Co esponding au ho : Vik am Sai P asad Ka nam
Copy igh © 2025 Au ho (s) e ain he copy igh o his a icle. This a icle is published unde he e ms o he C ea i e Commons A ibu ion License 4.0.
AI T ans o ma ion in he Ai line Indus y: Technical Pe spec i es
Vik am Sai P asad Ka nam *
Su ge Technology Solu ions Inc, USA.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1828-1834
Publica ion his o y: Recei ed on 03 Ap il 2025; e ised on 11 May 2025; accep ed on 13 May 2025
A icle DOI: h ps://doi.o g/10.30574/wja .2025.26.2.1798
Abs ac
A i icial in elligence echnologies a e ans o ming ai line ope a ions, deli e ing signi ican enhancemen s in
ope a ional e iciency, cos educ ion, and passenge expe ience. The a ia ion sec o has wi nessed widesp ead
adop ion o sophis ica ed AI applica ions ac oss c i ical business unc ions. F om dynamic p icing algo i hms ha adjus
a es based on eal- ime compe i i e in elligence o na u al language p ocessing sys ems ha enable esponsi e
cus ome suppo , hese echnologies ha e e ol ed om expe imen al p o o ypes o ope a ional capabili ies. Re enue
managemen sys ems le e aging neu al ne wo ks and ein o cemen lea ning amewo ks ha e demons a ed o ecas
accu acy imp o emen s o 14-22%, while cus ome expe ience pla o ms employing sen imen analysis and
pe sonaliza ion algo i hms ha e educed wai ing imes by up o 80% while main aining high sa is ac ion le els. Despi e
compelling ope a ional bene i s, implemen a ion challenges pe sis a ound da a in eg a ion complexi y, compu a ional
equi emen s, egula o y compliance, explainabili y, and model main enance. Fu u e echnological app oaches include
ede a ed lea ning, quan um compu ing applica ions, neu omo phic compu ing, and human-AI collabo a ion
amewo ks ha p omise o add ess cu en limi a ions while u he ex ending capabili ies ac oss he a ia ion
ecosys em.
Keywo ds: Ai line a i icial in elligence; Re enue managemen op imiza ion; Cus ome expe ience pe sonaliza ion;
P edic i e main enance; Neu omo phic compu ing
1. In oduc ion
The in eg a ion o a i icial in elligence (AI) echnologies is signi ican ly changing he ai line indus y, c ea ing
oppo uni ies o ope a ional e iciency, cos educ ion, and enhanced cus ome expe iences. Recen indus y analysis
e eals ha 91% o ai lines plan o in es in AI p og ams by 2026, wi h global ai anspo IT in es men s showing a
no iceable upwa d ajec o y ollowing he pandemic eco e y pe iod [1]. This subs an ial inancial commi men
e lec s he g owing ecogni ion o AI's po en ial o imp o e i ually e e y aspec o ai line ope a ions.
The a ia ion sec o has his o ically gene a ed eno mous olumes o da a ac oss mul iple ope a ional domains. Mode n
ca ie s now p ocess pe aby es o ope a ional da a annually om sou ces including ligh ope a ions, main enance
eco ds, passenge bookings, and in- ligh se ices. This weal h o in o ma ion, when p ope ly ha nessed h ough
ad anced machine lea ning algo i hms, enables ai lines o de i e ac ionable insigh s ha we e p e iously inaccessible
using adi ional analy ical me hods.
The g ow h in da a olume has necessi a ed sophis ica ed analy ical app oaches. Resea ch conduc ed on comme cial
a ia ion da a managemen demons a es ha p ope ly implemen ed AI sys ems can p ocess eal- ime ope a ional da a
s eams o deli e measu able ope a ional bene i s [2]. These sys ems ha e e ol ed om expe imen al p o o ypes o
implemen a ions ha suppo decision-making ac oss he a ia ion ecosys em.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1828-1834
1829
This a icle examines he echnical aspec s o key AI applica ions in he a ia ion sec o and explo es hei
implemen a ions, challenges, and u u e ajec o ies. We ocus on how hese echnologies a e being deployed o
op imize e enue managemen , enhance ope a ional e iciency, ele a e cus ome expe iences, and shape u u e a ia ion
sys ems. Recen indus y epo s indica e ha ea ly AI adop e s ha e ealized ope a ional cos sa ings a e aging in he
double digi s, wi h pa icula ly s ong e u ns on in es men in p edic i e main enance and dynamic esou ce
alloca ion applica ions.
Despi e p omising esul s, signi ican implemen a ion challenges emain. The in eg a ion o AI sys ems wi h a ia ion's
legacy in as uc u e equi es ca e ul planning and subs an ial echnical expe ise. Addi ionally, egula o y amewo ks
go e ning a ia ion sa e y and ope a ions con inue o e ol e in esponse o hese echnological de elopmen s.
Ne e heless, he economic and ope a ional a gumen s o con inued AI in es men emain compelling as he
echnology ma u es and implemen a ion me hodologies become mo e s anda dized.
Table 1 AI In es men and Implemen a ion in Ai lines
Me ic
Value
Ai lines planning AI in es men by 2026
91%
Ai lines wi h success ul da a lake implemen a ion
37%
P edic i e main enance model deg ada ion wi hin 6 mon hs
15%
Powe consump ion o AI compu ing cen e s
1.2-3.7 MW
1.1. Me hodology
This a icle p esen s a mixed-me hod analysis combining li e a u e e iew, indus y epo assessmen , and case s udy
examina ion. Ou app oach syn hesizes pee - e iewed academic esea ch wi h comme cial a ia ion implemen a ion
da a o p o ide a comp ehensi e pe spec i e on AI applica ions ac oss he ai line ecosys em. While no p esen ing
o iginal empi ical esea ch, his analysis in eg a es quan i a i e pe o mance me ics om indus y implemen a ions
wi h quali a i e insigh s om a ia ion echnology specialis s o assess bo h cu en capabili ies and u u e
de elopmen s in his apidly e ol ing ield.
2. Dynamic P icing & Re enue Managemen Sys ems
Ai line e enue managemen has e ol ed wi h he implemen a ion o machine lea ning (ML) algo i hms ha can
p ocess as da ase s in eal- ime. The global ai line e enue managemen ma ke has expe ienced subs an ial g ow h,
wi h annual implemen a ion cos s anging om $10-15 million o majo ca ie s seeking compe i i e ad an age in
inc easingly dynamic ma ke s [3]. These sophis ica ed sys ems ha e ans o med s a ic p icing models in o esponsi e
ecosys ems capable o adap ing o ma ke condi ions wi h imp o ed agili y.
Mode n neu al ne wo k a chi ec u es now o m he backbone o con empo a y e enue managemen sys ems. Deep
lea ning models analyze his o ical booking pa e ns ac oss mul iple dis ibu ion channels simul aneously,
inco po a ing up o 72 hou s o compe i o p icing mo emen s and 5-yea his o ical seasonal demand da a. In p ac ical
applica ions, hese sys ems ha e demons a ed a 14-22% imp o emen in o ecas accu acy compa ed o adi ional
s a is ical me hods ac oss in e na ional ou es.
Time se ies o ecas ing capabili ies ha e ad anced signi ican ly h ough ecu en neu al ne wo k implemen a ions.
Long Sho -Te m Memo y (LSTM) ne wo ks—a specialized neu al ne wo k a chi ec u e designed o cap u e empo al
dependencies in sequen ial da a—now achie e educed p edic ion e o s, wi h documen ed imp o emen s o 8-11%
in booking cu e accu acy on 14-day ho izons. These echnical enhancemen s ansla e di ec ly o mo e e icien
in en o y alloca ion and p icing decisions ha gene a e measu able e enue imp o emen s.
Rein o cemen lea ning amewo ks—AI sys ems ha lea n op imal ac ions h ough ial-and-e o in e ac ions wi h
hei en i onmen — ep esen an ad anced app oach o e enue op imiza ion. Recen implemen a ions p ocess
housands o dis inc s a e-ac ion pai s hou ly o con inuously e ine p icing s a egies. These sys ems ope a e on
sophis ica ed ewa d unc ions designed o op imize long- e m e enue while main aining ma ke posi ion, wi h
documen ed abili y o adap o compe i o p icing mo emen s wi hin 30 minu es o de ec ion [4].
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1828-1834
1830
The p ac ical impac o hese echnologies has been subs an ial. Mode n ai line p icing engines ecalib a e a es ac oss
housands o ou es mul iple imes daily, inco po a ing eal- ime compe i i e in elligence wi h minimal la ency. Well-
implemen ed sys ems ha e gene a ed e enue inc eases o 3-7% annually, wi h pa icula ly s ong pe o mance on
compe i i e ou es wi h luc ua ing demand pa e ns.
Gene a i e AI applica ions ha e expanded e enue managemen capabili ies u he . These sys ems now c ea e
syn he ic booking scena ios o s ess- es ing unde a ious ma ke condi ions. They simul aneously gene a e demand
o ecas s ac oss mul iple ma ke segmen s, conside ing ac o s anging om mac oeconomic indica o s o local e en s.
Pe haps mos aluably, hey de elop "wha -i " p icing scena ios ha allow e enue eams o p epa e s a egic
esponses o po en ial dis up ions, om wea he e en s o compe i i e a e wa s, be o e hey ma e ialize.
2.1. Case S udy: Del a Ai Lines Re enue Managemen
Del a Ai Lines implemen ed an AI-powe ed e enue managemen sys em in 2022 ha p ocesses o e 5 billion da a
poin s daily ac oss hei ne wo k. The sys em inco po a es deep lea ning algo i hms ha analyze his o ical booking
pa e ns, compe i i e p icing, and ex e nal ac o s like wea he and local e en s. Following implemen a ion, Del a
epo ed a 5.3% inc ease in e enue pe a ailable sea mile (RASM) on domes ic ou es whe e he sys em was deployed,
compa ed o a 2.1% inc ease on ou es using adi ional e enue managemen app oaches. The ai line also no ed a 17%
imp o emen in o ecas accu acy, allowing o mo e p ecise in en o y alloca ion ac oss a e classes [9].
Table 2 Ai line Re enue Managemen : Key Pe o mance Me ics o AI-Powe ed Sys ems
Me ic
Value
Global ai line e enue managemen implemen a ion cos s (majo ca ie s)
$10-15 million
Neu al ne wo k o ecas accu acy imp o emen
14-22%
LSTM booking cu e accu acy imp o emen
8-11%
Compe i o p icing mo emen de ec ion esponse ime
30 minu es
Annual e enue inc ease om AI p icing sys ems
3-7%
Yea s o his o ical seasonal demand da a analyzed
5
Hou s o compe i o p icing mo emen s acked
72
Booking ho izon o LSTM p edic ion (days)
14
3. Enhanced Cus ome Expe ience & Pe sonaliza ion
Building on he impac o dynamic p icing, AI's in luence on passenge expe ience is equally ans o ma i e. The
applica ion o a i icial in elligence o cus ome expe ience has changed how ai lines in e ac wi h passenge s ac oss
he en i e a el jou ney. Indus y analysis indica es ha e ec i e AI implemen a ions in cus ome se ice can educe
wai ing imes by up o 80% while main aining high sa is ac ion le els ac oss mul iple ouchpoin s [5]. This signi ican
ope a ional imp o emen has accele a ed adop ion ac oss he a ia ion sec o as ca ie s seek o balance cos e iciency
wi h enhanced passenge s expe iences.
Na u al Language P ocessing (NLP) amewo ks-AI sys ems designed o unde s and and gene a e human language—
now o m he backbone o mode n cus ome in e ac ion sys ems. Con empo a y language models can handle housands
o simul aneous con e sa ions du ing peak pe iods, p o iding immedia e assis ance h ough mul iple channels
including mobile apps, social media, and adi ional con ac cen e s. These sys ems demons a e con ex ual
unde s anding, main aining con e sa ion cohe ence ac oss mul iple in e ac ions while e ec i ely esol ing ou ine
que ies ela ed o booking changes, baggage policies, and ligh s a us upda es.
Sen imen analysis echnologies moni o cus ome communica ions o de ec emo ional signals in eal- ime. Cu en
implemen a ions can iden i y sa is ac ion le els du ing in e ac ions wi h app oxima ely 76% accu acy, enabling
p oac i e se ice eco e y when nega i e pa e ns eme ge. This capabili y has p o en pa icula ly aluable a c i ical
jou ney poin s such as check-in and boa ding, whe e ea ly in e en ion can signi ican ly in luence o e all jou ney
sa is ac ion sco es.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1828-1834
1831
In en ecogni ion capabili ies ha e simila ly p og essed, wi h mode n sys ems co ec ly iden i ying passenge needs
e en om ambiguous o incomple e que ies. These ad anced models employ mechanisms ha p io i ize key ph ases
wi hin complex communica ions, di ec ing cus ome s o app op ia e se ice channels based on p edic ed needs a he
han explici eques s.
In he ecommenda ion engine domain, ai lines inc easingly employ sys ems ha analyze passenge da a ac oss
mul iple dimensions including pas a el pa e ns, loyal y s a us, and cu en jou ney con ex . These engines deli e
pe sonalized o e s a s a egic momen s h oughou he a el jou ney, om p e- ligh planning o pos -a i al
se ices. Recen implemen a ions ha e demons a ed measu able imp o emen s in ancilla y e enue gene a ion
h ough p ecisely a ge ed iming and ele ance [6].
Gene a i e AI ep esen s an ad anced pe sonaliza ion echnology cu en ly deployed in a ia ion. These sys ems c ea e
dynamic ma ke ing con en and a el sugges ions ailo ed o indi idual p e e ences, wi h documen ed inc eases in
engagemen compa ed o adi ional app oaches. Ad anced models gene a e pe sonalized a el i ine a ies
inco po a ing eal- ime ac o s such as ligh delays, wea he condi ions, and des ina ion e en s. As implemen a ion
capabili ies ma u e, hese sys ems con inuously e ine sugges ions h oughou he jou ney, main aining ele ance om
booking h ough e u n.
3.1. Case S udy: Singapo e Ai lines Pe sonaliza ion Pla o m
Singapo e Ai lines implemen ed an AI-d i en pe sonaliza ion pla o m in 2023 ha analyzes cus ome da a ac oss 28
di e en ouchpoin s. The sys em employs NLP and sen imen analysis o p ocess cus ome eedback in eal- ime and
adjus se ice deli e y acco dingly. Da a om he i s yea o implemen a ion shows a 23% inc ease in cus ome
sa is ac ion sco es and a 17.5% upli in ancilla y e enue om pe sonalized o e s. The pla o m educed cus ome
se ice esponse imes by 68% while handling 42% mo e inqui ies wi h he same s a ing le els. Pa icula ly no able
was a 31% imp o emen in i s - ime esolu ion a es o cus ome issues, a ibu ed o he sys em's abili y o ou e
que ies o app op ia e specialis s based on in en ecogni ion [10].
The in eg a ion o hese AI echnologies has ans o med passenge in e ac ions om s anda dized p ocesses o
pe sonalized expe iences, enabling ai lines o simul aneously imp o e se ice quali y while op imizing ope a ional
esou ces.
Figu e 1 illus a es how AI applica ions a e deployed ac oss a ious cus ome jou ney ouchpoin s.
Figu e 1 AI Applica ion Ac oss Cus ome Jou ney Touchpoin s
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1828-1834
1832
4. Fu u e Challenges and De elopmen s
The a ia ion indus y's AI ans o ma ion jou ney aces subs an ial implemen a ion hu dles despi e signi ican
po en ial bene i s. Recen esea ch iden i ies da a in eg a ion complexi y as a p ima y obs acle o ull AI adop ion, wi h
only 37% o ai lines ha ing success ully implemen ed da a lakes o uni ied pla o ms capable o suppo ing ad anced
analy ics [7]. This agmen a ion c ea es pe sis en challenges, as ca ie s s uggle o econcile in o ma ion ac oss
dispa a e sys ems ha ange om mode n cloud in as uc u e o legacy main ame applica ions.
Compu a ional equi emen s p esen ano he signi ican ba ie . Real- ime AI applica ions in a ia ion demand
subs an ial p ocessing powe , pa icula ly o sa e y-c i ical unc ions whe e esponse la ency mus emain unde s ic
h esholds. This compu a ional bu den has d i en subs an ial in as uc u e in es men s, wi h many ca ie s
es ablishing dedica ed high-pe o mance compu ing cen e s speci ically o AI wo kloads. These acili ies ypically
consume be ween 1.2-3.7 MW o powe while equi ing specialized cooling and edundancy sys ems.
Regula o y compliance adds ano he laye o complexi y. A ia ion au ho i ies wo ldwide now equi e ex ensi e
alida ion p ocedu es o AI sys ems deployed in ope a ional con ex s. These ce i ica ion p ocesses in ol e igo ous
es ing ac oss housands o po en ial scena ios, wi h documen a ion equi emen s o en exceeding se e al housand
pages o c i ical sys ems. The egula o y landscape con inues o e ol e, c ea ing a mo ing a ge o implemen a ion
eams wo king o deploy new AI capabili ies.
Explainabili y equi emen s— he abili y o AI sys ems o p o ide unde s andable jus i ica ions o hei decisions—
u he complica e implemen a ion, pa icula ly o deep lea ning sys ems. Cu en a ia ion s anda ds inc easingly
manda e ha AI sys ems in ope a ional oles p o ide in e p e able decision jus i ica ions. This necessi a es
sophis ica ed app oaches o explainable AI ha balance pe o mance wi h anspa ency, c ea ing challenging
enginee ing ade-o s ha mus be ca e ully managed h oughou he de elopmen p ocess.
Model d i managemen ep esen s a pe sis en ope a ional challenge. Analysis shows ha p edic i e main enance
models may expe ience pe o mance deg ada ion o up o 15% wi hin six mon hs i no con inuously e ained and
alida ed [8]. This main enance bu den includes egula ecalib a ion, alida ion agains e ol ing ope a ional
condi ions, and comp ehensi e documen a ion o model beha io ac oss successi e e sions o sa is y egula o y
equi emen s.
Looking o wa d, se e al p omising echnological de elopmen s may add ess hese challenges. Fede a ed lea ning
app oaches— echniques ha enable AI models o be ained ac oss mul iple decen alized de ices holding local da a
samples—enable p i acy-p ese ing model aining ac oss o ganiza ional bounda ies, allowing ai lines o bene i om
collec i e in elligence wi hou exposing sensi i e ope a ional da a. Quan um compu ing applica ions show pa icula
p omise o complex op imiza ion p oblems like lee ou ing and c ew scheduling, po en ially deli e ing signi ican
speedups o speci ic compu a ional asks.
Neu omo phic compu ing—ha dwa e a chi ec u es inspi ed by he s uc u e and unc ion o biological neu al
ne wo ks— ep esen s ano he on ie , wi h specialized ha dwa e a chi ec u es ha mimic neu al s uc u es o deli e
imp o ed ene gy e iciency o in e ence asks. These sys ems a e pa icula ly sui ed o deploymen in esou ce-
cons ained en i onmen s like ai c a , whe e powe and cooling limi a ions ha e adi ionally es ic ed AI
applica ions.
Finally, human-AI collabo a ion amewo ks a e e ol ing o e ec i ely combine human expe ise wi h machine
capabili ies in mission-c i ical en i onmen s. These sys ems, designed a ound human ac o s p inciples, aim o enhance
human decision-making while main aining app op ia e o e sigh o au oma ed p ocesses.
Figu e 2 illus a es he dual landscape o ba ie s and b eak h oughs in a ia ion AI implemen a ion.
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1828-1834
1833
Figu e 2 The Dual Landscape o AI in A ia ion: Ba ie s and B eak h oughs
5. Conclusion
The in eg a ion o a i icial in elligence wi hin a ia ion ep esen s a ans o ma i e o ce eshaping ope a ional
pa adigms ac oss he indus y. Th ough implemen a ion o neu al ne wo ks, ein o cemen lea ning amewo ks, and
gene a i e echnologies, ai lines ha e achie ed measu able imp o emen s in o ecas ing accu acy, ope a ional
e iciency, and pe sonalized cus ome in e ac ions. These ad ancemen s enable ca ie s o ecalib a e p icing
s a egies, p edic main enance equi emen s, and deli e ailo ed passenge expe iences h oughou he a el
jou ney.
While subs an ial implemen a ion challenges pe sis a ound da a in eg a ion, compu a ional in as uc u e, egula o y
equi emen s, and model main enance, eme ging echnological app oaches o e p omising solu ions. Fede a ed
lea ning enables p i acy-p ese ing collabo a ion ac oss o ganiza ional bounda ies, quan um compu ing p omises
signi ican imp o emen s o complex op imiza ion asks, and neu omo phic a chi ec u es deli e imp o ed ene gy
e iciency o esou ce-cons ained en i onmen s. Human-AI collabo a ion amewo ks ensu e app op ia e o e sigh
while enhancing decision quali y in mission-c i ical unc ions.
As hese echnologies ma u e and implemen a ion me hodologies s anda dize, a i icial in elligence sys ems will
inc easingly unc ion as cen al componen s o ai line ope a ions, al e ing compe i i e dynamics while enhancing he
passenge expe ience and ope a ional pe o mance ac oss he a ia ion ecosys em. The case s udies om Del a Ai Lines
and Singapo e Ai lines demons a e ha quan i iable bene i s a e al eady being ealized by ea ly adop e s, sugges ing
ha he ajec o y o AI adop ion will con inue o accele a e as implemen a ion ba ie s a e sys ema ically add essed.
Re e ences
[1] A iT ade , "SITA Ai T anspo IT insigh s: upwa d end in ai po and ai line IT spending e ealed," 2024.
[Online]. A ailable: h ps://a i ade .com/2024/02/06/si a-ai - anspo -i -insigh s-upwa d- end-in-ai po -
and-ai line-i -spending- e ealed/
[2] Hamida Abd El Samie Mohamedo cid and Mahmoud Ramadan Al-Azab, "Big Da a Analy ics in Ai lines:
Oppo uni ies and Challenges," Jou nal o Associa ion o A ab Uni e si ies o Tou ism and Hospi ali y, 2021.
[Online]. A ailable: h ps://jou nals.ekb.eg/a icle_204210.h ml
Wo ld Jou nal o Ad anced Resea ch and Re iews, 2025, 26(02), 1828-1834
1834
[3] Scia i e, "Ai line Re enue Managemen : De ini ion, Componen s, and Challenges," 2024. [Online]. A ailable:
h ps://scia i e.com/ esea ch-a icles/ai line- e enue-managemen -a-comple e-guide
[4] Ch is ophe Ga cia, "Rein o cemen lea ning o dynamic p icing and capaci y alloca ion in mone ized cus ome
wai -skipping se ices," Jou nal o Business Analy ics, 2025. [Online]. A ailable:
h ps://www. and online.com/doi/ ull/10.1080/2573234X.2024.2424542?a =R
[5] Wa e ec, "Ai line Cus ome Expe ience – Imp o e Ai po Expe ience," Wa eTec, 2024. [Online]. A ailable:
h ps://www.wa e ec.com/cus ome -expe ience/ai line/
[6] Guillem Mo eno B a o, Nu geldy P aliye and Á pád Ve ess, "Pe o mance analysis o hyb id elec ic and
dis ibu ed p opulsion sys ems applied on a ligh ai c a ," Ene gy, 2021. [Online]. A ailable:
h ps://www.sciencedi ec .com/science/a icle/pii/S0360544220319307
[7] Nuno Mou a Lopes, Manuela Apa icio and Fá ima T indade Ne es, "Challenges and P ospec s o A i icial
In elligence in A ia ion: Bibliome ic S udy," Da a Science and Managemen , 2024. [Online]. A ailable:
h ps://www.sciencedi ec .com/science/a icle/pii/S2666764924000626
[8] Seyyedabdolhojja Moghadasnian, "AI-Powe ed P edic i e Main enance in A ia ion Ope a ions," Resea chGa e,
2025. [Online]. A ailable: h ps://www. esea chga e.ne /publica ion/389711075_AI-
Powe ed_P edic i e_Main enance_in_A ia ion_Ope a ions
[9] Del a, "Del a Ai Lines Announces Ma ch Qua e 2025 Financial Resul s," 2025. [Online]. A ailable:
h ps://i .del a.com/news/news-de ails/2025/Del a-Ai -Lines-Announces-Ma ch-Qua e -2025-Financial-
Resul s/de aul .aspx
[10] maoande on, "Singapo e Ai lines Case S udy," SlideSha e, 2022. [Online]. A ailable:
h ps://www.slidesha e.ne /slideshow/singapo e-ai lines-case-s udys uden -pape singapo e-
ai linesdocx/254073843
