Chap e 1
T ansla ion compe ence in he age o
gene a i e AI: Deba es, dilemmas,
di ec ions
Ga y Masseya& Mau een Eh ensbe ge -Dowa
aZHAW Zu ich Uni e si y o Applied Sciences, Swi ze land ( e .)
Un il ecen ly, he desc ip ion and modelling o he compe ences and skills needed
o ansla e success ully, and he ways in which hey de elop, ha e seen a s eady
e olu ion and p edic able expansion, la gely o accommoda e echnological ad-
ances and an inc easing awa eness o si ua edness. Howe e , he impac o neu al
MT and, now, gene a i e AI (GenAI) has been unp eceden ed in apidly ans o m-
ing he co e asks o ansla o s. Toge he wi h a p oli e a ion o c ea i e oles in
a di e si ying language indus y, he ola ili y indica es a pa adigm shi ha is
beginning o supplan e en he once s able epi he “ ansla o ”. I also ques ions
he e icacy o cu en desc ip ions o skills and con on s educa o s wi h dilem-
mas o balancing specialisa ion and gene alisa ion, ou inisa ion and adap i i y,
co e and ans e able skills. This chap e conside s ele an aspec s o modelling
compe ences and hei de elopmen and examines ela ed deba es and dilemmas
engaging educa o s and employe s in he cu en and o eseen language-indus y
clima e. I ou lines di ec ions o aining in he age o ansla ing wi h(ou ) GenAI,
p oposing an app oach ha , alongside co e ex ual, in e lingual ansla ion and
digi al skills, combines ans e able skills wi h human-machine/human-agen in-
e ac ion (HMI/HAI) compe ence.
Ga y Massey & Mau een Eh ensbe ge -Dow. 2026. T ansla ion compe ence in he
age o gene a i e AI: Deba es, dilemmas, di ec ions. In JC Pene , Joss Moo kens &
Masa u Yamada (eds.), Teaching ansla ion in he age o gene a i e AI: New pa adigm,
new lea ning?, 3–26. Be lin: Language Science P ess. DOI: 10.5281/zenodo.17641064
Ga y Massey & Mau een Eh ensbe ge -Dow
1 In oduc ion
1.1 Gene al
Wha elemen s cons i u e ansla ion compe ence (TC)1and how hey can be
de eloped ha e ep esen ed pi o al issues in applied T ansla ion S udies since
i s incep ion (Holmes 1988: 77). They con inue o igge conside able deba e
amongs esea che s, educa o s and p ac i ione s, and hey p esen educa o s
and hei o ganisa ions wi h dilemmas abou how and whe e o a ge esou ces.
In his chap e , we explo e some deba es and dilemmas igge ed by changes o
p o essional and educa ional p ac ices b ough abou by gene a i e a i icial in-
elligence (GenAI). We also ou line po en ial di ec ions o human agen s in he
age o ansla ing wi h and wi hou GenAI. In ou iew, his necessi a es an ap-
p oach combining ans e able skills wi h human-machine/human-agen in e -
ac ion (HMI/HAI) compe ence, alongside “old-school” ansla ion skills. Be o e
p oceeding o he deba es, dilemmas and di ec ions, howe e , we should de ine,
desc ibe and dema ca e wha exac ly we mean by p o essional in e lingual TC
in he age o GenAI.
1.2 De ini ions, desc ip ions and dema ca ion
The wo d in e lingual is used in T ansla ion S udies and he language indus y
o designa e a pa icula ype o language media ion ac i i y, namely ha which
akes place be ween na u al languages and he pa icula cul u es hey ep esen .
Alongside in e p e ing, ansla ion ep esen s a p o o ypical o m o p o essional
in e lingual language media ion. In a language indus y cha ac e ised by apidly
di e si ying job i les and asks (Bond 2018, Sla o 2020: 11–17), he p o o ypical
concep ualisa ion o wha so-called (p o essional) ansla o s do, i.e. ansla e
con en w i en in a sou ce-language (SL) documen in o a a ge -language (TL)
documen , appea s inc easingly ou moded. Language p o essionals adap i ely
engage in a whole ange o se ice p o ision as he lines be ween co e and adja-
cen se ices blu (Angelone 2023, Angelone e al. 2024: 3–5).
The majo d i e behind his s a e o a ai s has been he echnologisa ion o
he indus y as language se ice p o ide s (LSPs) ope a e mo e and mo e down-
s eam and ups eam o p e iously co e ansla ion, localisa ion and in e p e -
1This chap e applies he de ini ions o “skill” and “compe ence” om he la es EMT (2022: 3)
compe ence amewo k. A skill is he “abili y o apply knowledge and use know-how o com-
ple e asks and sol e p oblems”, while compe ence e e s o “ he p o en abili y o use knowl-
edge, social and/o me hodological abili ies, in wo k o s udy si ua ions and in p o essional
and pe sonal de elopmen ”.
4
1 T ansla ion compe ence in he age o gene a i e AI
ing ac i i ies. Indeed, he indus y i sel p e e s he gene ic e m linguis , and
wi h good eason. Since 2016, he highe accu acy and luency o neu al machine
ansla ion (NMT) engines ha e led o mo e ex ensi e au oma ion o he ansla-
ion p ocess. Linguis s ha e hus seen mo e deploymen downs eam o language
ans e (o con e sion) pe se in pos -edi ing (PE), ou pu e alua ion and quali y
assu ance asks. Mos ecen ly, he in oduc ion o au oma ed MT quali y es ima-
ion (QE) o de e mine how much human PE is equi ed is i sel eplacing mo e
cos ly human MT quali y e alua ion (Sla o 2024: 50–51). Ups eam o ans e
p ope , linguis s ha e been expanding hei epe oi e o pe o m asks ha in-
clude linguis ic consul ancy, MT p e-edi ing and mul ilingual con en c ea ion,
o example in ma ke ing, PR and co po a e communica ions (o en e e ed o
as ansc ea ion). Ne e heless, we shall e ain he e m ansla ion, bo h in line
wi h he i le o he p esen olume and in ecogni ion o he con incing a gu-
men made by do Ca mo & Moo kens (2021) ha he ul ima e e hos o ansla ion
is o help communica ion low be ween humans, ega dless o he new echnolo-
gies deployed and o he oles and ac i i ies hese necessi a e.
GenAI has now en e ed he echnological mix, and i s applica ions clea ly go
a beyond in e lingual media ion. Oppo uni ies and isks ha e been iden i ied
in highe educa ion in gene al (A las 2023, Gimpel e al. 2023, OECD 2023), whe e
he po en ial o GenAI as an educa ional ool ex ends o eache s and s uden s
alike. P ope ly used, i can suppo cou se design, ma e ials de elopmen , as-
sessmen , ex and image p oduc ion, coding, c i ical hinking and indi idualised
lea ning. Imp ope ly used, i can acili a e academic dishones y, inc ease echno-
logical dependency, a ophy human skills and agency, unde mine da a p o ec-
ion and in ellec ual p ope y igh s, ein o ce disc imina ion (due o da a bias)
and inc ease inequali y (due o unequal access).
Wi hin he na owe con ines o in e lingual ansla ion, GenAI di e s de-
cisi ely om AI-based language echnology such as NMT in being capable o
gene a ing mul ilingual ex , images and o he con en in esponse o p omp s,
including he p omp o ansla e. These p omp s elici s a is ically p obable ou -
pu gene a ed by deep-lea ning models able o p ocess and gene a e na u al lan-
guages, known as la ge language models (LLMs). The p omp - esponse cycle
can be i e a i e, allowing use s o ask o mo e in o ma ion o o p o ide mo e
de ailed p omp s should he esul s equi e imp o emen (Gimpel e al. 2023:
22). Adequa ely s uc u ing he ins uc ions ha p omp a GenAI model – also
known as p omp design o , a he mo e echnical le el, p omp enginee ing –
hus becomes cen al o i s e ec i e use. The skills and compe ences (see Foo -
no e 1) necessa y o emb ace GenAI in he educa ion and p ac ice o p o essional
5
Ga y Massey & Mau een Eh ensbe ge -Dow
ansla o s hus di e in majo espec s om hose needed o mas e CAT ools
and o he ansla ion echnologies.
I is indispu able ha such skills and compe ences a e essen ial o a language
indus y al eady ansi ioning om NMT o GenAI. Unbabel co- ounde and
CEO João G aça poin s ou ha LLMs a e a ac ing mo e esea ch and de el-
opmen , and can handle mo e complex asks han NMT, such as au oma ic PE,
sou ce co ec ion and cul u al adap a ion.2Lionb idge, one o he wo ld’s lead-
ing LSPs, claims GenAI is “ aking au oma ed ansla ion and localisa ion o new
heigh s”.3Sla o (2024: 52) epo s ha ansla ion managemen sys ems (TMS)
p o ide s a e al eady in eg a ing GPT models o le e age LLMs’ capaci y o
eph ase, summa ise and sugges new ansla ions. LLMs can be used o inco po-
a e QE in o wo k lows and gauge PE e o , do PE o e en p o ide linguis ic in-
sigh s h oughou he ansla ion wo k low. Sla o ’s esul s also show wo- hi ds
o MT p o ide s o e ing ine- uned LLMs and o e 80% o e ing MT-LLM hy-
b id solu ions. LSPs ha e he po en ial o become majo p o ide s and enable s
o mul ilingual, mul imodal GenAI con en h ough con en c ea ion, alida ion,
localisa ion, ansmission and managemen . In keeping wi h he b oadening po -
olios o LSPs, mul ilingual ex gene a ion is cu en ly “ he mos in-demand lan-
guage AI applica ion, a e machine ansla ion” (Sla o 2024: 74). Fine- uning
and p omp ing LLMs o c ea e and ansla e/localise con en a e eme ging ac i -
i ies o language p o essionals o complemen longe -es ablished PE and quali y
assu ance oles.
Compe en – o ideally expe – le e age o GenAI is a mus o all hose
aiming o wo k in he language indus y. Bu which skills and compe ences does
and will his equi e? And in a dawning age o (language) wo k inc easingly
domina ed by AI, wha alue can human in elligence and skills b ing o bea ? I
is ime o conside he deba es a ound p o essional in e lingual ansla ion and
he dilemmas ha educa o s a e acing.
2 Deba es and dilemmas
2.1 Agency
Assche (2023) aised he issue o whe he MT can ac ually be ega ded as a
p ope poin o in e es in T ansla ion S udies om a de ini ional pe spec i e.
He concludes ha i is indeed compa ible wi h bo h p esc ip i e, equi alence-
o ien ed de ini ions o wha ansla ion is as well as desc ip i e de ini ions based
2Sla o pod #216, 12 July 2024 (h ps://www.you ube.com/wa ch? =6smTEp3CXwQ)
3h ps://www.lionb idge.com/gene a i e-ai/
6
1 T ansla ion compe ence in he age o gene a i e AI
la gely on how ansla ions a e ecei ed. Bu he also poin s ou ha he pe cei ed
h ea o AI o he social and p o essional s a us o ansla o s, and o hose
who educa e hem (see also ELIS 2024: 24–25), may well ed aw de ini ional
bounda ies and dis inguish be ween MT and human ansla ion on he basis o
he pe cei ed c ea i e and mo al au hen ici y o conscious agency (Assche 2023:
14–16).
Assche ’s a gumen d aws on cu en posi ions on AI om he b oade hu-
mani ies and social sciences, such as mo al philosophy (e.g. Sebas ián & Rudy-
Hille 2021). Howe e , he e is e idence o simila p eoccupa ions in educa ion
policy and science, wi h emphasis alling on he impo ance o human agency,
including eache agency, and associa ed ans e able skills – esponsible ac ion,
c i ical hinking, sys ems hinking, logical easoning, cul u al agili y, p oblem
sol ing and emo ion egula ion (Gimpel e al. 2023, OECD 2023).
Wi hin he na owe con ines o T ansla ion S udies, i appea s ha agency
is al eady eme ging as he ulc um a ound which he alue o p o essional hu-
man ansla ion can be measu ed. Fo example, Viei a (2020: 327–329) p esen s
PE as a spec um anging om an MT-cen ed p ocess ypi ied by au oma ic PE
o human-cen ed PE whe e humans ha e ull con ol o ou pu . Human agency
mo es back and o h along he spec um as clien , commission and employe de-
mands equi e. The human-cen ed pole is mo e pala able o p o essional ansla-
o s (Viei a 2020: 327), whose esis ance o PE seems o de i e om anxie y o e
a pe cei ed loss o agency (Cadwell e al. 2018, Sakamo o 2019). In he discou se
su ounding PE, agency has hus become a ouchs one o p o essionalism and
p o essional sel -concep .
Simila ly, Rico & González Pas o (2022: 188) ound in hei p e-GenAI s udy
o a i udes o eaching MT ha he ansla ion educa o s made much o he “hu-
man ac o ”. Though he da a a e unclea on p ecisely whe e hey belie e ha
agency esides, he pa icipan s place he ansla o a he cen e o ansla ion
p oduc ion. In e es ingly, his p omp s he esea che s o posi ha “MT has o e -
come i s condi ion o ool”, and o claim ha eaching MT should adop a holis ic
app oach “beyond an ins umen alis agenda ha concen a es on he echnical
p ope ies o he echnologies” and which “e idences how he human ac o is
key o he ansla ion p ocess” (Rico & González Pas o 2022: 190, 193). Re e ing
o o he esea ch, hey also en a i ely conside whe he complemen a y e ec s
o cogni i e impai men o augmen a ion could in luence compe ence de elop-
men (Rico & González Pas o 2022: 192). One wide implica ion is ha , e en
be o e he ad en o GenAI, he pa adigma ic human agency o he ool-use
is being sub e ed by a mo e disce ning and sub le app ecia ion o he apidly
e ol ing in e ac i i y be ween ansla ion echnologies and hei use s.
7
Ga y Massey & Mau een Eh ensbe ge -Dow
The changed na u e o ha ela ionship appea s o be e lec ed in he
me apho s used o desc ibe GenAI sys ems: Cha GPT and simila sys ems a e
e e ed o as “con e sa ional agen s” (Moo kens & Gue be o A enas 2024: 75),
and educa ional sou ces desc ibe hem as a “language pa ne ”, “w i ing pa -
ne ”, “w i ing collabo a o ”, “co-pa ne o o mula e ex ”, “lea ning pa ne ”,
“in ellec ual spa ing pa ne ”, o “pa ne o gene a e codes” (e.g. A las 2023: ii:
64, Gimpel e al. 2023: 19–24). Business consul an s and IT case esea che s alike
e e o human-GenAI in e ac ion as “co-c ea ion” (e.g. Eapen e al. 2023, Nah
e al. 2023: 296).4
GenAI has now ad anced well beyond he ma e ial agen s in he “dance o
agency” so de ly explo ed by Olohan (2011: 342) in in e ac ions be ween ans-
la o s and TM. This is a ques ion no jus o deg ee (speed, accu acy and ade-
quacy o sugges ions, e c.) bu also, and mos pe inen ly, o quali y ( ecip ocal
in e ac ions and lea ning). The model pi o al o Olohan’s (2011: 344) concep ual-
isa ion o agency decisi ely dis inguishes be ween he capaci y o human agen s
o bo h esis and accommoda e o echnologies and o non-human agen s o
simply esis . Bu GenAI is pa en ly capable o accommoda ion in i s own igh
(deep lea ning). This ende s i ele an he pe cei ed di ide be ween he poles
o social and echnological de e minism ha Olohan (2011: 345) obse es among
ansla o s commen ing on TM, and which O’B ien (2024) c i icises in he call
o o e come he an agonism be ween echnology and ansla o s by ocusing on
a human-cen ed AI (HCAI) ha ampli ies a he han emula es human abili ies
(c . Shneide man 2020). As Risku & Windhage (2013: 36–37) poin ou , any ech-
nology o a e ac used by ansla o s in hei wo k aligns wi h he concep o
non-human “ac an s” in ac o -ne wo k heo y (ANT, e.g. La ou 2005). GenAI,
howe e , qui e ob iously goes u he in ul illing he condi ions o wha he
human-machine in e ac ion (HMI) sub- ield o human-agen in e ac ion (HAI)
has called “join ac i i y” (B adshaw e al. 2011: 288): “ he essence o join ac i -
i y is in e dependence […] o p oduce some hing ha is a genuine join p oduc ”.
Toge he wi h o he e ec s o (Gen)AI on he p ocesses and p ac ices o p o es-
sional ansla ion, he agency issue yields mul iple po en ial ami ica ions o he
de elopmen and exe cise o p o essional TC. I is o hese ha he nex sec ions
u n.
4The an h opomo phism is mo e han a concei . Resea ch on con e sa ional agen s used as
social companions shows ha human-AI iendship is simila o human-human ela ionships
and con ibu es o social heal h (B and zaeg e al. 2022, Cha u edi e al. 2023, Guing ich &
G aziano 2024). Vi ual agen s capable o ecip ocal adap i e beha iou a e being es ed o
use in cogni i e beha iou he apy and social skills aining (Woo e al. 2024).
8
1 T ansla ion compe ence in he age o gene a i e AI
2.2 Modelling compe ence
The ecip oci y and mu uali y o GenAI al eady go some way owa ds ealis-
ing aspec s o human-cen ed augmen ed ansla ion so engagingly discussed
by O’B ien (2024). She a gues o HCAI as a amewo k o ampli ying ansla-
o s’ abili ies and empowe men while main aining human con ol. We would
claim ha GenAI ep esen s a collabo a i e echnology capable o complemen -
ing human agency and suppo ing empowe men o a deg ee unseen in dedica ed
(augmen ed) ansla ion echnologies o da e. E en be o e Cha GPT launched he
GenAI e a la e in 2022, Fügene e al. (2021: 1552) could con iden ly asse ha
“human pe o mance can imp o e indi idually by ecei ing AI ad ice […] due
o complemen a y knowledge”. Al hough hei esul s also showed “signi ican
downsides” when AI ad ice is p o ided as a “one-size- i s-all” solu ion in g oup
decision-making, he indi idualisa ion made possible by app op ia e p omp ing
would sugges ha his is no longe an issue.
As in all complemen a y pa ne ships, he ac an s ( o use he ANT e m) mus
b ing hei own s eng hs o he able. Compe en ansla o s will o hei pa
s ill need he ans e able human skills ha enable hem o egula e cogni i e
and a ec i e pe o mance wi hou AI suppo , he expe ience and knowledge
on which hey a e p edica ed, and he compe ences necessa y o deploy hem
wi hin he socio echnical en i onmen s whe e hey wo k. These will ha e o
be aligned wi h he new key skill o in e ac ing wi h GenAI by p omp ing and
ep omp ing LLMs,5based on knowledge and expe ience o how GenAI wo ks.
We u he p opose ha his skill could be embedded in a b oade a ea o HMI/
HAI compe ence (see below) which akes due accoun o GenAI’s echnological
agency.6
I is he e ha cu en TC models e eal inadequacies. Al hough hey gi e
due weigh o he ans e able pe sonal and in e pe sonal skills ha acili a e
human agency, hey ha e pe pe ua ed an ins umen alis eaching agenda by o-
cussing almos exclusi ely on aspec s o CAT ool use (including MT) ha no
only unde play he human ac o in HMI/HAI bu also assume no echnologi-
cal agency. Fo example, in he case o he EMT compe ence amewo ks, he
only hin o possible ecip ocal e ec s be ween he echnological en i onmen
and human pe o mance has been he inclusion o an o ganisa ional and phys-
ical e gonomic desc ip o as an aspec o pe sonal compe ence (EMT 2022: 10).
5See Chap e 5 by Yamada in he p esen olume.
6This is no o say ha he e is equi alence be ween conscious human agency and AI agency.
Howe e , hey sha e ea u es ha sugges “ amily esemblances” in he sense de eloped by
Wi gens ein (e.g. 1958: 32).
9
Ga y Massey & Mau een Eh ensbe ge -Dow
Tellingly, i makes no men ion o e gonomic ac o s o echnology design and
use ha play in o cogni i e pe o mance, despi e empi ical e idence ha hey
can be majo enhance s, dis ac o s o s esso s (Eh ensbe ge -Dow e al. 2016).
O he wise, he EMT echnology compe ence ca ego y ocuses on non-in e ac i e
aspec s o echnology use: c i ically assessing, adap ing o and e ec i ely deploy-
ing a ailable esou ces, including app op ia e use o MT, ile managemen and
TMS, and demons a ing da a li e acy. In e ac i i y is isible only in in e pe -
sonal ela ions: checking, e iewing, e ising and e alua ing he wo k o o he s,
wo king in mul icul u al and mul ilingual eams, in e ac ing wi h clien s and
ne wo king wi h language p o essionals and LSPs (EMT 2022: 8–11).
The olde bu no less in luen ial PACTE G oup’s compe ence model (Hu ado
Albi 2017: 35–41) is simila in his espec . I includes echnology unde wha
is called ins umen al sub-compe ence, “ ela ed o he use o documen a ion e-
sou ces and in o ma ion and communica ion echnologies applied o ansla ion”
(Hu ado Albi 2017: 40). The desc ip o again demons a es how compe ence
models ha e uelled he ins umen alis agenda c i icised abo e.7
Ne e heless, he PACTE model has been a majo sou ce o o he s, who ha e
supplemen ed i wi h addi ional componen s whe e app op ia e. Those addi ions
e lec a g owing awa eness ha ansla ion is a less a disc e e cogni i e ac
pe o med solely by he mind o an indi idual han an e en o embodied cogni-
ion pe o med by complex sys ems in ol ing humans, hei social, echnical and
physical en i onmen s oge he wi h hei cul u al a e ac s (Risku 2010: 103).
Fo example, an in e pe sonal componen is added o he mos ecen PACTE-
in o med model: P ie o Ramos’s (2024) adap a ion o his p e ious model o legal
TC (P ie o Ramos 2011) o ins i u ional ansla ion. The model’s “ins umen al
compe ence” ca ego y goes beyond he PACTE model by speci ying he use o
eliable esou ces o in o ma ion mining, o compu e ools o ansla ion and
e ision asks, and o MT (P ie o Ramos 2024: 155). Bu i does no speak o pos-
sible implica ions o human in e ac i i y wi h agen ic AI sys ems.
Despi e some o e (and acknowledged) simila i ies, P ie o Ramos (2024: 153)
is highly c i ical o he la es EMT amewo k (2022) o ailing o ed ess wha
he sees as a downg ading o hema ic compe ence (i.e. domain specialisa ion,
e.g. in law) in i s p e ious i e a ion (EMT 2017). I is ha d o a gue agains him,
since hema ic compe ence igu es high on he lis o p o essional expec a ions
unco e ed by his own esea ch and ha o o he s, especially wi h espec o
7The ele an desc ip o s in he highes le el o he mo e ecen compe ence amewo k (EF-
FORT) a e simila , in ha hei ocus is on non-in e ac i e aspec s o echnology. h ps://
www.e o p ojec .eu/wp-con en /uploads/Le el-C-en.pd
10
1 T ansla ion compe ence in he age o gene a i e AI
legal and ins i u ional legisla i e ansla ion (Es andia i e al. 2019, La ebe 2023,
P ie o Ramos & Guzmán 2023). The esea ch ecei es equal c edence om emic
(inside ) language indus y heu is ics. Desc ibing he Sla o (2022) “expe in
he loop” model, o example, indus y commen a o Flo ian Faes esponds o a
ques ion by he i s au ho ha only language p o essionals wi h high le els o
domain and language expe ise, combined wi h app op ia e echnological and
p omp design/enginee ing compe ence, will p o ide he added human alue
needed o wo k wi h (Gen)AI and ha such expe s will s ay in high demand
(Faes & Massey 2024: 27–29).
In o he espec s, howe e , he EMT amewo k (2022: 10) alls in wi h he
b oad consensus, al eady a icula ed abo e, ha ans e able pe sonal and in e -
pe sonal skills (called gene ic o so skills) a e cen al o employabili y and o
he adap abili y needed in oday’s wo k en i onmen . The poin is echoed by
a ecen PE model (Ni zke & Hansen-Schi a 2021: 69–79), isualised as a house,
whe e he basic bilingual, ex a-linguis ic and in o ma ion esea ch compe ences
ha p o essional ansla o s possess o m he ounda ions. The oo is made o
so skills such as isk assessmen , se ice compe ences, sel -e icacy, a p o es-
sional sel -concep , an e hical a i ude, concen a ion, s ess- esis ance, logical
easoning, analy ical hinking and an a ini y wi h echnology. The model is one
o se e al de eloped o asks such as e ision and PE (Robe e al. 2023), which
we e once conside ed in eg al aspec s o TC. These e lec he g owing di e si-
ica ion o language-indus y asks and he concomi an need o language p o-
essionals o ha e ans e able skills enabling hem o adap o new ac i i ies.
2.3 De eloping compe ence
Indus y di e si ica ion has p omp ed Angelone (2023) o call o os e ing adap-
i e (as opposed o ou inised) expe ise ac oss ansla o aining cu icula by
consis en ly exposing lea ne s o challenging si ua ions whe e hey need o ap-
ply hei knowledge lexibly – i sel a ans e able skill se . Bu wha balance
should be s uck be ween ou ine and adap i i y, be ween co e and ans e able
skills? These a e wo dilemmas con on ing ansla ion eache s and hei ins i-
u ions wi h limi ed ma e ial and empo al esou ces a hei disposal. Rela ed o
hem is a hi d, namely employabili y ensions be ween domain o ask speciali-
sa ion (P ie o Ramos 2024) and a gene alis app oach mo e cong uen wi h adap-
i i y aining (EMT 2022). In all hese cases, choices ha e o be made in speci ic
educa ional con ex s as o he mos likely p ospec s o g adua e employmen ,
wi h cu iculum de elope s nimble enough o adap quickly o ma ke changes.
11
Ga y Massey & Mau een Eh ensbe ge -Dow
i le. E en when ansla o s a e explici ly being ec ui ed, hough, a b oad ange
o skills seems o be expec ed. In hei analysis o job no ices om 2005–2020,
P ie o Ramos & Guzmán (2023: 53–55) ound ha he du ies o ansla o s a
sup ana ional and in e go e nmen al o ganisa ions also included (in o de o
o e all a e age men ion): assis ance wi h o he asks, e minology wo k, e i-
sion, CAT- ool managemen and edi ing.
On he opic o how au oma ion is changing he ansla ion p o ession, Pym
& To es-Simón (2021) discuss a ious ecommenda ions o ansla o s o ocus
on wha machines canno (ye ) do, such as language se ice ad ice, se ice p o-
ision, language consul ing and high-s akes communica ion. This is consis en
wi h commen s om indus y obse e s ha ex p oduc ion wi h GenAI is less
sui able o “con en ha has a egula o y o echnical pu pose” (Sla o 2024:
75). The same epo also p edic s inc eased demand o mul ilingual expe s o
e ise co po a e-gene a ed ou pu o mee expec ed quali y le els (Sla o 2024:
63).
LSPs al eady o e a wide ange o AI- ela ed se ices (Sla o 2024: 101), many
o which equi e linguis ic expe ise ha may s ill be di icul o ind (c . Faes
e al. 2024). The la es Eu opean language indus y su ey (ELIS 2025: 34) shows
he only inc ease in echnology implemen a ion o e 2024 o be in GenAI. This
ep esen s an oppo uni y o bo h aining ins i u ions and hei g adua es and,
encou agingly, 64% o he s uden s pa icipa ing in he same su ey epo ed
using GenAI in hei aining, 19% egula ly (ELIS 2025: 37).
This aises one las impo an ques ion abou hose who ain ansla ion s u-
den s. The eache s hemsel es pa en ly equi e he (Gen)AI li e acy needed o
unde s and AI echniques, c i ically assess AI p oduc ions and ecommenda ions,
and use AI c ea i ely in hei eaching (OECD 2023: 401).11 And ins i u ions need
o ha e he s a de elopmen p ocedu es in place ha empowe hei eache s
o do so.
4 Final ake-aways
Whe he employed as ansla o s, linguis s, ansc ea o s, localise s, consul an s
o language p o essionals, ou g adua es ha e an impo an ole o play in he
language indus y despi e o because o he in oduc ion o GenAI. Bu hey will
need app op ia e skills o do so. And hose skills will, abo e all, ha e o accom-
moda e he dynamic in e ac ions o human and AI agency. Though deg ees o
domain specialisa ion a e open o deba e and will need o be de e mined by local
11Ideas o cou se design, ma e ials and eaching scena ios ha align well wi h he skills and
compe ences ou lined he e can be ound in Pym & Hao (2025).
18
1 T ansla ion compe ence in he age o gene a i e AI
condi ions, cu en TC models al eady adequa ely co e he old-school ansla-
ion and ex ual skills ha indus y demands. Bu hei ans e able and ech-
nological and digi al skills a e unde speci ied, and he in oduc ion o GenAI
calls o g ea e p ecision. K üge ’s (2024) AI li e acy amewo k is a esponse,
ou lining he digi al compe ences needed o ha ness AI e icien ly, e hically and
sus ainably. I co e s echnical ounda ions (ope a ing p inciples, aining, e c.),
assessing AI’s use ulness o (domain-)speci ic asks, in e ac ing wi h (Gen)AI,
implemen ing AI in wo k lows, and unde s anding e hical and socie al aspec s.
Pa ially o e lapping wi h elemen s o K üge ’s amewo k, ou own b ie ecom-
menda ions ep esen a se o conc e e p io i ies o emb acing GenAI in ans-
la o educa ion.
We ecommend ha collabo a i e expe ien ial lea ning should always in e-
g a e decisions abou whe he and how o deploy GenAI, oge he wi h c i ical
e alua ions o he esul an p ocesses and p oduc s. Ou comes should ocus on
de eloping speci ic ans e able skills:
• c i ical hinking
• adap i i y
• c ea i e p oblem-sol ing
• cogni i e and emo ional sel - egula ion
• sel -e icacy
• accoun abili y and a sense o e hics
• collabo a i e abili y in human and AI in e ac ions.
The ex ension o he las poin aims o de elop wha we ha e e med HMI/
HAI compe ence, which includes:
• designing e ec i e p omp s ( ansla ion pu pose, clien speci ica ions, a -
ge gen e, audience, e c.), e alua ing GenAI esponses wi h supplemen a y
esea ch (whe e app op ia e) and ep omp ing
• ecognising p iming e ec s and mi iga ing nega i e ones
• iden i ying and elimina ing bias, addi ions, omissions, hallucina ions
• unde s anding LLM da a sou ces and knowledge ime-lags.
None o he abo e should be ega ded as op ional add-ons bu mus ecei e
a leas equal weigh o so-called co e skills. Only hen will ou g adua es be
p ope ly equipped o he GenAI age, wha e e oles hey ake on.
19
Ga y Massey & Mau een Eh ensbe ge -Dow
Re e ences
Angelone, E ik. 2023. Wea ing adap i e expe ise in o ansla o aining. In
Ga y Massey, Elsa Hue as-Ba os & Da id Ka an (eds.), The human ansla o
in he 2020s, 60–73. London, UK: Rou ledge.
Angelone, E ik, Ga y Massey & Mau een Eh ensbe ge -Dow. 2024. In oduc-
ion: Con ex ualizing language indus y s udies. In Ga y Massey, Mau een
Eh ensbe ge -Dow & E ik Angelone (eds.), Handbook o he language indus y:
Con ex s, esou ces and p o iles, 1–13. Be lin, Ge many: De G uy e Mou on.
Assche , Om i. 2023. The posi ion o machine ansla ion in ansla ion s udies:
A de ini ional pe spec i e. T ansla ion Spaces 12(1). 1–20. DOI: 10.1075/ s.22035.
ass.
A las, S ephan. 2023. Cha GPT o highe educa ion and p o essional de elopmen :
A guide o con e sa ional AI. Tech. ep. Kings on: Uni e si y o Rhode Island.
h ps://digi alcommons.u i.edu/cba_ acpubs/548.
Bond, Es he . 2018. The s unning a ie y o job i les in he language indus y. h ps:
//sla o .com/ he-s unning- a ie y-o -job- i les-in- he-language-indus y/.
Bowke , Lynne & Jai o Bui ago Ci o. 2019. Machine ansla ion and global e-
sea ch: Towa ds imp o ed machine ansla ion li e acy in he schola ly commu-
ni y. Leeds, UK: Eme ald Publishing. DOI: 10.1108/9781787567214.
B adshaw, Je ey M., Paul J. Fel o ich & Ma hew Johnson. 2011. Human-agen
in e ac ion. In Guy A. Boy (ed.), The handbook o human-machine in e ac ion,
283–300. Boca Ra on, USA: CRC P ess. h ps : / / www . aylo ancis . com /
chap e s/edi /10.1201/9781315557380-14/human%E2%80%93agen -in e ac ion-
je ey-b adshaw-paul- el o ich-ma hew-johnson.
B and zaeg, Pe e Bae, Ma i a Skju e & Asbjø n Føls ad. 2022. My AI iend:
How use s o a social cha bo unde s and hei human–AI iendship. Human
Communica ion Resea ch 48. 404–429. DOI: 10.1093/hc /hqac008.
Cadwell, Pa ick, Sha on O’B ien & Ca los S. C. Teixei a. 2018. Resis ance and
accommoda ion: Fac o s o he (non-) adop ion o machine ansla ion among
p o essional ansla o s. Pe spec i es 26(3). 301–321. DOI: 10.1080/0907676X.
2017.1337210.
Ce ezo He e o, En ique, As id Schmidho e & Meli a Kole nik. 2021. An LSP
amewo k o ansla ion and in e p e ing pedagogy. Re is a de lenguas pa a
ines especi icos 37(2). 140–156. h ps://ojsspdc.ulpgc.es/ojs/index.php/LFE/
a icle/ iew/1406.
Cha u edi, Rkjul, Sanjee Ve ma, Ronnie Das & Yogesh K. Dwi edi. 2023. Social
companionship wi h a i icial in elligence: Recen ends and u u e a enues.
20
1 T ansla ion compe ence in he age o gene a i e AI
Technological Fo ecas ing & Social Change 193. 122634. DOI: 10.1016/j. ech o e.
2023.122634.
Ch is ensen, Tina Paulsen, Ma ian Flanagan & Anne Schjoldage . 2017. Mapping
ansla ion echnology esea ch in T ansla ion S udies. An in oduc ion o he
hema ic sec ion. He mes 56. 7–20. DOI: 10.7146/hjlcb. 0i56.97199.
Dale, Da id, Elena Voi a, Loic Ba aul & Ma a R. Cos a-jussà. 2023. De ec ing
and mi iga ing hallucina ions in machine ansla ion: Model in e nal wo kings
alone do well, sen ence simila i y e en be e . In P oceedings o he 61s Annual
Mee ing o he Associa ion o Compu a ional Linguis ics, ol. 1, 36–50. h ps:
//aclan hology.o g/2023.acl-long.3.pd .
do Ca mo, Félix & Joss Moo kens. 2021. Di e en ia ing edi ing, pos -edi ing and
e ision. In Maa i Koponen, B ian Mossop, Isabelle S. Robe & Gio anna
Scocche a (eds.), T ansla ion e ision and pos -edi ing: Indus y p ac ices and
cogni i e p ocesses, 35–49. London, UK: Rou ledge.
Eapen, Tojin T., Daniel J. Finkens ad , Josh Folk & Lokesh Venka aswamy. 2023.
How gene a i e AI can augmen human c ea i i y. Ha a d Business Re iew
(July-Augus ). h ps://hb .o g/2023/07/how- gene a i e-ai-can-augmen -
human-c ea i i y.
Eh ensbe ge -Dow, Mau een, And ea Hunzike Heeb, Ga y Massey, U sula Mei-
de , Silke Neumann & Heid un Becke . 2016. An in e na ional su ey o he
e gonomics o p o essional ansla ion. ILCEA 27. 1–23. h p://ilcea. e ues.
o g/4004.
ELIS. 2024. ELIS 2024 Eu opean language indus y su ey. Tech. ep. ELIS Re-
sea ch. 1–57. h ps://elis- su ey.o g/wp- con en /uploads/2024/03/ELIS-
2024-Repo .pd .
ELIS. 2025. ELIS 2025 Eu opean language indus y su ey. Tech. ep. ELIS Re-
sea ch. 1–53. h p://elis- su ey.o g/wp- con en /uploads/2025/03/ELIS-
2025_Repo .pd .
EMT. 2017. Eu opean mas e ’s in ansla ion compe ence amewo k 2017. Tech. ep.
B ussels: Eu opean Commission. 1–12.
EMT. 2022. Eu opean mas e ’s in ansla ion compe ence amewo k 2022. Tech.
ep. B ussels: Eu opean Commission. 1–12. h ps://commission.eu opa.eu/
sys em/ iles/2022-11/em _compe ence_ wk_2022_en.pd .
Es andia i, Mohammad Reza, Nas in Shok pou & Fo ough Rahimi. 2019. An
e alua ion o he EMT: Compa ibili y wi h he p o essional ansla o ’s needs.
Cogen A s & Humani ies 6(1: 1601055). 1–17. DOI: 10 . 1080 / 23311983 . 2019 .
1601055.
Faes, Flo ian, Alex Edwa ds & Anna Wyndham. 2024. Sla o Q1 2024 esea ch
b ie ing. online. h ps://sla o .com/e en /sla o -q1-2024- esea ch-b ie ing/.
21
Ga y Massey & Mau een Eh ensbe ge -Dow
Faes, Flo ian & Ga y Massey. 2024. Cha ing he language indus y: In e iew
wi h an indus y obse e . In Handbook o he language indus y, 17–31. Be lin,
Ge many: De G uy e Mou on. DOI: 10.1515/9783110716047-002.
Fa ell, Michael. 2023. Do ansla o s use machine ansla ion and i so, how?
Resul s o a su ey held among p o essional ansla o s. In Joss Moo kens &
Vilelmini Sosoni (eds.), P oceedings o T ansla ing and he Compu e 44, 49–60.
Gene a, Swi ze land: Edi ions T adulex. h ps://www. adulex.com/ a ia/
TC44-luxembou g2022.pd .
Fügene , And eas, Jö n G ahl, Alok Gup a & Wol gang Ke e . 2021. Will humans-
in- he-loop become bo gs? Me i s and pi alls o wo king wi h AI. MIS Qua -
e ly 45(3). 1527–1556. DOI: 10.25300/MISQ/2021/16553.
Gimpel, Henne , Ca oline Ruine , Man ed Schoch, Ma eike Schoop, Luis Läm-
me mann, Nils U bach, K is ina Hall, To s en Eymann, Maximilian Röglinge ,
S e en Vandi k, Alexande Mädche & S e an Decke . 2023. Unlocking he
powe o gene a i e AI models and sys ems such as GPT-4 and Cha GPT o highe
educa ion: A guide o s uden s and lec u e s. Wo king pape 02-2023. Hohen-
heim, Ge many: Uni e si y o Hohenheim. 4–46. h ps://hdl.handle.ne /10419/
270970.
Guing ich, Rose E. & Michael S. A. G aziano. 2024. Cha bo s as social companions:
how people pe cei e consciousness, human likeness, and social heal h bene i s in
machines. h ps://a xi .o g/abs/2311.10599.
Hao, Yu & An hony Pym. 2023. Whe e do ansla ion s uden s go? A s udy o he
employmen and mobili y o mas e g adua es. The In e p e e and T ansla o
T aine 17(2). 211–229. DOI: 10.1080/1750399X.2022.2084595.
Holmes, James S. 1988. T ansla ed! Pape s on li e a y ansla ion and ansla ion
s udies. Raymond an den B oeck (ed.), ol. 7 (App oaches o T ansla ion S ud-
ies). Ams e dam, Ne he lands: Rodopi.
Hu ado Albi , Ampa o (ed.). 2017. Resea ching ansla ion compe ence by PACTE
g oup. Ams e dam, Ne he lands: John Benjamins.
Kappus, Ma in. 2024. Language echnology de elope s. In Ga y Massey, Mau-
een Eh ensbe ge -Dow & E ik Angelone (eds.), Handbook o he language in-
dus y: Con ex s, esou ces and p o iles, 121–141. Be lin, Ge many: De G uy e
Mou on.
Ki aly, Don. 2000. A social cons uc i is app oach o ansla o educa ion: Empow-
e men om heo y o p ac ice. London, UK: Rou ledge.
Ki aly, Don. 2013. Towa ds a iew o ansla o compe ence as an eme gen phe-
nomenon: Thinking ou side he box(es) in ansla o educa ion. In Don Ki aly,
Sil ia Hansen-Schi a & Ka in Maksymski (eds.), New p ospec s and pe spec-
22
1 T ansla ion compe ence in he age o gene a i e AI
i es o educa ing language media o s, 197–224. Tübingen, Ge many: Na Ve -
lag.
Kolb, Wal aud. 2024. ‘I am a bi su p ised’. Li e a y ansla ion and pos -edi ing
p ocesses compa ed. In And ew Ro hwell, Andy Way & Roy Youdale (eds.),
Compu e -assis ed li e a y ansla ion, 53–68. London, UK: Rou ledge.
Koponen, Maa i , B ian Mossop, Isabelle S. Robe & Gio anna Scocche a. 2021.
T ansla ion e ision and pos -edi ing: Indus y p ac ices and cogni i e p ocesses.
London, UK: Rou ledge.
K üge , Ralph. 2016. Con ex ualising compu e -assis ed ansla ion ools and
modelling hei usabili y. T ans-kom 9(1). 114–148. h ps : / / d - nb . in o /
1105491919/34.
K üge , Ralph. 2024. Ou line o an a i icial in elligence li e acy amewo k
o ansla ion, in e p e ing and specialised communica ion. Lublin S udies in
Mode n Languages and Li e a u e 48(3). 11–23. DOI: 10.17951/lsmll.2024.48.3.11-
23.
K üge , Ralph & Janiça Hackenbuchne . 2024. A compe ence ma ix o machine
ansla ion-o ien ed da a li e acy eaching. Ta ge 36(2). 245–275. DOI: 10.1075/
a ge .22127.k u.
La ebe , Anne. 2023. Skills and knowledge equi ed o ansla o s in ins i u ional
se ings. In Tomáš S oboda, Łucja Biel & Vilelmini Sosoni (eds.), Ins i u ional
ansla o aining, 30–48. London, UK: Rou ledge.
La ou , B uno. 2005. Reassembling he social - An in oduc ion o ac o -ne wo k-
heo y. Ox o d, UK: Ox o d Uni e si y P ess.
Massey, Ga y. 2021. T ansla ion li e acy in addi ional language lea ning: Closing
he concep ual di ide be ween ansla ion and language educa ion. Bulle in
suisse de linguis ique appliquée 114. 31–45. DOI: 10.5169/seals-1030136.
Mellinge , Ch is ophe D. 2017. T ansla o s and machine ansla ion: Knowledge
and skills gaps in ansla o pedagogy. The In e p e e and T ansla o T aine
11(4). 280–293. DOI: 10.1080/1750399X.2017.1359760.
Moo kens, Joss & Ana Gue be o A enas. 2024. A i icial in elligence, au oma-
ion and he language indus y. In Ga y Massey, Mau een Eh ensbe ge -Dow
& E ik Angelone (eds.), Handbook o he language indus y: Con ex s, esou ces
and p o iles, 71–97. Be lin, Ge many: De G uy e Mou on.
Nah, Fiona Fui-Hoon, Ruilin Zheng, Keng Siau & Lang ao Chen. 2023. Gene a-
i e AI and Cha GPT: Applica ions, challenges, and AI-human collabo a ion.
Jou nal o In o ma ion Technology Case and Applica ion Resea ch 25(3). 277–304.
DOI: 10.1080/15228053.2023.2233814.
Ni zke, Jean & Sil ia Hansen-Schi a. 2021. A sho guide o pos -edi ing. Be lin,
Ge many: Language Science P ess. DOI: 10.5281/zenodo.5646896.
23
Ga y Massey & Mau een Eh ensbe ge -Dow
No d, Ch is iane. 2005. Tex analysis in ansla ion. heo y, me hodology, and di-
dac ic applica ion o a model o ansla ion-o ien ed ex analysis. T ans. by
Ch is iane No d & Penelope Spa ow. Ams e dam, Ne he lands: Rodopi.
O’B ien, Sha on. 2021. Pos -edi ing. In Y es Gambie & Luc an Doo slae (eds.),
Handbook o ansla ion s udies (Vol. 5), 177–183. Ams e dam, Ne he lands: John
Benjamins. DOI: 10.1075/h s.5.pos3.
O’B ien, Sha on. 2024. Human-cen e ed augmen ed ansla ion: Agains an ag-
onis ic dualisms. Pe spec i es 32(3). 391–406. DOI: 10.1080/0907676X.2023.
2247423.
OECD. 2023. Oppo uni ies, guidelines and gua d ails o e ec i e and equi able
use o AI in educa ion. In OECD digi al educa ion ou look 2023: Towa ds an
e ec i e digi al educa ion ecosys em, 396–407. Pa is, F ance: Édi ions OECD.
DOI: 10.1787/2b39e98b-en.
Olohan, Mae e. 2011. T ansla o s and ansla ion echnology: The dance o
agency. T ansla ion S udies 4(3). 342–357. DOI: 10.1080/14781700.2011.589656.
P e o ius, Lyne e. 2023. Fos e ing AI li e acy: A eaching p ac ice e lec ion.
Jou nal o Academic Language & Lea ning 17(1). T1–T8. h ps://jou nal.aall.
o g.au/index.php/jall/a icle/ iew/891.
P ie o Ramos, Fe nando. 2011. De eloping legal ansla ion compe ence: An in-
eg a i e p ocess-o ien ed app oach. Compa a i e Legilinguis ics 5. 7–22. DOI:
10.14746/cl.2011.5.01.
P ie o Ramos, Fe nando. 2024. Re isi ing ansla o compe ence in he age o a i-
icial in elligence: The case o legal and ins i u ional ansla ion. The In e p e e
and T ansla o T aine 18(2). 148–173. DOI: 10.1080/1750399X.2024.2344942.
P ie o Ramos, Fe nando & Diego Guzmán. 2023. Ins i u ional ansla ion p o iles:
A compa a i e analysis o desc ip o s and equi emen s. In Tomáš S oboda,
Łucja Biel & Vilelmini Sosoni (eds.), Ins i u ional ansla o aining, 49–72.
London, UK: Rou ledge.
Pym, An hony. 2003. Rede ining ansla ion compe ence in an elec onic age. In
de ence o a minimalis app oach. Me a 48(4). 481–497. DOI: 10.7202/008533a .
Pym, An hony & Yu Hao. 2025. How o augmen language skills. Gene a i e AI
and machine ansla ion in language lea ning and ansla o aining. London,
UK: Rou ledge.
Pym, An hony & Es e To es-Simón. 2021. Is au oma ion changing he ansla-
ion p o ession? In e na ional Jou nal o he Sociology o Language 2021(270).
39–57. DOI: 10.1515/ijsl-2020-0015.
Rico, Celia & Diana González Pas o . 2022. The ole o machine ansla ion in
ansla ion educa ion: A hema ic analysis o ansla o educa o s’ belie s. The
24
1 T ansla ion compe ence in he age o gene a i e AI
In e na ional Jou nal o T ansla ion and In e p e ing Resea ch 14(1). 177–197.
DOI: 10.12807/ i.114201.2022.a010.
Risku, Hanna. 2010. A cogni i e scien i ic iew on echnical communica ion and
ansla ion. Do embodimen and si ua edness eally make a di e ence? Ta ge
22(1). 94–111. DOI: 10.1075/ a ge .22.1.06 is.
Risku, Hanna & Flo ian Windhage . 2013. Ex ended ansla ion: A sociocogni i e
esea ch agenda. Ta ge 25(1). 33–45. DOI: 10.1075/ a ge .25.1.04 is.
Robe , Isabelle S., Jim J. U eel & I is Sch ij e . 2023. T ansla ion, ansla ion
e ision and pos -edi ing compe ence models: Whe e a e we now? In Ga y
Massey, Elsa Hue as-Ba os & Da id Ka an (eds.), The human ansla o in
he 2020s, 44–59. London, UK: Rou ledge.
Sakamo o, Akiko. 2019. Unin ended consequences o ansla ion echnologies:
F om p ojec manage s’ pe spec i es. Pe spec i es 27(1). 58–73. DOI: 10.1080/
0907676X.2018.1473452.
Sawye , Da id B., F ank Aus e mühl & Vanessa En íquez Raído (eds.). 2019. The
e ol ing cu iculum in in e p e e and ansla o educa ion. Ams e dam, Ne he -
lands: John Benjamins.
Sebas ián, Miguel Ángel & Fe nando Rudy-Hille . 2021. Fi s -pe son ep esen-
a ions and esponsible agency in AI. Syn hese 199. 7061–7079. DOI: 10.1007/
s11229-021-03105-8.
Shneide man, Ben. 2020. Human cen e ed a i icial in elligence: Th ee esh
ideas. AIS T ansac ions on Human-Compu e In e ac ion 12(3). 109–124. DOI:
10.17705/1 hci.00131.
Sla o . 2020. Sla o 2020 language indus y ma ke epo . Tech. ep. Sla o . h ps:
//sla o .com/sla o -2020-language-indus y-ma ke - epo /.
Sla o . 2022. Sla o 2022 language indus y ma ke epo . Tech. ep. Sla o . h ps:
//sla o .com/sla o -2022-language-indus y-ma ke - epo /.
Sla o . 2024. Sla o 2024 language indus y ma ke epo . Tech. ep. Sla o . h ps:
//sla o .com/2024-language-indus y-ma ke - epo -language-ai-edi ion/.
Teubne , Timm, Ch is oph M. Fla h, Ch is o Weinha d & Wil an de Aals .
2023. Welcome o he e a o Cha GPT e al. The p ospec s o la ge language
models. Business & In o ma ion Sys ems Enginee ing 65(2). 95–101. DOI: 10 .
1007/s12599-023-00795-x.
Vandepi e, Sonia, Lie e Jooken, Robe M. Maie & Binghan Zhang. 2018. Lin-
guis ics. In Lie en D’huls & Y es Gambie (eds.), A His o y o Mode n T ans-
la ion Knowledge, 357–366. Ams e dam, Ne he lands: John Benjamins.
Viei a, Lucas Nunes. 2020. Pos -edi ing o machine ansla ion. In Minako
O’Hagan (ed.), The Rou ledge handbook o ansla ion and echnology, 319–335.
London, UK: Rou ledge.
25
Ga y Massey & Mau een Eh ensbe ge -Dow
Wi gens ein, Ludwig. 1958. Philosophical in es iga ions. T ans. by G. E. M.
Anscombe. 2nd edn. Ox o d, UK: Blackwell.
Woo, Jieyeon, Kazuhi o Shida a, Ca he ine Acha d, Hi oki Tanaka, Sa oshi Naka-
mu a & Ca he ine Pelachaud. 2024. Adap i e i ual agen : Design and e alu-
a ion o eal- ime human-agen in e ac ion. In e na ional Jou nal o Human-
Compu e S udies 190. 103321. DOI: 10.1016/j.ijhcs.2024.103321.
26
