E hics o AI and
Da a in
Educa ion
The Si ua ed ETH-TECH
Pe spec i e: a F amewo k o
P ac ice
P ojec Deli e able –Wo k Package II «ETH-TECH F amewo k»
The P ojec
& Pa ne ship
Ancho ing E hical Technology (AI and Da a)
Usage in he Educa ional p ac ice (ETH-TECH)
The ETH-TECH p ojec aims o p omo e an e hical
app oach o he use o echnologies, especially
da a and a i icial in elligence (AI), in highe
educa ion (HE) eaching and lea ning. I is
g ounded on Eu opean Union ini ia i es, such as
he e hical guidelines on he use o AI and da a in
educa ion published in 2022, which encou aged
educa o s o e lec on he consequences o he
digi al ans o ma ion.
ETH-TECH channels he complex p o essional
expe ise o uni e si y educa o s and
esea che s om 4 EU coun ies, o p o ide a
cul u ally sensi i e and con ex ualized
amewo k o he e hical use o AI in HE.
This p ojec deli e able is he esul o join collabo a ion be ween all pa ne s.
Final documen au ho ed by BBU and UNIPD eams (Wo k Package Leade s). 2
Le ’s make
a s a !
ETH-TECH objec i es
To aise awa eness among u u e educa o s
abou ETH-TECH pe spec i es in HE cou ses.
To de elop a se o p ac ical ools o sel -
assessmen o he e ec i e in eg a ion o an ETH-
TECH app oach in HE, based on EU guidelines.
To suppo he de elopmen o an ETH-TECH
app oach h ough a se o OER.
To os e c ea i e engagemen o key
s akeholde s owa ds he ETH-TECH.
ETH-TECH ac i i ies
Awa eness- aising sessions, a e he
pa icipa o y analysis o syllabi and eaching
p ac ices on AI e hics and da a use in educa ion.
C ea ion o a sel -assessmen amewo k and
ools o p omo e dialogue on cu en gaps and
u u e challenges o he ETH-TECH app oach.
De elopmen o Open Educa ional Resou ces
and a 25-hou cou se (1 ECTS) o acili a e he
ETH-TECH app oach in ins i u ions.
O ganiza ion o Educa hon e en s o alida e he
ETH-TECH app oach and esou ces wi h NGOs,
SMEs, and ci il socie y.
Unde s anding
he ETH-TECH
objec i es and
ac i i ies will help us
unde s and he alue
o he p esen
amewo k
3
The aim o his amewo k is o p o ide suppo o
u u e educa o s (and eache s and educa o s’
aine s) o e lec abou he e hics o AI and da a in
educa ion*. ETH-TECH c i ically in eg a es he 7
p inciples on he use o AI and da a in educa ion
p oposed by he Eu opean Commission in 2022.
Mo eo e , we conside he ecen de elopmen s
ela ed o he AI Ac (Eu opean Commission, 2024)
o Highe Educa ion.
P inciple Sho explana ion and key guiding ques ion
Human Agency
and O e sigh
AI can help eaching and lea ning goals and collabo a i e wo k
wi h colleagues. I is key ha educa o s ha e con ol and
o e sigh o he AI
-suppo ed p oduc s being de eloped.
Who con ols and moni o s he AI you use o educa ion?
T anspa ency
AI sys ems need o clea ly explain how hey unc ion and wha
da a hey p ocess.
Do you know how he AI you use o educa ion wo ks?
Di e si y, non
-
Disc imina ion
and Fai ness
AI sys ems should be designed o accommoda e he di e si y
and needs o all eache s and s uden s, who should be able o
use hese sys ems equally.
Could he AI sys em lead o disc imina ion o un ai ness
owa ds some use s?
Socie al and
En i onmen al
Wellbeing
The inc easing use o AI ools o educa ional ac i i ies should
no ha e a nega i e impac on b oade e hical conce ns, such
as oxic dependence, lack o academic in eg i y o
en i onmen al impac .
How does he AI you use o educa ion make you eel?
P i acy and
Da a
Go e nance
In he “
da a ied” socie y use in e ac ions wi h AI sys ems a e
eco ded and po en ially mone ized. I is impo an o
unde s and how pe sonal da a is collec ed, s o ed, and used
by AI sys ems.
How is you da a collec ed when you use AI o educa ion
and how is i used? Who has access o his da a?
Technical
Robus ness and
Sa e y
Many AI sys ems ope a e as “black boxes,” which means ha
hei algo i hms wo k in unclea ways, o he in e aces do no
explain how da a is handled. You depend on sys ems you don’
ully unde s and, which can unde mine us and pedagogical
alignmen .
Is you in e ac ion wi h AI o educa ion secu e and you can
ace each s ep o he in e ac ion?
Accoun abili y
Teache s and uni e si ies need o unde s and and moni o how
Ed
-Tech enhanced by AI wo ks, o be in con ac wi h hei
de elope s o oubleshoo ing, and o ackle po en ial
p oblems.
Who can you con ac when some hing goes w ong wi h you
using AI o educa ion?
Table 1. Syn he ic p esen a ion o he 7 p inciples on he use o AI and da a in educa ion
(Eu opean Commission, 2022)
Take you
ime...
And explo e he
se e al p inciples.
Use he ques ions in
he able o hink
abou he AI-
powe ed sys ems in
use in he ins i u ion
whe e you ake pa
as a eache o a
s uden .
Du ing an ini ial analysis ca ied ou o e mo e han 300 syllabi ac oss ou membe s a es (Ge many,
I aly, Romania, Spain) we ound a oubled pic u e o how he E hics o AI and da a is augh o u u e
educa o s. Read mo e HERE.
4
Ce ainly,
hese
p inciples
a e e y
abs ac
and non-
con ex ual.
The ETH-TECH amewo k
ecommends o c i ically
in eg a e EU egula ions
on e hical use o AI.
He e, you will see se e al
hings o e lec upon
be o e conside ing he
e hical p inciples.
5
Analyse
he ole o
cul u e
Each local con ex has a dominan cul u al
o ien a ion ( o example, collec i ism e sus
indi idualism), which is e lec ed on how
legisla o s, ins i u ions, and educa o s app oach
AI o educa ion.
These cul u al o ien a ions place mo e o less
esponsibili y on he indi idual (adminis a o ,
eache , s uden ) o he legisla o s (na ional o
local adminis a i e ins i u ions) o de ine, en o ce,
and suppo e hical AI use in educa ion.
The us ha people ha e in he sys em is he
i s line o analysis ha needs o be
conside ed.
Guiding e lec ion ques ions
Who egula es he e hical use o AI in
educa ion in you coun y?
How much lexibili y does you uni e si y ha e
in en o cing e hical use o AI in educa ion?
Do you eel p o ec ed ye esponsible when
you expe ience a b each in e hical use o AI in
educa ion in you p ac ice?
6
Conside
he
na ional
and local
socio-
economic
dynamic
Local (embedded in na ional) con ex s may ha e
mo e o less esou ces in educa ing HE
ins i u ions abou he e hical use o AI in
educa ion.
This c ea es inequali ies a na ional and egional
le el and equi es a ailo ed app oach o wha is
possible o be implemen ed in e ms o
p ocedu al egula ions and aining o eache s.
Guiding e lec ion ques ions
Who p o ides esou ces (in o ma ion,
aining, assis ance o c i ical p oblems) o
e hical use o AI in you egions?
Is he e hical use o AI o educa ion legally
egula ed and a e he e esou ces o ensu e
he implemen a ion o hese egula ions?
Do eache s and s uden s ha e he socio-
economic esou ces ( ime, educa ion abou
how AI wo ks, ools and skills o c i ical
e lec ion) o ensu e e hical use o AI in hei
educa ional p ac ice?
7
Unde s and
he
s uc u e
o he
speci ic
educa ional
sys em
Each educa ional sys em wo ks di e en ly,
s a ing om how HE is s uc u ed o who can
access HE. Mo eo e , he au onomy HE
ins i u ions ha e di e s among EU coun ies.
These s uc u al eali ies g ea ly shape WHAT and
WHEN can be done o implemen e hical use o AI
in educa ion.
Guiding e lec ion ques ions
Does you uni e si y o you na ional Minis y
o Educa ion decide on he e hical
guidelines o AI in educa ion?
Do eache s and s uden s in you uni e si y
bene i om aining on he e hical use o AI
in educa ion?
Can you au onomously in eg a e e hical
guidelines o AI use in educa ion in you
cou se syllabi?
Who can you con ac when he e is an
e hical b each o AI use in educa ion?
8
The ETH-TECH eam conduc ed Awa eness Raising
Sessions wi h s uden s and eache s in he 4 p ojec
coun ies*. These sessions unco e ed he
mul ilaye ed dynamic o e hical use o AI in educa ion.
Pa icipan s in he Awa eness Raising sessions
concep ualized h ee hie a chical le els o e hics in
p ac ice:
1. No ma i e/Technological;
2. Ins i u ional;
3. Pe sonal ( eache , s uden , class oom as
egula ed in e ac ion o indi iduals).
This dynamic is syn he ized in Figu e 1 and de ailed
below.
*The coun y-le el epo s de ailing he sessions and na ional conclusions a e a ailable online:
Ge many (HSU eam) - I aly (UNIPD eam) - Rumania (BBU eam) - Spain (UB eam)
No ma i e/
Technological
Le el
Ins i u ional
Le el
Pe sonal
Le el
•In e na ional egula ions.
•Na ional and ins i u ional
in o ma ion a ailable.
•Ins i u ional in o ma ion
a ailable.
•Ins i u ional ech ools.
•Ins i u ional egula ions
and codes o p ac ice.
•Teaching p ac ice and
p o essionalism.
•Academic in eg i y.
•Pe sonal posi ionali y.
Le els o
unde s and
he e hics
o AI and
da a in he
educa ional
p ac ice
9
Ed-Tech enhanced by AI should help eache s and
s uden s each hei eaching and lea ning goals
and wo k wi h colleagues o c ea e be e
academic and educa ional wo k. A each poin , i is
key ha eache s ha e con ol and o e sigh o he
AI-suppo ed p oduc s so hey can in e ene in
cases o e o s, misin o ma ion, disc imina ion and
s uden o e eliance on he sys ems.
Human
Agency
and
O e sigh
F iendly
De ini ion
In a so wa e ce i ica ion cou se, he e is a
e y ac i e g oup o pa icipan s who p omo e
in o mal suppo o he s udy. In his ega d,
hey ha e opened a Wha sApp channel o
suppo each o he in hei lea ning e o s.
Wi hin his g oup, i eme ges ha he use o AI
ools such as Claude o Copilo is pe ec o
w i ing a p og amming assignmen equi ed in
one o he eachings. The eache is no awa e
o he ool and does no ha e access o
e icien ools o de ec ing AI-gene a ed
con en , as hey ha e no been de eloped ye .
The e o e, many s uden s c ea e hei en i e
assignmen wi h AI. Despi e some su p ise a
he unusually high wo k quali y o his
gene a ion o s uden s, he eache does no
wo y much: he mo e pa icipan s a e
ce i ied, he highe he success a e o he
cou se, he be e he emune a ion. The use o
AI in s uden s' wo k is no discussed du ing he
cou se, and s uden s begin o use AI as a
sho cu o comple e hei assignmen s a he
han a ool ha can assis hem in hei
lea ning.
Case
S udy
Guiding Ques ions
Can you ela e o his si ua ion? In you local educa ional con ex , do you hink
s uden s can use AI as a ool ha impai s hei lea ning p ocess, despi e
seemingly leading o good esul s?
16
Human
Agency
and
O e sigh
Ac ion Poin s: Ins i u ion
P oblem: Teache s ask he uni e si y leade ship o cla i y he ole
o genAI in EdTech p oduc s ha a e o mally used by he
uni e si y o acili a e eaching and s uden lea ning. They s ess
ha mo e genAI has been in oduced in he las e sions o hese
p oduc s and ea ha s uden decisions a e now con olled by AI.
Ac ion poin : Requi e human o e sigh o au oma ed decisions.
AI-guided decisions o di e en ea u es need o be anspa en
o he eache s who can hen acili a e o es ic s uden access
o hese ea u es. Especially when AI ea u es e lec con en
c ea ion o s uden g ading, eache s need o be able o make
decisions ega ding ea u e ac i a ion o cus omiza ion.
P oblem: Teache s and s uden s complain ha hey don’ ully
unde s and he limi a ions o genAI use o educa ion. The o me
ask o ins i u ional egula ions and he la e claim ha hey can
decide hemsel es how o use AI o educa ion, as he e a e no
clea ules.
Ac ion poin : Conside ing na ional and uni e si y egula ions
ega ding he use o AI o educa ion, c ea e an e hical code and
implemen a ion ules (in e ms o ac ions o speci ic issues) ha
all eache s mus adhe e o and in eg a e in o hei syllabi and all
s uden s mus o mally acknowledge a he beginning o he
academic yea .
P oblem: The e hics commi ee o you uni e si y aces new and
mo e complex e hical complain s b ough by eache s and
s uden s ega ding he audulen use o genAI and canno keep
up wi h upda ing he e hical code o he uni e si y o hese issues.
Ac ion poin : Make a da abase o ypes o p oblems and
complain s on he une hical use o AI and genAI in educa ion, o
g adually c ea e ca ego ies o p oblems. By making legal ad iso s,
ech specialis s, eache s, and s uden s awa e o ypes o
p oblems, a communi y ocusing on anspa en p oblem-sol ing
can be de eloped. This can be a p ocedu al manne o ensu e
anspa en and pa icipa o y decision-making o AI- ela ed
issues in educa ion.
17
Human
Agency
and
O e sigh
Ac ion Poin s: Teache
P oblem: The lea ning communi y o a cou se has
been g ea ly a ec ed by he ambi alence o AI use o
educa ion, which some s uden s pe cei e as chea ing
and o he s see as a na u al enhancemen o hei
academic lea ning p ocess.
Ac ion poin : Ensu e meaning ul consen o AI use in
educa ion and o e al e na i es wi hou penal y, by
co-c ea ing "didac ic con ac s" a he cou se ou se .
P oblem: The assignmen s s uden s u n in ha e
many ake e e ences (including o you own
publica ions) and use a complex and somehow
unna u al ype o ph asing. You suspec ha hey ha e
been gene a ed wi h he use o genAI.
Ac ion poin : Shi owa ds p ocess-o ien ed s uden
assessmen , ha elies on he e-and-now ac ions and
is no in luenced by gen AI. O al e alua ions and
p ojec wo k ha occu s du ing semina s as “in-
baske ” o si ua ional es s can be e cap u e
s uden s’ compe ences.
P oblem: S uden s don’ unde s and how he
excessi e use o genAI o educa ion can nega i ely
impac hei c i ical hinking.
Ac ion poin : Teach s uden s o c i ically e lec on
con en s gene a ed by genAI o ques ions you use in
you own cou se. You can in oduce he de elopmen
o c i ical e lec ion on genAI use o educa ion as a
ans e sal skill ha you cou se de elops.
18
Human
Agency
and
O e sigh
Ac ion Poin s: S uden
P oblem: Teache s do no clea ly s a e in he cou se
syllabi i and which genAI you can use o academic
lea ning.
Ac ion poin : Openly discuss wi h he p o esso ia e
as he beginning o he academic yea which is he
o icial posi ion o he uni e si y and each eache ’s
speci ic cou se ega ding he use o AI o academic
lea ning.
P oblem: All you colleagues ha e been using genAI
o mos academic wo k and you eel you also mus
use i o emain compe i i e.
Ac ion poin : C i ically analyze wha “compe i i e”
means o a speci ic cou se. Using genAI o
gene a ing academic wo k does no enhance you
knowledge and compe ences and o en c ea es jus
an illusion o compe ency.
P oblem: The answe s ha he genAI con e sa ional
agen has been gi ing you o a ques ion ha you
eache app oached in a cou se a e e y di e en om
wha he eache p esen s du ing he cou se.
Ac ion poin : C i ically analyze each answe and
discuss wi h he eache he di e ences you
iden i ied. Fac checking and ca e ul eading o
e e ences used by he AI and he eache a e e y
impo an in unde s anding why he answe s a e
di e en .
19
AI sys ems need o clea ly explain how hey
unc ion, wha da a hey collec and o wha
pu poses. S uden s, eache s and
uni e si ies should be in o med abou hese
aspec s so hey can gi e hei in o med
consen when using AI sys ems.
F iendly
De ini ion
A uni e si y in oduces an AI-powe ed
so wa e ha assis s s uden s in hei
lea ning. I wo ks as a i ual assis an
which gi es s uden s de ailed ins uc ions
and eedback on hei asks bu also
includes emo ional suppo o help
s uden s manage hei men al heal h
du ing imes o academic s ess. Bo h
s uden s and eache s a e happy o use
his ee sys em: s uden s app ecia e he
immedia e and pe sonalized assis ance,
while eache s app ecia e he educed
wo kload. Howe e , some s uden s no ice
ha hey began ecei ing ads o paid s udy
ma e ials, online cou ses and u o ing
se ices. Some o hem also ecei ed ads
o men al heal h se ices, and apps
a ge ed on issues like hose discussed
wi h he so wa e’s cha bo . O e ime,
s uden s and eache s become su e ha
he da a is sha ed wi h hi d pa ies and
used o comme cial p o iling.
Case
S udy
Guiding Ques ions
Do you know wha da a is collec ed du ing you in e ac ions wi h AI sys ems you use in you
educa ional con ex and how i is la e used?
Would knowing you da a is sha ed o comme cial p o iling in luence i and how you (as s uden ,
eache o ins i u ional educa ional s a ) use AI sys ems?
T anspa ency
20
Ac ion Poin s: Ins i u ion
T anspa ency
P oblem: Teache s and s uden s poin ou ha he AI
sys ems o icially used by he uni e si y a e linked o
comme cial pla o ms ha o e pe sonalized
solu ions based on class wo k.
Ac ion poin : P omo e anspa ency abou how da a
and algo i hms wo k. Reques he AI sys em p o ide
o p o ide and explain how da a collec ed om
uni e si y use s a e used and sha ed.
P oblem: Adminis a o s and eache s poin ha
exis ing uni e si y-le el a e no a ailable o ou da ed
o deal wi h genAI e hical issues in HE.
Ac ion poin : Make sha ed decisions abou ool
usage and cu iculum-in eg a ed e hics as
app oaches aligned wi h pa icipa o y pedagogies and
democ a ic educa ion.
P oblem: Though he uni e si y p o ides an e hical
code o AI use o e e ences he EU-le el egula ions,
eache s and s uden s do no unde s and hese
egula ions and canno link hem o hei eaching and
lea ning p ac ice.
Ac ion poin : P o ide aining o bo h eache s and
s uden s on e hical AI use, using co-design scena ios
ha deal wi h lea ne s and communi y p oblems.
21
Ac ion Poin s: Teache
T anspa ency
P oblem: In eg a ion o AI-gene a ed ma e ials in he
cou se can be use ul as p ac ical examples o many
subjec s. Ne e heless, when he eache gene a es
ma e ial using complex p omp s o a genAI agen , i
is no clea who owns he p oduc and how i will be
used by he AI de elope company in he u u e.
Ac ion poin : When gi en he possibili y, always op
o an open-sou ce solu ion o gene a e new cou se
ma e ial o a genAI solu ion ha is anspa en abou
da a s o age and sha ing.
P oblem: As eache s use mul iple AI and genAI ools
in hei eaching p ac ice, i may be di icul o keep
ack which ools a e anspa en on hei da a sha ing
p ac ices.
Ac ion poin : Take ime a he beginning o each yea
o choose he AI ools you will use o educa ion. Then
ga he in o ma ion on how he in o ma ion you inpu
in o AI sys ems o educa ion is being s o ed, sha ed,
and who owns he p oduc s you de elop wi h he help
o AI.
P oblem: I is no clea how GenAI powe ed lea ning
ools e alua e s uden knowledge and compe ence
acquisi ion, which makes i ha d o he eache o
accu a ely assess how much s uden s ha e lea n .
Ac ion poin : Become in o med how he AI sys ems
you use o s uden lea ning ga he in o ma ion o
p o ide assessmen s o s uden lea ning p og ess.
22
Ac ion Poin s: S uden
T anspa ency
P oblem: genAI is easy o use and o e s quick
answe s s a ing om e y simple ques ions. Bu
s uden s become inc easingly awa e ha hei
choices when using AI o educa ion a e ans e ed o
o he pla o ms ha may no be linked o educa ion.
Ac ion poin : Unde s and wha he “digi al oo p in ”
means. I p i a e companies a e behind AI ools, da a
will be collec ed o comme cial pu poses. The e a e
mul iple nega i e e hical implica ions o educa ional
con en s and educa ional decisions ha a e
mone ized wi hou he awa eness and ac i e consen
o he use .
P oblem: I is di icul o unde s and how AI and
genAI wo k and whe e o sea ch o easy- o-
unde s and in o ma ion on how hese sys ems
unc ion, who is esponsible o hei mechanics, and
how da a is being used.
Ac ion poin : Read and wa ch p esen a ions
de eloped by legisla o s ha y o egula e AI o
educa ion. He e a e p esen a ions o he AI Ac
de eloped by he Eu opean Commission (2024):
h ps://a i icialin elligenceac .eu/ *
Take in o accoun ha he complexi y o he
echnological sys ems behind genAI canno be ully
g asped om a sho p esen a ion..
Websi e main ained by he Fu u e o Li e Ins i u e (FLI). FLI is an independen non-p o i wo king o educe la ge-scale,
ex eme isks om ans o ma i e echnologies. h ps://a i icialin elligenceac .eu/abou /
23
All eache s and s uden s should be
able o access he AI (o Ed-Tech
enhanced by AI) in he same manne
and he AI sys em should be
designed o accommoda e o he
di e si y o all s uden s, including
hose wi h special needs.
AI sys ems should no acili a e
disc imina ion o o he inequi able
p ac ices.
F iendly
De ini ion
A p o esso a a mul icul u al uni e si y
c ea ed a p esen a ion o he uni e si y
o p ospec i e s uden s.
Dall-E (an image gene a ion sys em) and
Can a ( eemium e sions) a e used o
gene a e he p esen a ion. In c ea ing
some o he images, he p o esso
ealizes ha all he images o scien is s
gene a ed by AI include middle-aged
men, usually Caucasian and shown in a
cen al posi ion.
When he p omp s a e changed o ask
o emale and disabled scien is s, hey
a e usually p esen ed in a suppo ing
ole.
Case
S udy
Guiding Ques ions
Do you hink AI can ein o ce p e-exis ing s e eo ypes and biases in you con ex ?
Do you hink ha in you uni e si y/educa ional ins i u ion AI is equally accessible o all s uden s,
ega dless o backg ound and possible special needs?
Di e si y,
non-
Disc imina ion
and Fai ness
24
P oblem: AI and genAI educa ional p oduc s
implici ly p omo e s e eo ypical p esen a ion o
lea ne s and expec ed lea ning ou comes. Gende
s e eo ypes, men al heal h s e eo ypes, pe o mance
s e eo ypes, educa ional achie emen expec a ions
a e embedded in images, ypes o asks, and
e alua ion op ions.
Ac ion poin : Choose an AI ools o educa ion ha is
cus omizable o he cha ac e is ics o he s uden
lea ne s in you uni e si y and ha o e eache s he
possibili y o cus omiza ion o a speci ic cou se.
P oblem: Many AI solu ions o educa ion do no
accommoda e he lea ning needs and lea ning pace
o s uden s wi h physical and/o cogni i e
disabili ies. They p omo e he idea ha e e yone can
become compe en and can each he same high
le els o pe o mance.
Ac ion poin : P o ide eache s echnical suppo o
ailo AI ools ha can be adop ed by di e se
s uden s. Technical s a needs o ha e ope a ional
compe ences in choosing he mos app op ia e AI
ools o educa ion o speci ic ypes o s uden
disabili ies. Re use AI ools ha does no align wi h
di e si y-
P oblem: AI ins umen s o educa ion do no
acknowledge he di e si y o lea ne needs and do
no p omo e he ole o accep ance o di e si y om
eache s and s uden s.
Ac ion poin : Focus on o e ing eache s aining
p og ams ha help hem ecognize and wo k wi h
s uden di e si y by os e ing inclusion and
accep ance.
Di e si y,
non-
Disc imina ion
and Fai ness
Ac ion Poin s: Ins i u ion
25
