Bo ho ship: AI Cha bo Au ho ship A e Two Yea s
B ady D. Lund
Abs ac
Pu pose: Bo ho ship – “bo au ho ship”, o he use o a i icial in elligence ools o suppo
w i ing ac i i ies – has ans o med publishing in he ew yea s since he eme gence o Cha GPT
in la e 2022. The bane o he publishe ’s exis ence, bu a boon o w i e s, hese ools suppo
enhanced w i ing quali y and educe he amoun o ime and e o needed o u n esea ch
indings in o an accep able manusc ip . This pape discusses some o he key aspec s o
Bo ho ship as hey ha e eme ged in he pas wo yea s.
Design: This pape explo es ecen publica ions and discou se su ounding AI con ibu ions o
schola ly publica ions.
Findings: While he e a e subs an ial downsides o AI use in schola ly communica ions, he e a e
also emendous bene i s. Bo ho ship can le el he playing ield o non-na i e English speake s
ha ing o na iga e an a ena (schola ly publishing) whe e English is he lingua anca.
O iginali y/Value: This pape discusses key issues ela ed o bo ho ship and AI con ibu ions o
publica ions. I e iews and p esen s a pe spec i e on he u u e o AI au ho ship and copyedi ing
o manusc ip s.
Despi e he bes e o s om edi o s and publishe s, con en gene a ed by a i icial in elligence
(AI) cha bo s has ound i s way in o coun less publica ions since he eme gence o Cha GPT
nea ly wo yea s ago. O en imes, i is impossible o de ini i ely iden i y his usage, as AI
checke s a e any hing bu pe ec , wi h he mos ob ious clue o misconduc being he gene a ion
o ic i ious e e ences. None heless, publishe s y hei bes o police i s usage and equi e
au ho s o acknowledge he suppo o machine lea ning models in composing hei wo k. In
Lund and Naheem (2024), we ind ha i ually all academic publishe s ha e a policy abou AI
usage in publishing, wi h a ew banning i s usage al oge he . Should his be he case, hough?
A e he policies and condemna ion ill-placed?
Bo ho ship (“bo au ho ship”) is o en iewed as p oblema ic o a nuisance – a ely as a po en ial
posi i e. In he case o AI w i ing an en i e pape based on a p omp , his c i icism is wa an ed.
Howe e , his de elopmen o bo ho ship also ep esen s a key s ep owa ds elimina ing
language and g amma di e ences as a majo ac o in de e mining he ou come o a manusc ip .
In ha sense, i can be a signi ican ool o global equi y. The objec i e o mos wo ks o
li e a u e and schola ship oday is o con ey in o ma ion, no o demons a e an au ho ’s ap i ude
o w i ing in g amma ically co ec English. In ha case, why should i ma e i he pa y
esponsible o he language and g amma behind he w i ing is a human au ho o digi al bo ho ,
as long as he ideas hemsel es come om he human(s) claiming in ellec ual con ibu ion?
Song e al. (2013) and Boon and Pinx en (2021) bo h ound ha esea che s only spend abou 15
– 25% o hei ime w i ing pape s; howe e , publishing is c i ical o ob aining enu e,
p omo ion, and sala y aises. This illus a es he ime-consuming na u e o he emaining
obliga ions o esea che s. I a esea che mus in es g ea e ime in w i ing ac i i ies, due o
limi a ions in hei unde s anding o English, he lingua anca o he publishing wo ld, o
spelling and g amma issues ha a e no ge mane o he ac ual ou comes o he esea ch being
conduc ed, hen i consumes ime ha could be spen pe o ming addi ional esea ch, pe ec ing
me hodologies, and e alua ing implica ions o his wo k. Addi ionally, he le el o language
issues o some au ho s may mean ha high quali y esea ch canno be accep ed in high
quali y/highly isible publica ions. No only a e hese esea che s le behind compa ed o hei
pee s, bu he en i e science ecosys em su e s due o he delay in dissemina ion o quali y
esea ch indings.
Undoub edly, i is c i ical ha he esea ch ha is published be he in ellec ual c ea ion o hose
claiming au ho ship, ha he indings a e us wo hy and he e e ences a e accu a e. AI should
no be esponsible o he gene a ion o en i ely o iginal con en – conduc ing a li e a u e e iew
o analyzing esul s. Ra he , AI can e ise exis ing w i ing (o which he language and
g amma ical quali y may be lacking) o c ea e a well-w i en pape ha encapsula es he o iginal
esea ch o he esea che . Using AI e hically in he c ea ion o a manusc ip equi es ca e ul
human o e sigh . Figu e 1 p oposes one pa h. In his case, all o iginal “ esea ch” in ol ed in he
manusc ip mus be conduc ed by human au ho s o ools u ilized and acknowledged by hese
au ho s. Human au ho s a e also esponsible o p oducing he i s d a o he pape . They may
hen use an AI bo o e ise he quali y o he w i ing. Howe e , hey mus hen e iew he
e ised manusc ip o ensu e ha e e y hing emains accu a e and he bo has no changed he
meaning o c ea ed alse ideas, hallucina ions, o e e ences wi hin he manusc ip .
Figu e 1. P inciples o E hical Bo ho ship
<Figu e 1 Abou He e>
One po en ial s icking poin o many publishe s is whe he he use o AI ool mus be
acknowledged in a manusc ip . Assuming ha he abo e p inciples a e ollowed, i could be
a gued ha his acknowledgemen should no be equi ed. Acknowledging AI use, e en when
done so acco ding o publishe policies, c ea es a da k ma k – a symbol ha hese au ho s
u ilized assis ance in w i ing he manusc ip – ha causes hem and hei wo k o unnecessa ily
s and ou in a nega i e ligh . Os ensibly, i AI is used acco ding o he abo e p inciples, i is no
di e en han he usage o G amma ly, Spell and G amma Check in Mic oso Wo d, o any
numbe o o he ools ha millions o au ho s use wi hou acknowledgemen . Ul ima ely,
unce ain y abou AI and conce n abou hose who abuse i canno be allowed o in luence policy
o all au ho s. No ably, his does mi o he policy o Lib a y Hi Tech News (2024), whose
au ho guidelines s a e ha , “Copy-edi ing an a icle using a gene a i e AI ool/LLM in o de o
imp o e i s language and eadabili y would be pe missible as his mi o s s anda d ools al eady
employed o imp o e spelling and g amma , and uses exis ing au ho -c ea ed ma e ial, a he
han gene a ing wholly new con en , while he au ho (s) emains esponsible o he o iginal
wo k.”
Publishe s s ill appea o be unsu e abou how o add ess AI usage, as hough hey an icipa e
g ea e changes and challenges o come. Conside ing ha he Gene a i e P e-T ained
T ans o me model behind Cha GPT has e ol ed om using 117 million pa ame e s o o e 1
illion pa ame e s in a pe iod o jus i e yea s – an exponen ial g ow h in he capaci y o hese
models – his belie is pe haps well- ounded (Zhou e al., 2024). Hallucina ions and ic i ious
e e ences may soon be a hing o he pas , and AI models d a ing en i e manusc ip s may be bu
a ew yea s away. I hese de elopmen s come o ui ion, he ans o ma ion o he publishing
landscape will a exceed he scope o a simple policy. Ye , au ho s a e impac ed by policies o
oday and canno ope a e in he a ena o hypo he icals and u u ism.
The in eg a ion o AI in manusc ip p epa a ion o e s bo h bene i s and challenges. Bo ho ship
can expand oppo uni y and ele a e he quali y and pace o published wo k – highligh ing wo ks
based on in ellec ual me i a he han polished p ose – bu he limi a ions o la ge language
models mus be acknowledged. Publishe s may desi e a comp ehensi e and pe manen policy o
bo ho ship, bu wi h he pace ha hese models ha e e ol ed, ha is unlikely o be a success ul
app oach. Publishe s mus emain adap able, upda ing hei policies o e lec he cu en
capabili ies and limi a ions o hese ools – seeing hem no as a nuisance bu a he a g ea
oppo uni y o ad ance schola ly inqui y.
Re e ences
Bonn, N. A., & Pinx en, W. (2021). Ad ancing science o ad ancing ca ee s? Resea che s’
opinions on success indica o s. PLOS ONE, 16(2), e0243664.
Lib a y Hi Tech News. (2024). Au ho guidelines.
h ps://www.eme aldg ouppublishing.com/jou nal/lh n#jou nal_aims_scope
Lund, B. D., & Naheem, K. T. (2024). Can Cha GPT be an au ho ? A s udy o a i icial
in elligence au ho ship policies in op academic jou nals. Lea ned Publishing, 37(1), 13-21.
Song, D., Abedi, N., Macadam, S., & A neja, J. S. (2013). How many wo k hou s a e equisi e o
publish a manusc ip ? Plas ic and Recons uc i e Su ge y – Global Open, 1(1), e9-e10.
Zhou, Y., Ong, H., Kennedy, P., Wu, C. C., Kazam, J., Hen el, K., ... & Peng, Y. (2024).
E alua ing GPT-V4 (GPT-4 wi h Vision) on De ec ion o Radiologic Findings on Ches
Radiog aphs. Radiology, 311(2), e233270.
