou nals (o any kind, whe eby he esea ch
hey ha e e iewed con ibu es o he ield
o medicine). Eligible pa icipan s mus ha e
comple ed and submi ed a leas one pee
e iew epo o a leas one p o essional-
le el medical jou nal (MEDLINE-indexed)
wi hin he pas 24 mon hs. Those who ha e
pee e iewed exclusi ely o s uden jou nals
( o example, high school, unde g adua e,
o g adua e jou nals) will no be eligible o
pa icipa e.
Rec ui men o pa icipan s
P ospec i e pa icipan s will be ec ui ed om
a ious academic disciplines wi hin he medi-
cal ield. We will use con enience sampling
o ec ui pa icipan s, a ge ing medical
esea che s iden i ied h ough ou sampling
amewo k. An email will be sen o po en ial
pa icipan s con aining a ec ui men message,
app o ed by he Uni e si y Hospi al Tübingen
esea ch e hics boa d, ou lining he pu pose
o he s udy and a link o he su ey. Upon
clicking he link, pa icipan s will be di ec ed
o a web page wi h an in o med consen o m.
Pa icipan s mus indica e consen on he o m
be o e p oceeding wi h he su ey. This will be
a closed su ey, meaning only in i ed pa ici-
pan s will be able o pa icipa e.
I pa icipan s do no espond o he ini ial
in i a ion email, eminde emails will be sen
a e e e y 2 weeks. The hi d eminde will
be ollowed by a 4-week wai ing pe iod o
accommoda e any emaining in e es ed pa -
icipan s be o e he su ey is closed pe ma-
nen ly. Pa icipan s will ha e a o al o 8 weeks
o comple e he su ey.
The su ey will be adminis e ed online ia
Su eyMonkey. The e will be no mone a y
compensa ion o e ed o pa icipan s, and
pa icipa ion will be comple ely olun a y.
Pa icipan s can skip ques ions hey do
no wish o answe and can wi hd aw om
he su ey a any ime by simply closing
he b owse window. Pa icipa ion will be
anonymous and con iden ial h oughou he
s udy; howe e , se ings will allow acking
he esponden s o pu poses o ollow-up
eminde s and limi ing he esponse o only
one om each pa icipan .
A i udes and pe cep ions owa ds he use o a i icial in elligence
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 8 / 12
Sample size
Gi en ha ou ini ial lis o names and email
add esses is likely o con ain duplica es, non-
unc ioning emails, and o he inconsis en-
cies, we es ima e ha app oxima ely 40,000
co esponding au ho s will be con ac ed a e
emo ing duplica es. Fo a icles wi h wo
o mo e co esponding au ho s, each will
be in i ed. This es ima e was based on he
ollowing assump ions: app oxima ely 5300
jou nals will be selec ed, and we will e ie e
he PMID numbe s o a icles published
wi hin hese jou nals o e he 2 mon hs p e-
ceding he sea ch. A leas one PMID numbe
pe jou nal pe mon h will be e ie ed, based
on ea lie s udies using his sampling ame-
wo k, leading o a o al o 120,000 PMIDs
om MEDLINE. A e p ocessing 120,000
PMIDs, we will ha e mo e han 70,000
unique names and email add esses. Based on
pas su eys conduc ed,29,30 we an icipa e a
esponse a e o 3%–5%.
Su ey ins umen
A comple e copy o he su ey o be used
o pilo ing can be ound he e: h ps:// os .
io/c 748y. The su ey will be c ea ed, dis ib-
u ed, and collec ed using Su eyMonkey, a
secu e online su ey ool (h ps:// www.su
eymonkey .com/). The su ey ins umen was
de eloped h ough a e iew o he li e a u e
and inpu om expe s in AI and scien i ic
esea ch. All au ho s o he s udy p o ocol
e iewed he su ey d a and will e iew he
edi ed e sion p io o i being ci cula ed.
The su ey begins wi h a sc eening ques-
ion o con i m ha pa icipan s iden i y
hemsel es as medical esea che s who ha e
submi ed a pee e iew epo wi hin he pas
24 mon hs o a MEDLINE-indexed jou nal.
Following his, he e a e eigh demog aphic
ques ions, asking pa icipan s abou hei
cu en posi ion, esea ch a ea, sex, age, and
coun y o employmen . Pa icipan s will
hen answe ano he eigh ques ions abou
hei amilia i y wi h and expe ience o using
AI cha bo s, ollowed by eigh mo e ques ions
abou he iews o he pa icipan s on he
ole o AI cha bo s in he pee e iew p o-
cess. Subsequen ly, pa icipan s will answe
ou mo e ques ions abou he pe cei ed
bene i s and challenges o using AI cha bo s
in pee e iew. The su ey concludes wi h an
open-ended ques ion, allowing pa icipan s
o p o ide addi ional eedback o hough s
abou he use o AI cha bo s in pee e iew.
In o al, he su ey includes 30 ques ions
and is expec ed o ake abou 15 minu es o
comple e.
Da a managemen and analysis
All esponses will be collec ed ia
Su eyMonkey, and he da a will be expo ed
o analysis using Mic oso Excel. Desc ip i e
s a is ics, such as equencies and pe cen ages,
will be calcula ed o summa ise he su ey
esponses. Demog aphic in o ma ion, such as
pa icipan ole and yea s o expe ience, will
also be analyzed o p o ide insigh s in o any
ends o pa e ns using in e en ial s a is ics.
To explo e po en ial di e ences in a i udes
and pe cep ions be ween di e en subg oups
( o example, senio e sus ea ly-ca ee pee
e iewe s), c oss abs may be c ea ed o key
demog aphic a iables p o ided he sample
size is adequa e. Quali a i e da a collec ed
h ough open-ended ques ions will unde go
induc i e coding and hema ic analysis by
wo au ho s31,32 To ensu e consis ency in cod-
ing, a pilo coding p ocess will be conduc ed,
wi h wo au ho s independen ly coding he
i s h ee su ey esponses. A e wa ds, he
au ho s will collabo a e o de elop a uni ied
coding amewo k. Once a consensus on he
codes is eached, all esponses will be g ouped
in o hema ic ca ego ies, which will be clea ly
A i udes and pe cep ions owa ds he use o a i icial in elligence
Ng e al. / doi.o g/10.3897/ese.2025.e159921 Page 9 / 12
de ined and desc ibed. The esul s will be
epo ed in he o m o ables, wi h dis inc
hemes and illus a i e quo es whe e ele an .
E hical conside a ions
E hics app o al was ob ained om he
Uni e si y Hospi al Tübingen Resea ch E hics
Boa d (REB Numbe : 080/2025BO2) o con-
duc his s udy.
Pa icipa ion in he su ey will be olun-
a y, and pa icipan s will ha e he igh o
wi hd aw om he s udy a any poin be o e
submi ing hei comple ed su ey. All
da a collec ed will be kep con iden ial and
anonymous, and no iden i ying in o ma ion
will be collec ed. The e o e, once he su ey
is submi ed, pa icipan s will no be able o
wi hd aw om he s udy as hei esponses
will be collec ed wi hou any pe sonal iden i-
ie s, including IP add esses.
Discussion
The main pu pose o his s udy is o ga he
he a i udes and pe cep ions o pee e iew-
e s o medical jou nals ega ding he use o
AI cha bo s in he pee e iew p ocess, and
he po en ial impac o such use on how pee
e iew is conduc ed a a ime when hese
AI cha bo s a e being inc easingly deployed
in schola ly publishing. AI cha bo s can
au oma e ime-consuming asks in he pee
e iew p ocess, p o ide w i ing and edi -
ing assis ance, and po en ially imp o e he
e iciency and consis ency o e iews. They
may also help lowe e iewe wo kload,
enhance he o e all quali y o eedback p o-
ided o au ho s, and allow human e iew-
e s mo e ime o e alua ing he no el y,
impo ance, and quali y o esea ch and
o p o ide insigh s ha AI cha bo s canno
p o ide. Howe e , he use o AI cha bo s
in pee e iew aises signi ican conce ns,
including e hical issues, isks o bias, isk o
manipula ion o he pee e iew p ocess,
and challenges ela ed o he accu acy and
anspa ency o au oma ed e iews. These
a e po en ial bene i s and isks o AI cha bo s
ha equi e alida ion wi h u he da a and
p ac ice in he esea ch p ocess.
To guide he esponsible in eg a ion o
AI cha bo s in pee e iew, a clea e hical
amewo k is necessa y. Unde s anding pee
e iewe s’ pe cep ions and conce ns abou
AI cha bo s is c ucial o shaping policies and
p ac ices ha add ess hese challenges. The
esul s o his su ey could in o m he de el-
opmen o guidelines by academic jou nals
o publishe s on he esponsible use o AI
cha bo s in he pee e iew p ocess, balanc-
ing hei po en ial bene i s wi h he need o
igo ous, unbiased, and anspa en e iew
p ac ices.
S eng hs and limi a ions
This p oposed s udy uses a c oss-sec ional
su ey design, which has se e al s eng hs as
well as limi a ions. The key s eng hs a e ha
he app oach is cos -e ec i e and ela i ely
quick o adminis e , and allows us o ga he
da a om a la ge sample o pee e iewe s o
medical jou nals, making he esul s gene al-
izable o he b oade communi y o medical
esea che s. As he esea che s in ou sample
a e likely o ep esen di e se medical disci-
plines, we expec a b oad ange o opinions
ega ding he use o AI cha bo s in he pee
e iew p ocess, p o iding aluable insigh s.
Addi ionally, by collec ing names and email
add esses only om he pas wo mon hs,
we minimize he likelihood o encoun e -
ing inac i e o bounced emails, which helps
ensu e he accu acy o ou con ac lis .
As o he limi a ions o he design, he e
is he po en ial o ecall bias, common o
