Re iew A icle h ps://doi.o g/10.12973/ijem.11.2.203
In e na ional Jou nal o Educa ional Me hodology
Volume 11, Issue 2, 203 - 216.
ISSN: 2469-9632
h p://www.ijem.com/
In eg a ion o A i icial In elligence and Machine Lea ning in Educa ion: A
Sys ema ic Re iew
Manuel Reina-Pa ado*
Uni e si y o Se illa, SPAIN
Ped o Román-G a án
Uni e si y o Se illa, SPAIN
Ca los He ás-Gómez
Uni e si y o Se illa, SPAIN
Recei ed: Janua y 30, 2025 ▪ Re ised: Ma ch 12, 2025 ▪ Accep ed: Ap il 14, 2025
Abs ac :
This PRISMA-based sys ema ic e iew analyzes how a i icial in elligence (AI) and Machine Lea ning (ML) a e
in eg a ed in o educa ional ins i u ions, examining he challenges and oppo uni ies associa ed wi h hei adop ion. Th ough a
s uc u ed selec ion p ocess, 27 ele an s udies published be ween 2019 and 2023 we e analyzed. The esul s indica e ha AI
adop ion in educa ion emains une en, wi h signi ican ba ie s such as limi ed eache aining, echnological accessibili y gaps,
and e hical conce ns. Howe e , indings also highligh p omising applica ions, including AI-d i en adap i e lea ning sys ems,
in elligen u o ing, and au oma ed assessmen ools ha enhance pe sonalized educa ion. The geog aphical analysis e eals
ha mos esea ch on AI in educa ion o igina es om No h Ame ica, Eu ope, and Eas Asia, while de eloping egions emain
unde ep esen ed. Wi hou s a egic in eg a ion, he une en implemen a ion o AI in educa ion may widen social inequali ies,
limi ing access o inno a i e lea ning oppo uni ies o disad an aged popula ions. Consequen ly, his s udy unde sco es he
u gen need o policies and eache aining p og ams o ensu e equi able AI adop ion in educa ion, os e ing an inclusi e and
echnologically p epa ed lea ning en i onmen .
Keywo ds: A i icial in elligence, Cha GPT, educa ion, machine lea ning, eache aining.
To ci e his a icle: Reina-Pa ado, M., Román-G a án, P., & He ás-Gómez, C. (2025). In eg a ion o a i icial in elligence and
machine lea ning in educa ion: A sys ema ic e iew. In e na ional Jou nal o Educa ional Me hodology, 11(2), 203-216.
h ps://doi.o g/10.12973/ijem.11.2.203
In oduc ion
Inc easingly, echnologies a e doing hings ha p e iously only humans could do. This is so un il a ime comes when I
do p ac ically all o hem. This is wha we ha e been calling echnological globaliza ion (Kule o e al., 2021; Rod íguez-
Ga cía e al., 2020).
In educa ion, hese echnologies a e ha ing an amazing impac , enabling access o mo e and di e en educa ional
esou ces (Hoosain e al., 2020). New online lea ning pla o ms and mul imedia con en a e eme ging o enhance
eaching quali y. These ools le e age a i icial in elligence (AI) o analyze s uden pe o mance, iden i ying pa e ns
such as inc eased ailu e a es in speci ic asks, p olonged esponse imes in exams, o dec eased engagemen wi h he
pla o m o e ime. By de ec ing hese ends, educa o s can in e ene mo e e ec i ely o suppo s uden lea ning.
Despi e wha i may seem, he inco po a ion o AI in he educa ional ield is s ill in a de eloping phase, and is
cha ac e ized by a slow adop ion p ocess. This is because eme ging echnologies end o a i e in educa ion a e
consolida ing hemsel es in o he sec o s, such as p oduc ion o social, and because he e is a his o ical pe cep ion ha
eaching is a ask ha belongs only o human beings (Nicole i & de Oli ei a, 2020). Th ough di e en media, i has
been possible o show ha some p o essionals in he educa ion sec o a e eluc an o inco po a e AI (Cha e jee &
Bha acha jee, 2020; Kadhim & Hassan, 2020).
Likewise, i is clea o hink ha AI ep esen s a ool wi h eno mous po en ial o add ess c i ical p oblems such as
demo i a ion and school d opou (Salas-Rueda e al., 2020), challenges ha signi ican ly a ec he cu en educa ion
sys em, especially since he e a e many eache s who a e no able o p o ide solu ions ela ed o his issue, and AI can
* Co esponding au ho :
Manuel Reina-Pa ado, Uni e si y o Se illa, Spain.  m [email protected]
© 2025 The au ho (s); licensee IJEM by RAHPSODE LTD, UK. Open Access - This a icle is dis ibu ed unde he e ms and condi ions o he
C ea i e Commons A ibu ion License (h ps://c ea i ecommons.o g/licenses/by/4.0/).
204  REINA-PARRADO ET AL. / AI and ML in Educa ion: Sys ema ic Re iew
p o ide hem wi h poin s o iew no con empla ed un il hen. Howe e , i s applica ion in his con ex emains an
unexplo ed e i o y, o e ing mul iple oppo uni ies o inno a ion and imp o emen o educa ional p ocesses.
The ield o AI in educa ion is a ac ing inc easing in e es due o i s inno a i e na u e and he challenges aced by
eache s in e ms o hei aining in compu a ional hinking. The lack o p e ious expe ience and he complexi y o his
discipline om i s ounda ions make i c ucial o explo e how AI is being applied in educa ional con ex s and wha
me hods a e mos sui able o inco po a e i e ec i ely (Chang e al., 2022).
To unde s and he cu en landscape, i is p oposed o ca y ou a sys ema ic e iew ha analyzes he use o Machine
Lea ning as pa o AI. This app oach will p o ide an inno a i e pe spec i e on how hese echnologies a e
ans o ming he educa ional ield, wi h he aim o p epa ing s uden s o ake ad an age o he echnological ools
a ailable in he u u e.
Acco ding o Zawacki-Rich e e al. (2019), he pu pose o a sys ema ic e iew is o answe speci ic ques ions using a
s uc u ed, anspa en , and ep oducible sea ch me hodology, using clea inclusion and exclusion c i e ia o selec
ele an s udies. This p ocess includes coding and da a ex ac ion, which acili a es he syn hesis o indings o iden i y
bo h hei p ac ical applica ions and exis ing con adic ions o limi a ions.
The inco po a ion o ad anced echnologies such as AI in he class oom ep esen s a complex and p og essi e p ocess
(P endes-Espinosa & Ce dán-Ca agena, 2021). In his sense, a ho ough e iew o he mos ecen esea ch on he
applica ion o AI in educa ion can o e a de ailed and c i ical analysis o he cu en s a e o his eme ging ield.
The PRISMA (P e e ed Repo ing I ems o Sys ema ic Re iews and Me a-Analyses) s a emen is de eloped as a guide
in ended o p o ide a s anda d app oach o conduc ing sys ema ic e iews. I s main pu pose is o uni y p ocedu es,
ensu ing ha he esul s ob ained a e consis en and use ul o u u e esea ch in he a ea o s udy (Page e al., 2021;
U ú ia & Bon ill, 2010). Al hough PRISMA is no a sys ema ic e iew in i sel , i is an essen ial ool o ca y i ou in a
igo ous and s uc u ed manne .
PRISMA includes 27 elemen s ha mus be conside ed du ing he de elopmen o he esea ch. These poin s allow o
he gene a ion o well- ounded conclusions ha e lec he s a e o knowledge on a speci ic opic, de ined acco ding o
he selec ion c i e ia es ablished o he e iew (Page e al., 2021; U ú ia & Bon ill, 2010).
Since sys ema ic e iews a e dynamic, i is necessa y o delimi a ime ame ha de e mines which a icles will be
included. Howe e , i is ecommended o upda e hem pe iodically o inco po a e new s udies ha expand and en ich
he analysis (Page e al., 2021; Su e al., 2022; Talan, 2021; U ú ia & Bon ill, 2010).
This esea ch pu sues he main objec i e o analyzing how AI and ML a e in eg a ed in o educa ional ins i u ions,
examining he challenges and oppo uni ies associa ed wi h hei adop ion. To achie e his, he s udy es ablishes he
ollowing speci ic objec i es: o iden i y and compile key bibliog aphic sou ces ela ed o he mos ou s anding
publica ions in he ield; and examine he indings o such publica ions o assess he impac o using a i icial
in elligence h ough ML-powe ed cha bo s in educa ion.
In o de o achie e hese pu poses, speci ic objec i es ha e been de ined ha allow hese issues o be add essed in a
s uc u ed way h ough analysis: o explo e he ways in which AI, h ough ML-based cha bo s, is being implemen ed in
he educa ional ield; o in es iga e eache s' pe cep ions pe cep ions o he educa ional alue o AI and s uden s
de i ed om he e iew on he in eg a ion o AI in he class oom; and iden i y he AI ools and p og ams mos used in
he educa ional con ex .
Me hodology
This pape p esen s a sys ema ic e iew o scien i ic publica ions ocused on he use o AI in he educa ional ield. Fo
i s de elopmen , he PRISMA me hodology was used (Hu on e al., 2016; Page & Mohe , 2017; U ú ia & Bon ill, 2010).
PRISMA is s uc u ed in o 27 elemen s ha se e as a e e ence o ensu e ha sys ema ic e iews a e use ul and
unde s andable o eade s (Hu on e al., 2016). The ini ial e sion o PRISMA, published in 2009, gained wide
accep ance and applica ion in a ious ields. Howe e , he upda ed 2020 e sion, used in his s udy, in oduces
signi ican imp o emen s, including he possibili y o conduc ing dynamic sys ema ic e iews, also known as "li e",
which can be con inuously upda ed based on new da a (Page e al., 2021).
Da abase Selec ion and A icle Selec ion
The sys ema ic e iew ocused on h ee undamen al inclusion c i e ia: Machine Lea ning (ML), Educa ion, and AI. The
eason why hese h ee c i e ia ha e been used was he ollowing:
a) ML is a undamen al b anch o AI ha allows machines o analyze da a and lea n om i o make p edic ions o make
decisions. This c i e ion was included due o i s g owing impac on he de elopmen o educa ional ools and
applica ions. ML echniques such as classi ica ion algo i hms, eg ession, and neu al ne wo ks a e he basis o many
sys ems o pe sonalizing lea ning, adap i e assessmen , and analyzing s uden beha io . I s inclusion allows us o
In e na ional Jou nal o Educa ional Me hodology  205
analyze how hese echnologies a e being used in he design and implemen a ion o inno a i e educa ional
me hodologies.
b) The educa ional ield is he key con ex o his e iew, as i seeks o explo e how AI-based echnologies a e
ans o ming eaching and lea ning me hods. Including educa ion as a c i e ion ensu es ha he selec ed s udies a e
di ec ly ela ed o he impac o hese echnologies on educa ional ins i u ions, pedagogical p ac ices, and he aining
o s uden s and eache s. In addi ion, his c i e ion helps o unde s and he speci ic bene i s and challenges ha
educa ional communi ies ace when inco po a ing AI in o hei p ocesses.
c) AI is he gene al amewo k unde which applica ions such as ML and o he sub ields a e de eloped. This c i e ion is
undamen al because i allows us o iden i y esea ch ha no only deals wi h he p ac ical use o AI, bu also wi h i s
e hical, social and pedagogical implica ions in he educa ional ield. By including AI as a c i e ion, a b oade ision is
gua an eed ha encompasses bo h speci ic applica ions and heo e ical e lec ions on i s ole in he ans o ma ion o
educa ion.
A icles ha me he h ee es ablished c i e ia we e selec ed o analysis. This selec ion p ocess was ca ied ou using
da abases in e na ionally ecognized o hei ele ance in he indexing o scien i ic li e a u e, such as SCOPUS, Web o
Science (WoS) and ERIC. The choice o hese sea ch sou ces is based on hei ele ance, co e age and in e na ional
ecogni ion in he indexing o scien i ic and academic li e a u e.
The combina ion o SCOPUS, WoS and ERIC ensu es comp ehensi e co e age o ele an s udies, balancing dep h o
analysis in he ield o educa ion (ERIC) wi h he b ead h and quali y o mul idisciplina y publica ions (Scopus and
WoS). This allows o a mo e comple e iew o how a i icial in elligence and machine lea ning a e impac ing he ield
o educa ion, while ensu ing ha he sou ces selec ed a e igo ous and eliable.
The sea ch was ca ied ou using a deduc i e app oach, using keywo ds as he main il e and applying sea ch s ings
based on Boolean ope a o s, speci ically: "Machine Lea ning" AND "Educa ion" AND "A i icial In elligence". The
selec ed a icles we e expo ed o a sp eadshee in Excel o ma o acili a e hei e iew and subsequen o ganiza ion.
Subsequen ly, hey we e ans e ed o an ex e nal pla o m o he managemen o bibliog aphic e e ences: Mendeley
(desk op e sion). This so wa e, which is eely accessible, is designed o collec , o ganize and ci e esea ch. I allows
da a o be impo ed di ec ly om compa ible websi es and ecognized o ma s, which acili a es he managemen o
bibliog aphic in o ma ion (Ba sky, 2010).
Documen Fil e ing and Selec ion
Nex , he esul s we e limi ed o documen s wi h access o he ull ex and published in inal e sions (excluding
p ep in s, since hey a e no de ini i e and could be al e ed in he inal publica ion). The inclusion/exclusion c i e ia
we e as ollows:
a) Inclusion c i e ia
- Focused on ML as pa o AI applied o he educa ional ield.
- Add esses he use o AI based on ML echniques.
- Published be ween 2019 and 2023.
- Applicable o any educa ion sys em, wi hou geog aphical o con ex ual es ic ions.
- Includes p ac ical applica ions o AI o case s udies ha explo e po en ial educa ional uses o hese echnologies.
- I is limi ed o a icles published in academic jou nals.
- W i en in Spanish o English.
- A ailable in i s en i e y wi h ull access o he ex .
- Final documen s.
b) Exclusion c i e ia:
- I does no add ess machine lea ning o AI as main axes.
- I is limi ed o dealing wi h a speci ic opic whe e AI is used only as a seconda y ool o achie e o he objec i es.
- Published in 2018 o in p e ious yea s.
- Focused exclusi ely on a speci ic geog aphical con ex .
- I does no include p ac ical applica ions o AI o case s udies ha explo e po en ial educa ional uses o his
echnology.
206  REINA-PARRADO ET AL. / AI and ML in Educa ion: Sys ema ic Re iew
- I does no co espond o a icles om academic jou nals.
- W i en in languages o he han Spanish o English.
- The a icle is no a ailable o ull eading.
- P ep in s.
The selec ion o a icles om 2019 onwa ds ensu es ha he included s udies a e ep esen a i e o he mos cu en
echnologies, me hodologies and policies, maximising he ele ance and impac o he esul s o his sys ema ic e iew.
I is p ecisely om 2019 ha a no able inc ease in he adop ion o AI-based ools in educa ional con ex s has been
obse ed. This pe iod coincides wi h he ise o pla o ms such as Cha GPT, adap i e lea ning sys ems, and educa ional
cha bo s, making s udies published in his ime in e al especially ele an o analysis.
A e his i s il e ing, he 297 esul s ob ained a e p esen ed in Figu e 1.
Figu e 1. Ini ial Sc eening
The ini ial p ocessing o he collec ed da a was ca ied ou using a sp eadshee in Excel o ma . Fo he SCOPUS and
WoS da abases, he p ocedu e consis ed o selec ing he p e iously il e ed a icles and expo ing hem in CSV (Comma
Sepa a ed Values) o ma , which is compa ible wi h Excel and allows di ec in eg a ion.
In he case o ERIC, he expo gene a es a ile in nbib o ma , a ile ype used p ima ily in he PubMed da abase. This
o ma is no di ec ly compa ible wi h Excel, so i was necessa y o use he Zo e o e e ence manage (Alonso-A é alo,
2015).
The PubMed da abase was no used o manusc ip sc eening because i s que y could ha e inco po a ed s udies wi h a
bias owa ds biomedical applica ions o AI, which is no he main objec i e o he analysis.
Once he esul s o he h ee da abases we e ob ained in sepa a e Excel o ma iles, hey we e manually combined in o
a single documen . This consolida ion allowed he da a o be uni ied in o a single XLSX ile, om which he subsequen
e iew and analysis was ca ied ou .
A icle Re iew
The e iew began wi h a o al o 297 a icles (Figu e 2), which we e consolida ed in o a single Excel sp eadshee o
acili a e hei ini ial managemen .
0
52
000
52
0
111
6
18 14
149
34
53
9
0 0
96
0
20
40
60
80
100
120
140
160
Con e ence
p oceedings
Jou nal a icles Book chap e s Commen s Con e ence
pape s
TOTAL
ERIC SCOPUS WoS
In e na ional Jou nal o Educa ional Me hodology  207
Figu e 2. Flow Diag am o he Phases Acco ding o he PRISMA Model
The eco ds we e o ganized and hose ha we e duplica e (44) we e elimina ed, no ing he da abases o o igin o each
a icle. A e his p ocess, 253 documen s emained o con inue wi h he analysis.
The i s il e applied consis ed o selec ing only a icles published in academic jou nals, educing he numbe o 164.
Documen s disca ded a his s age we e a chi ed o possible u u e esea ch ela ed o his line o s udy. These 164
a icles we e hen e alua ed by e iewing hei i les, abs ac s, and keywo ds. We included o excluded hem on he
basis ha hey me he objec i es o he e iew.
A e applying he inclusion and exclusion c i e ia, 112 s udies we e excluded due o una ailabili y o ull ex ,
i ele ance o he esea ch objec i es, o lack o empi ical da a. Following his p ocess, a o al o 52 a icles we e
downloaded and managed using he bibliog aphic e e ence so wa e Mendeley o de ailed eading and e alua ion,
aligning wi h he p inciples o Open Science. Du ing his comp ehensi e e iew, p e iously es ablished inclusion and
exclusion c i e ia we e eapplied. Among he main easons o disca ding i ems we e he ollowing:
- The a icles deal wi h AI and ML angen ially, ocusing on he speci ic con en ha was sough o wo k wi h hese
echnologies, which does no mee he c i e ion ha he ocus should be on AI and ML as cen al elemen s.
- The s udies could no be ex apola ed o b oad educa ion sys ems, as hey we e limi ed o e y speci ic con ex s o
condi ions, ailing o mee he c i e ion o being applicable o any educa ion sys em.
- Al hough hey add essed opics ela ed o he objec o s udy, hey did no include p ac ical applica ions o AI o case
s udies ha showed speci ic uses in he educa ional ield, which con a enes he es ablished c i e ia.
Iden i ica ion
Sc eening
Selec ion
Iden i ied eco ds o :
Da abase (n=3)
Reco ds (n=297)
Reco ds dele ed be o e
sc eening: Duplica e eco ds
(n=44)
Excluded eco ds (n=89)
Repo s eques ed
o eco e y
(n=164)
Reco ds no eco e ed (n=112)
Excluded epo s: Th ough a
ho ough eading
(n=27)
Repo s E alua ed
o Eligibili y
(n=52)
Re ised Reco ds
(n=253)
S udies included
in he e iew
(n=25)
Iden i ica ion o s udies h ough da abases and egis ies

208  REINA-PARRADO ET AL. / AI and ML in Educa ion: Sys ema ic Re iew
- Some pape s iden i ied as case s udies u ned ou o be sys ema ic e iews, ailing o mee he ype o app oach
sough o his e iew.
Finally, a e his p ocess, 25 manusc ip s we e iden i ied ha me all he inclusion c i e ia and we e selec ed o be pa
o he analysis (Figu e 2).
Final Selec ion o A icles
A e applying he inclusion and exclusion c i e ia, and ca ying ou a de ailed analysis o he selec ed manusc ip s, 25
inal documen s we e ob ained. These we e o ganized in a speci ic sub olde wi hin he Mendeley bibliog aphic
manage and, subsequen ly, expo ed o an Excel sp eadshee o acili a e hei handling and subsequen analysis
(Table 1).
Table 1. Fi s Sc eening
No.
Yea
Au ho
Educa ional
Le el
Ti le o he A icle
Da abase
1
2023
Billingsley e
al.
K-12
Can a obo be a scien is ? De eloping
s uden s' epis emic insigh h ough a
lesson explo ing he ole o human
c ea i i y in as onomy
SCOPUS
--
--
2
2022
Jokhan e al.
Highe
Educa ion
Inc eased digi al esou ce consump ion
in highe educa ional ins i u ions and
he a i icial in elligence ole in
in o ming decisions ela ed o s uden
pe o mance
SCOPUS
Wos
--
3
2022
Nuankaew
Highe
Educa ion /
Gene al
Sel - egula ed lea ning model in
educa ional da a mining
--
--
ERIC
4
2022
Niyogisubizo
e al.
No speci ied
Ti le no a ailable in e e ences
SCOPUS
Wos
ERIC
5
2022
G unhu e al.
Medical
Educa ion
Needs, challenges, and applica ions o
a i icial in elligence in medical
educa ion cu iculum
SCOPUS
--
--
6
2022
Zammi e al.
K-12
Lea n o machine lea n ia games in he
class oom
SCOPUS
--
--
7
2022
Vi -Singh and
Kan -Hi an
Highe
Educa ion
The impac o AI on eaching and
lea ning in highe educa ion echnology
SCOPUS
--
--
8
2021
S adelmann e
al.
Gene al /
Hyb id
The AI-A las: Didac ics o eaching AI
and machine lea ning on-si e, online,
and hyb id
SCOPUS
Wos
--
9
2021
Kule o e al.
Highe
Educa ion
Explo ing oppo uni ies and challenges
o a i icial in elligence and machine
lea ning in highe educa ion ins i u ions
SCOPUS
--
--
10
2021
Lampos e al.
Special
Educa ion /
Au ism
An a i icial in elligence app oach o
selec ing e ec i e eache
communica ion s a egies in au ism
educa ion
SCOPUS
--
--
11
2021
Ha a i e al.
Gene al / K-
12
Assessmen and lea ning in knowledge
spaces (ALEKS) adap i e sys em impac
on s uden s' pe cep ion and sel -
egula ed lea ning skills
SCOPUS
--
--
12
2021
Ac ion
No speci ied
Ti le no a ailable in e e ences
--
Wos
--
13
2021
Pu e al.
Gene al /
Bibliome ic
Iden i ica ion and analysis o co e opics
in educa ional a i icial in elligence
esea ch: A bibliome ic analysis
--
Wos
--
14
2021
Kanglang
Highe
Educa ion
A i icial in elligence (AI) and
ansla ion eaching: A c i ical
pe spec i e on he ans o ma ion o
educa ion
SCOPUS
--
--
In e na ional Jou nal o Educa ional Me hodology  209
Table 1. Con inued
No.
Yea
Au ho
Educa ional
Le el
Ti le o he A icle
Da abase
15
2021
D uzhinina e
al.
Ma hema ics /
Gene al
De elopmen o an in eg a ed complex
o knowledge base and ools o expe
sys ems o assessing knowledge o
s uden s in ma hema ics
SCOPUS
Wos
ERIC
16
2020
Salas-Rueda e
al.
Gene al /
Highe Ed
Impac o he web applica ion o he
educa ional p ocess on he compound
in e es conside ing da a science
--
Wos
--
17
2020
Ma ques e al.
K-12
Teaching machine lea ning in school: A
sys ema ic mapping o he s a e o he
a
SCOPUS
--
--
18
2020
Muniasamy
and Alasi y
No speci ied
Ti le no a ailable in e e ences
--
--
ERIC
19
2020
Rod íguez-
Ga cía e al.
K-12 /
Gene al
Lea ningML: A ool o os e
compu a ional hinking skills h ough
p ac ical a i icial in elligence p ojec s
--
Wos
ERIC
20
2020
Kadhim and
Hassan
Highe
Educa ion
Towa ds in elligen e-lea ning sys ems:
A hyb id model o p edic ing he
lea ning con inui y in I aqi highe
educa ion
SCOPUS
--
--
21
2019
How & Hung
K-12 / STEAM
Educing AI- hinking in science,
echnology, enginee ing, a s, and
ma hema ics (STEAM) educa ion
SCOPUS
Wos
--
22
2019
Ruipé ez-
Valien e e al.
Highe Ed /
MOOCs
Using machine lea ning o de ec
'mul iple-accoun ' chea ing and analyze
he in luence o s uden and p oblem
ea u es
--
--
ERIC
23
2019
Palasund am
e al.
Highe
Educa ion /
Cha bo s
Sequence o sequence model
pe o mance o educa ion cha bo
SCOPUS
--
--
24
2019
Sha ma e al.
Highe Ed /
Gene al
Building pipelines o educa ional da a
using AI and mul imodal analy ics: A
'g ey-box' app oach
--
Wos
--
25
2019
Luckin and
Cuku o a
Gene al
Designing educa ional echnologies in
he age o AI: A lea ning sciences-d i en
app oach
SCOPUS
--
--
Sc eening Upda e
In a i s phase, he sys ema ic e iew conside ed he a icles a ailable in he da abases up o Feb ua y 2023. Howe e ,
be o e concluding he i s epo in July 2023, a second sea ch was conduc ed o include a icles published be ween
he wo pe iods, which had no been ini ially e alua ed.
This addi ional sea ch ollowed he same c i e ia and p ocedu es p e iously es ablished, al hough he ime ange was
adjus ed o include only documen s published in 2023. A e applying he inclusion and exclusion c i e ia, wo new
a icles we e iden i ied (Table 2) ha me he equi emen s and p o ided ele an conclusions o he s udy. Thus, he
inal e iew included a o al o 27 a icles.
Table 2. New I ems Added
No.
Yea
Au ho
Educa ional Le el
Ti le o he A icle
Da abase
26
2023
Gilson
e al.
Medical Educa ion
How does Cha GPT pe o m on he Uni ed
S a es medical licensing examina ion? The
implica ions o la ge language models o
medical educa ion and knowledge assessmen
SCOPUS
--
--
27
2023
Chung
e al.
Gene al / AI
Applica ions
Technology accep ance p edic ion o obo-
ad iso s by machine lea ning
SCOPUS
--
--
Figu e 2 p esen s a clus e map gene a ed wi h VOS iewe om he keywo ds ex ac ed om he analyzed a icles.
This map shows he close connec ion be ween machine lea ning (ML) and a i icial in elligence (AI), highligh ing how
210  REINA-PARRADO ET AL. / AI and ML in Educa ion: Sys ema ic Re iew
bo h concep s a e in e ela ed and complemen each o he in he p ocessing and classi ica ion o da a using hese
echnologies.
In addi ion, Figu es 2 and 3 shows ha AI is linked o a ious subjec a eas, while ML is di ec ly associa ed wi h he
da a ha AI collec s and p ocesses.
Connec ed o smalle nodes, such as "adap i e educa ion" o "da a p ocessing," hese e ms can be in e ed o ep esen
speci ic applica ions o a eas o in e es ela ed o ML and AI.
Ano he iden i ied clus e is composed o e ms such as "STEM," "educa ional assessmen ," and " echnology in he
class oom," indica ing ha se e al a icles speci ically explo e he applica ion o AI and ML echnologies in eaching and
e alua ion p ocesses wi hin educa ional con ex s (How & Hung, 2019; Sha ma e al., 2019).
Ano he g oup o key wo ds such as "adap i e lea ning," "pe sonalized educa ion," and "s uden engagemen " e lec s
he g owing esea ch in e es in AI-d i en sys ems designed o ailo educa ional expe iences o indi idual lea ne
needs (G unhu e al., 2022; Zammi e al., 2022).
Finally, ano he g oup ocuses on "e hical conce ns," " eache aining," and " echnological ba ie s," highligh ing he
challenges ha educa o s ace when in eg a ing hese echnologies in o hei p ac ices (Kadhim & Hassan, 2020; Singh
& Hi an, 2022). Toge he , hese clus e s illus a e he di e si y o esea ch opics wi hin he ield and ein o ce he
mul idimensional impac o AI and ML on educa ion.
Figu e 2. Main Keywo ds Ex ac ed om he Re iewed S udies on AI and ML in Educa ion
This ein o ces he idea ha ML and AI a e in e dependen componen s, unde lining he ele ance o his s udy and i s
con ibu ion o he unde s anding o hese echnologies in he educa ional ield.
Figu e 3. Map o he Rela ionship Be ween A icles. Made Wi h VOS iewe
Resul s
A e sc eening scien i ic manusc ip s, a o al o 27 s udies published be ween 2019 and 2023 in a ious da abases
we e analyzed. The esul s show ha he main sou ces o in o ma ion used we e SCOPUS, Web o Science (WoS) and
ERIC.
The o al numbe o s udies pe da abase has been:
In e na ional Jou nal o Educa ional Me hodology  211
- SCOPUS: 19 s udies.
- Web o Science (WoS): 10 s udies.
- ERIC: 6 s udies.
The empo al dis ibu ion e lec s a p og essi e g ow h in he publica ion o esea ch ela ed o AI and ML:
- 2023: 3 s udies (Billingsley e al., 2023; Chung e al., 2023; Gilson e al., 2023).
- 2022: 6 s udies (G unhu e al., 2022; Jokhan e al., 2022; Niyogisubizo e al., 2022; Nuankaew, 2022; Singh & Hi an,
2022; Zammi e al., 2022).
- 2021: 8 s udies (D uzhinina e al., 2021; Ha a i e al., 2021; Kanglang & A zaal, 2021; Kule o e al., 2021; Lampos e al.,
2021; Pu e al., 2021; S adelmann e al., 2021; Talan, 2021).
- 2020: 6 s udies (Kadhim & Hassan, 2020; Ma ques e al., 2020; Muniasamy & Alasi y, 2020; Rod íguez-Ga cía e al.,
2020; Salas-Rueda e al., 2020).
- 2019: 4 s udies (How & Hung, 2019; Palasund am e al., 2019; Ruipé ez-Valien e e al., 2019; Sha ma e al., 2019).
Following he analysis o he selec ed s udies, a hema ic classi ica ion was de eloped o align he esul s wi h he
esea ch objec i es and o be e unde s and how AI and ML a e being in eg a ed in o educa ional ins i u ions. Th ee
main hemes eme ged, e lec ing he di e se applica ions and challenges iden i ied in he li e a u e. This ca ego iza ion
also highligh s he po en ial and limi a ions o AI and ML in educa ional con ex s.
1) Cu iculum de elopmen o AI educa ion (7 s udies):
These s udies ocus on in eg a ing AI li e acy and compu a ional hinking in o educa ional cu icula, pa icula ly a he
K-12 le el. How and Hung (2019) p oposed inco po a ing AI- hinking in o STEM educa ion, aiming o os e analy ical
skills om ea ly s ages. Ma ques e al. (2020) conduc ed a sys ema ic mapping o machine lea ning eaching in schools,
iden i ying a g owing in e es in p ac ical AI educa ion. Zammi e al. (2022) examined game-based lea ning
app oaches o each AI concep s, demons a ing posi i e e ec s on s uden engagemen . Simila ly, Rod íguez-Ga cía e
al. (2020) in oduced he Lea ningML ool o p omo e compu a ional hinking skills h ough AI p ojec s, while
S adelmann e al. (2021) explo ed didac ic s a egies o eaching AI in hyb id en i onmen s. Talan (2021) ein o ced
he impo ance o including AI in educa ion h ough a bibliome ic s udy, and Kanglang and A zaal (2021) c i ically
examined he ole o AI in ansla ion eaching, s essing cu iculum adap a ion needs.
2) Implemen a ion o AI and ML ools in educa ional pla o ms (11 s udies):
This heme includes s udies ha analyze he use o AI-powe ed ools and pla o ms designed o enhance lea ning
expe iences and educa ional p ocesses. Palasund am e al. (2019) es ed he e ec i eness o cha bo s in suppo ing
s uden lea ning. Vázquez-Cano e al. (2021) de eloped a cha bo o imp o e Spanish punc ua ion skills, enhancing
lexible lea ning en i onmen s. Kadhim and Hassan (2020) p oposed a hyb id AI model o p edic lea ning con inui y
in highe educa ion. Salas-Rueda e al. (2020) demons a ed he impac o web applica ions using da a science o
eaching compound in e es . Jokhan e al. (2022) analyzed AI’s ole in digi al esou ce consump ion and decision-
making ega ding s uden pe o mance. Ha a i e al. (2021) e alua ed he adap i e ALEKS sys em’s impac on sel -
egula ed lea ning. Addi ionally, G unhu e al. (2022) and Gilson e al. (2023) s udied AI applica ions in medical
educa ion, pa icula ly he po en ial o la ge language models like Cha GPT in knowledge assessmen . Nuankaew
(2022) de eloped a sel - egula ed lea ning model based on educa ional da a mining. Sha ma e al. (2019) p oposed AI
and mul imodal analy ics pipelines, while Ruipé ez-Valien e e al. (2019) applied ML o de ec chea ing beha io s in
MOOCs.
3) Ba ie s and challenges o AI adop ion in educa ion (9 s udies):
The inal g oup o s udies ocused on iden i ying obs acles o e ec i e AI in eg a ion in educa ion. Singh and Hi an
(2022) emphasized he digi al di ide and lack o educa o eadiness as signi ican ba ie s. D uzhinina e al. (2021)
explo ed he complexi y o expe AI sys ems in ma hema ics lea ning en i onmen s. Pu e al. (2021) conduc ed a
bibliome ic analysis e ealing geog aphical and educa ional le el dispa i ies in AI esea ch co e age. Kule o e al.
(2021) examined challenges ela ed o AI and ML implemen a ion in highe educa ion ins i u ions. Biu un (2023)
highligh ed socie al-le el conce ns, such as na ional es ic ions on ools like Cha GPT. Simila ly, Mu phy-Kelly (2023)
discussed e hical isks and global calls o cau ion in AI de elopmen . Nicole i and de Oli ei a (2020) p oposed ML-
based models o d opou p edic ion, unde lining he need o u he esea ch on equi y and accessibili y. Lampos e
al. (2021) analyzed AI’s po en ial o suppo eache s in au ism educa ion, iden i ying he need o be e in eg a ion
s a egies. Finally, C uz-Jesus e al. (2020) add essed he use o AI o assess academic achie emen , s essing he
impo ance o conside ing con ex ual ba ie s.