GEF Madrid 2024: Ethical implications of AI
Summary
Please replace the link and try again.
Takeaways
- 🌐 La sesión del Foro Global de Educación se centra en las implicaciones éticas de la IA, destacando la importancia de abordar cuestiones morales y de justicia en el desarrollo de la tecnología.
- 👨🏫 Gregory Enu, profesor de la Universidad de Mason y editor de la MIT Sloan Management Review, comparte su experiencia con la IA desde su introducción en el aula, y cómo la IA puede ser usada para avanzar en educación pero también plantea retos éticos significativos.
- 🤖 La IA y los modelos grandes tienen el potencial de transformar la educación, pero también presentan desafíos éticos y otros problemas que deben ser considerados cuidadosamente.
- 📊 Se destaca la falta de investigación académica en las implicaciones éticas de la IA, señalando una deficiencia importante en el ámbito educativo.
- 🔧 Hgo, jefe de IA en la Asociación Españolola de la Economía Digital, enfatiza la importancia de la transparencia y explicabilidad en los sistemas de IA para construir una IA responsable.
- 🏛️ Se discute la necesidad de trabajar con empresas y reguladores para comprender plenamente los términos de transparencia y explicabilidad, lo que incluye la conformidad con regulaciones y la comunicación con partes interesadas.
- 🎨 Ahmed Gal, de la Universidad R, explora la intersección del arte y la IA, y cómo la IA puede afectar la identidad del creador, planteando cuestiones éticas sobre la autoría y la originalidad en el arte.
- 📚 Se argumenta que la educación y la IA pueden ser utilizadas para abordar las desigualdades globales y promover la equidad en el acceso a la educación de calidad.
- 🌍 Se cuestiona el papel de la IA en la educación global y cómo puede ayudar a reducir las desigualdades geográficas y sociales, planteando preguntas éticas sobre su implementación y efecto en la equidad.
- 🤝 Se destaca la importancia de la colaboración entre el sector público y privado para el desarrollo de modelos de lenguaje en español y otras lenguas cooficiales en España y América Latina, sin dejar de lado la importancia de la calidad y la falta de sesgo en los datos.
- 🧐 Se abordan los retos de la IA en la educación, como la generación de datos no sesgados y la necesidad de comprender y abordar las desigualdades preexistentes en el sistema educativo antes de la implementación de la IA.
Q & A
¿Quién es Gregory Enu y qué papel desempeña en el panel sobre ética en la IA?
-Gregory Enu es el Arison profesor en la Universidad George Mason y guest editor de sostenibilidad en la MIT Sloan Management Review. Actúa como moderador del panel y experto en ética y implicaciones éticas en la IA.
¿Cuál es la preocupación principal de Car antias según el panel de la Comisión de Asesoramiento en IA de la ONU?
-Car antias resalta que las cuestiones que debemos enfrentar no son solo políticas, legales y de política pública, sino también cuestiones morales, es decir, cuestiones de correcto o incorrecto, bueno o malo.
¿Qué deficiencia en la investigación académica señala George Seamons en relación con la IA?
-George Seamons señala que la principal deficiencia en la investigación académica es el estudio de las implicaciones éticas de la IA.
¿Cómo describió Gregory Enu su experiencia con Chat GPT en el aula de alfabetización digital?
-Gregory Enu describió su experiencia con Chat GPT como reveladora del potencial vasto que la inteligencia artificial y los modelos grandes pueden aportar a la educación, así como de los desafíos éticos y otros que esto conlleva.
¿Qué papel desempeña hgo como Chief AI Officer en la Asociación Españolola de la Economía Digital?
-Hgo lidera un equipo que trabaja junto con las empresas miembros para entender las complejidades de la IA y las tecnologías emergentes, promoviendo la transparencia y la explicabilidad en los sistemas.
¿Qué es la preocupación principal de hgo en términos de ética en la IA según el panel?
-La preocupación principal de hgo es la transparencia y explicabilidad de los sistemas, lo que incluye cómo se comprenden los propios modelos que se construyen o se utilizan, y cómo se comparte esta información con los interesados.
¿Qué rol desempeña la identidad en la creación artística y cómo la afecta la IA según Ahmed Gal?
-Según Ahmed Gal, la identidad es fundamental para los creadores en el ámbito artístico y la música. La IA introduce problemas de autoría y originalidad, ya que la aleatoriedad en los sistemas de IA puede generar resultados diferentes para prompts similares.
¿Qué es la preocupación ética que Mark Kaban destaca en relación con los programas de educación STEM financiados por fabricantes de armas?
-Mark Kaban destaca la preocupación ética de la influencia de fabricantes de armas en los programas de educación STEM en escuelas K12, cuestionando si esto puede ser una forma de inducción o grooming de jóvenes para convertirse en futuros fabricantes de armas.
¿Cómo considera Mark Kaban que la tecnología de la IA pueda afectar la equidad educativa global?
-Mark Kaban cuestiona el papel de la IA en la educación y cómo puede abordar las desigualdades geográficas y sociales a nivel mundial, planteando si la educación en IA será liderada por corporaciones o por instituciones públicas.
¿Cuál es la reflexión final de Gregory Enu sobre el papel de la IA en el futuro de la educación?
-Gregory Enu reflexiona sobre cómo la IA puede cambiar nuestra comprensión del papel de la educación, destacando la necesidad de enseñar habilidades como la motivación, la perseverancia y el pensamiento crítico, ya que el conocimiento ya está accesible.
Outlines
😀 Introducción a la Ética en la IA
Gregory Enu, profesor de la Universidad George Mason y editor de MIT Sloan, inicia un panel sobre ética y consecuencias éticas en la IA. Destaca la importancia de abordar cuestiones morales en la tecnología y menciona el déficit en la investigación académica sobre este tema. Comparte su experiencia con Chat GPT en 2002 y la revelación de su potencial en la educación, así como los desafíos éticos que presenta.
🌐 Transparencia y Explicabilidad en la IA
Hugo, el Chief AI Officer de Adal, la Asociación Españolola de la Economía Digital, enfatiza la necesidad de transparencia y explicabilidad en los sistemas de IA para garantizar su ética y sostenibilidad. Discute la complejidad de los modelos de IA y la importancia de convertir su funcionamiento en algo comprensible para los usuarios y reguladores, para cumplir con las regulaciones y expectations de la sociedad.
🎨 Identidad y Ética en el Arte y la IA
Ahmed Gal, de la Universidad R, destaca su trayectoria en AI y su enfoque en el arte y la IA. Plantea la cuestión de la identidad y la autoría en la creación artística con la ayuda de la IA, señalando los problemas éticos que surgen cuando la IA genera resultados aleatorios que pueden ser utilizados por otros creadores, desafiando la noción de originalidad y derechos de autor.
📚 Ética en la Educación y la Industria de Armas
Mark Kaban, con experiencia en psicología y liderazgo educativo, expone su preocupación por la influencia de la industria de armas en los programas de STEM en escuelas K12, financiados por corporaciones como Lockheed Martin. Cuestiona la ética de vincular las misiones educativas con la producción de armas y el posible impacto en los estudiantes, especialmente en el contexto de universidades y la resistencia a la guerra.
🌍 Desafíos Éticos en la Educación Global con AI
Se discuten los retos éticos de implementar la IA en la educación a nivel global, enfocándose en la reducción de inequidades y el cumplimiento de los Objetivos de Desarrollo Sostenible. Se plantea la cuestión de si la IA puede ayudar a superar las brechas geográficas y sociales, y cómo abordar la equidad en el acceso y la calidad de la educación con tecnologías emergentes.
🤖 Aprendizaje Basado en Curiosidad con AI
Se explora cómo la IA puede fomentar un aprendizaje basado en la curiosidad y la resolución de problemas, contrastando con el enfoque tradicional basado en currículos. Se destacan las limitaciones actuales de la IA, como la 'alucinación' y las sesgos en los datos y los modelos, y se sugiere que la educación debe adaptarse a estas nuevas formas de aprendizaje potenciadas por la tecnología.
📘 El Cierre del Parentésis de Gutenberg y el Futuro de la Educación
Se reflexiona sobre el impacto de la imprenta en el aprendizaje y cómo la tecnología, incluida la IA, podría estar cerrando el 'Parenthesis de Gutenberg', cambiando la forma en que entendemos y modelamos nuestro mundo. Se sugiere que la IA puede cambiar nuestra percepción del aprendizaje, posiblemente de regreso a métodos más interactivos y basados en la discusión.
🧐 Desafíos de la Personalización y Asistencia Virtual en la Educación
Se debaten los posibles desafíos y oportunidades de la personalización en la educación mediante asistentes virtuales y AI. Se mencionan cuestiones de privacidad, la relación que los estudiantes podrían desarrollar con estas herramientas y las implicaciones éticas de estas interacciones, como la creación de vínculos personales con la tecnología educativa.
🤝 El Papel del Educador en el Futuro con AI
Se reconoce la importancia del rol del educador en el futuro, incluso con la presencia de la IA, para motivar y enseñar habilidades como la perseverancia y el pensamiento crítico. Se destaca que la IA no resuelve por sí sola la motivación para aprender, y es crucial la intervención del educador para guiar a los estudiantes en su camino educativo.
🔮 Consideraciones Finales y Desafíos Éticos Futuros
En conclusión, se destacan las consideraciones finales sobre cómo la IA y la tecnología digital son comercializadas y difundidas en la sociedad, a menudo sin prever las consecuencias éticas y morales a largo plazo. Se sugiere que el debate sobre la ética en la IA es un tema que debe continuar en el futuro, adaptándose a los cambios y desafíos que surjan con el tiempo.
Mindmap
Keywords
Please replace the link and try again.
Highlights
Gregory enu introduces the Global Educational Forum session on ethics in AI, emphasizing the importance of moral considerations in technology.
Car antias from the UN AI Advisory Commission stresses that ethical questions in AI are not just political or legal, but fundamentally moral.
Academic research is identified as lacking in the area of AI's ethical implications, according to George Seamons.
Gregory enu shares his experience with AI in education, highlighting the potential and challenges of AI in teaching.
Hgo discusses the importance of transparency and explainability in AI systems for ethical and regulatory compliance.
Ahmed gal from R University emphasizes the ethical issue of identity in AI, particularly for creatives using AI in their work.
Mark kaban raises concerns about the influence of arms manufacturers in STEM education and the ethical implications of such partnerships.
The panel explores the potential of AI to democratize education by overcoming geographic and social divides.
Bias in AI is discussed, including cultural bias and the importance of considering local contexts in AI applications.
The role of AI in global education is examined, with a focus on addressing inequities and promoting quality education.
The potential for AI to enable personalized learning experiences is highlighted, with ethical considerations regarding data privacy.
The ethical challenges of AI 'hallucinations' are discussed, where AI provides incorrect or made-up information.
The importance of critical thinking and discernment in the era of AI is underscored, especially for students in less-resourced schools.
A counterpoint is made about the positive use of AI for motivating autistic children, showing the potential for tailored educational support.
The final discussion centers on the role of educators in fostering motivation and critical thinking, which AI cannot replace.
The session concludes with a reflection on the ethical considerations of AI in education and the need for ongoing dialogue.
Transcripts
okay all right welcome everyone this is
the obviously the the session here in in
the uh Global educational
Forum my name is Gregory enu I'm uh the
Arison professor at George Mason
University I'm also the sustainability
guest editor at the MIT Sloan management
review um and the topic of this panel is
ethics and ethical implications in AI
we've seen in the plenary sessions quite
a bit of discussion about ethics
um car antias from the AI advisory
Commission of the UN highlighted that
the tech the questions we need to face
are not only political legal and policy
but also Moral Moral questions are
questions of right or wrong good or bad
and those are the considerations we're
charged with here we also found out from
George seamons that uh the acade
Academia is doing a poor job the number
one deficit in academic research is the
ethical implic ations of AI so um we're
talking about a very very important
topic today um I have an interesting
sort of background I'll share uh quickly
with uh around AI technology and
education uh in 20 2002 I was asked to
revamp our digital literacy course for
our undergraduate students for the 20
year-old students and so we were
designing a course that could teach uh
students to use digital Technologies to
advance their personal and professional
goals use technology and not be be used
by technology and in November of that
year in the middle of the class chat GPT
was launched and so as a professor of
the class on digital digital literacy I
logged in got my account and typed in my
the first thing I ever typed in is how
should I teach chat gbtb to my students
and within seconds I got a very well
reasoned out lesson plan and with just a
couple of touches of the keyboard I
recognized immediately the vast
potential that artificial intelligence
and large m models could bring to
education and then all the vast
potential challenges ethical and
otherwides that it can bring so that's
what we'll tap into here today and we
have uh a series of very uh qualified
and expert panelists to discuss this
area that we're still discovering uh the
some ethical concerns we are aware of
many were not aware of so what I would
like to ask is we'll go around uh the
panel here ask each of you to introduce
yourself tell us a little bit about your
background and then just select one
ethical issue that you think the
participants of this uh this forum
should be considering and uh being
perhaps thinking about responses or
Solutions too so please hello hello
everyone hello
everyone so my name is hgo I'm the chief
AI officer at adal adigital is the
Spanish Association of the digital
economy so we comprise around a bit more
than 500 companies uh that go from the
small digital marketing outlet with to
people to the big uh the big Tech
corporates startups and scale UPS um my
role in the in the association is is
it's a new role is is is I think today a
year since we started is is to lead a
team that work along the the the member
companies uh around the complexities of
AI so this Association went from Pure
public policy to let's say opening the
umbrella and trying to help uh companies
understand Ai and in general emerging
Technologies uh in a much deeper way
um while performing public policy of
course and and continuing with with the
things that that an association like
that typic typically that typically do
uh my background is technical so I don't
come from P policy I have a PhD in
computer science so that's basically
what I bring to the table in in the
association uh with the rest of the
expert team in public policy regulation
and the laugh and and the thing and and
also some startups that I founded around
the data the data world so just trying
to answer your your your question um I
believe that the and what we're trying
to do not just just our talk but Walk
The Talk um anything that we talk when
we think about ethical principles when
we think about sustainability when we
think about uh uh algorithmic justice or
Equity nothing can happen without
transparency and explainability of the
systems uh and I believe we are still in
this moment where we need to work with
companies work with Regulators to fully
understand what transparency and
explainability means transparency
meaning in a very simple way uh how do
you do commend how do you understand the
own models that you are building or you
are using uh because the regulation
affects not only uh what you build but
also what you use so you may be using a
third party uh Mo AI U AI models and
you're still have to be compliant with
the regulation and not also what you
document but also how you share it with
your stakeholders which can be your CFO
but can be your user your customer or
the regulator of course or the Society
because you believe or you are forced to
depending on regulation to publish that
information to the to the to the public
Society uh and also explainability and I
think that's obvious but we will discuss
about that because um models have
different levels of complexity uh the
most complex and Powerful models at the
same time difficult to understand
typically people say that is blackbox no
they they're not Black Box we are the
Black Box uh if you ask uh uh an
autonomous uh car about how why it
turned right instead of left is going to
say perfect I'm going to give you the
250 million reasons why I chose to turn
right uh the problem is how we are able
to turn that into the 10 or 15 key
reasons why the that we can understand
why the car turned right and this is I
believe the the core from which we can
start building the rest of the uh
ethical principles and in order to build
the responsible AI wonderful thank you
that's very very good point if for
building um intelligent model models
that we can understand uh first of all
it's difficult for us to accept and
trust their results but also you point
out very clearly if we don't understand
what they're doing regulation is is very
challenging so excellent very great
points all right so I introduce myself
um uh my name is Ahmed gal from R
University um I am an AI by training I
have been doing AI for the last 30 years
um in computer science and uh being um a
university Professor I also teach AI um
undergrad and grad and uh being in the
intersection um of education and uh
doing AI research is really interesting
um in the last 15 years I switched my
focus into um the area of Art and AI in
particular so I have a lab called the
art and AI Lab at redgar which really
give me um the opportunity to to
interact with people in Humanities um
and uh people um in in art schools um
and and that was an eye opener for me um
to talk to these people and understand
um their concern and and their the
issues they are dealing with and that
was before the current hype of AI um so
now I mean uh when I hear many of the
discussion um I have uh totally
different perspectives of of many of the
people uh who are just started to think
about these kind of problems so for
today's um for this uh banel I mean uh
there are many many things we can talk
about um
it's multifold in terms of we talk about
implication of AI in ethics so um um
I'll keep it open uh for the discussion
great so just to be clear your lab um
you've been speaking with artists and
their concerns are about their the
rights to their creative yes a lot of uh
discussion about um um the ethical
aspects of uh using AI the ethical
aspects of uh uh training AI um issues
about uh artist identity um um uh there
are another banel later today about art
and AI I don't want to but may might not
be attending there so I can I can also
get dig into this so um one of the
particular issues about ethical AI in
general which I don't hear many people
discussing is the issue of identity um
which um for creatives is very
fundamental uh if you are a Creator
doing art or music or anything that and
decide to use AI in your process
obviously there is a problem of um what
the AI is trained on and and how that's
being used but you as a Creator uh when
you write a prompt for example and
create an image um and you claim
authorship of that as you you don't
realize that um um somebody else might
use very similar prompt and generate
something different right and it's
different uh not because of what you did
or what you add the system it's just
different because there's a random
random random number generator at the
back end like a fing a coin that give
you different outcome from that other
person outcome so now the issue of
authorship which is fundamental um to um
to many many creative domain and
fundamental in
Humanities uh is is problematic yes and
and this start to emerge now um and and
um once you are using AI now it becomes
a big problem big ethical problem yes
well and it's interesting since we're
talking about educational context what
what you just described when a human
does it we would call it or a student
does it we call it plagiarism but when a
machine does it what do we call it
anyway we can leave that open and I'll
uh pass on to our next panelist thank
you I appreciate everyone for being here
thank you for coming together for this
discussion my name is Mark kaban and my
my doctoral training is in behavior and
cognitive
psychology and and adult learning adult
development um but for the last 15 years
I've been leading uh an NGO for Refugee
youth that been displaced by War and
then I spent three years as a director
um of an ed leadership master's program
in a very Innovative um kind of Boutique
graduate school and I was surprised on
my first week on the job I got an email
um that the students that are connected
to our University were having a free
stem program and I thought that's really
fantastic um so I clicked on it and I
saw that it was uh Skunk Works which is
one of the divisions of loed Martin that
makes a lot of their weapons and uh they
were not only funding this program and
offering it for free but they were
actually teaching it themselves and so I
got really interested and what I did and
if you want to do this go on Google not
now but just do a quick search um of all
these arms manufacturers Alba locked
Martin uh Ron Boeing and then put stem
in K12 schools and what you'll see is
that they've been investing millions of
dollars um into these programs around
the country and not a lot of people are
talking about it when I say not a lot I
don't it's it's kind of zero if if you
look it up online they're just kind of
articles that are written in local
Publications announcing this as a very
good thing um I was a bit shocked by
that because if you watch the videos um
you know it's a it's part partly
indoctrination I remember one of the
first people speaking on the video from
Lockheed was saying that I'm proud to
make the foreign policy tools of our
country he's talking about the
F35 and so then I thought wow these
16-year-old kids are being groomed to
being arms manufacturers why does this
matter now because across the United
States there's been this Uprising that's
happening in 140 plus schools where this
is coming to question that should the
educational missions of universities be
tied up um in you know creating weapons
and making wealth off of that um
and so the ethical question is going to
be what's going to happen at the
universities right now and and how this
plays out and if this is going to down
trickle down to the K12 system will
these um quote unquote stem AI programs
going to be questioned for their ethical
uses as well why do I care about this I
care about this because my family was
displaced by a war an Imperial War I
care about this because I spent uh 10
years of my life working with children
that have been displaced um by very us
Imperial Wars and yet at the same time
it's a topic that's seldom spoken about
um in the US if you even say the word
empire people look at you like um you're
from a different planet like what are
you talking about uh because the United
States was an Anti-Imperialist kind of
country and its origin story so I think
uh I like the niche kind of topics that
we have you have a tough job ahead of
you I know about how to tie these
together but that's the thing that's on
my mind today no that's very good and
that's you know um one of the things you
point out is that the uh creators of
these Technologies are largely
corporations and we've had the point
about understanding how these
Technologies work there's an asymmetry
at information right and the and the the
creators of the technology have way more
information about the technology um and
they also have demands I've worked also
with Intel and Intel has um what they
called originally AI for youth but it
was a k312 program and they have a whole
series now of programs they're they're
framing is they're trying to uh prevent
the creation of an artificial
intelligence digital divide but you're
absolutely correct do we want artificial
intelligence education stem education
being led by corporations and their
needs or do we want it led by private
colleges or do we want it led by the
public big questions and the challenge
there is that the people that control
the knowledge about how these things
work that information is very
concentrated in the creators and not
well spread which makes it a challenge
for the uh public institutions to deal
with it so we are talking about this is
the global education Forum so um let's
talk about this idea of global education
the role AI can play in global education
um we look frequently in terms of sort
of the ethical questions about you know
sustainable development to the
sustainable development goals we have
one of the sustainable development goals
is around quality education number four
the other one is about lowering
inequities um there are many divides
Geographic divides uh social divides and
the question is will what's the role of
AI and educa what's the role of
education and the role of AI in
education in helping us confront those
divides and from your perspectives do
you have any ethical considerations
about that question about what AI will
do in terms of the the equity around the
globe and and those situations we can
start we can go however you want but if
you want to go around the circle again
please yeah absolutely um great and
difficult question um we're now working
in a in a project in in in order to
bring responsible AI in in in not not to
bring because it already exist but to um
push the the concept of responsible AI
in Latin americ and Caribbean and um
there I mean of course we we treat it in
in in different standpoints one is to
increase in compet the competitivity of
um competitiveness sorry of of small
companies and things like this but that
there are three elements that we're
starting to see to look in in detail and
probably you you know lot about this but
just to share um one is the vulnerable
population both in terms of economy and
in terms of exclusion of course there
another important thing and critical
which is gender and then this this thing
of sustainability and uh what we're
trying to do and we starting so it's
just kind of a
of an initial initial work is to try to
understand uh how this is done because
what one this the the difficulty of this
of this specific project is that it's
not a very tactical project it's a
strategic project so on the one hand we
have to bring responsibil to the region
or to work with the uh existing uh uh
local agents already there in order to
kind of bring the let's say experience
that we are having with the with the AI
act and all the things that are
happening here in Europe uh but at the
same time uh without having specific
projects that can bias us to the
specific things uh but I think those are
the three elements that is that we've
recognized as as critical or the most
critical that we've seen at least in
that area in order to to work in on the
role of AI and that is affected by
education okay so how can we first use
how can we educate in AI but how can we
use AI in that education to scale as
much as possible and that brings us
again to the concept of bias and the
concept of you know how do we do there
so for example I don't know you know but
for example the Spanish government uh
just announced a couple weeks ago that
they're uh going to have relationships
with different companies and different
institutions in order to build uh the
Spanish and other co-official languages
in Spain and and Latin America uh
language model okay uh for us that
brings us questions about whether that's
that's the best way to bring you know
the new AI uh to the region or is much
better and that's a question of uh
whether what a government like this
should just focus on the data to have
real quality data unbiased data as much
and bias as possible in order to leave
companies to bring llms or other models
to to the region so I don't have the
answer right now but basically those are
the areas that that we believe are going
to be critical in following years that's
interesting so you you brought in we
we've had one type of bias and that's
the the owner of the technology and the
way they you brought in the bi a couple
of biases one is the biased uh data and
then the other is the biased model all
of which can distort educational
outcomes Etc and then Regulators can
focus on the models or just Pro and and
to create in themselves or just getting
the best unbiased data set if there is
such a thing and make that available to
the private sector interesting which
brings if I may add just one thing there
a cultural bias I mean trying to bring a
single model that works for every single
country in this case Latin America
doesn't make any sense yes uh you know
each has a different approach to
knowledge to understanding to learning
uh which brings also additional issues
yes
okay want to take a shot at the uh role
of Education AI in in global inequities
and yes definitely I think um AI um um
now starting to give us ideas about how
to approach education a very different
way and I think we all experienced that
uh when we started dealing with SH gbt
in a question answer kind of shat uh
educational experience and how that's
very different from reading a book when
you reading a book everything is linear
the author already organized the topics
for you here when you shatting the
system you ask question get answer that
invoke your curiosity to ask another
question so it open a door for uh
problem solving oriented curiosity based
learning which is a fundamental way of
doing learning if you look at thousand
year ago and how learning was done it
was like that it's not was not
curriculum based learning that that we
have established in the last 200 years
it was mainly problem solving curiosity
based learning and now ai will enable
that uh and in a global and and scalable
way um uh so so now comes many many
questions uh in terms of how educational
institution which have been really
focused on curriculum based Mass
certificate given uh institution uh will
transform um in terms of these new ways
of of new and old ways of of of learning
and um and what um what the role of uh
AI in this institution and what's the
role of AI uh in IND
outside this institution but also comes
all the question about um
um um AI hallucination uh is the data or
the outcome that AI give you is really
correct or not what are the sources
these coming from all these are unsolved
problem in terms of AI at these point
and and that limits its use of
Education um but there are many many
other um uh issues as well we mentioned
the biases here which I um I hear a lot
about talking about bias in AI however
we don't really um confess that the
issue of bias exist in our education
system and culture system way before the
AI I mean we who who decide which
textbook we are using and what's go
inside that textbook to start with I
mean that has nothing to do with AI um
and and um is one big issue I can talk
about in details but this is one of the
fundamental issues yes
I think that's you're right I think
we're recognizing that even outside of
AI there there's implicit and bias and
and structural inequities built into
many things that we ignored previously
that are becoming very uh poignant as
the technology can exacerbate those yeah
and that's interesting so you you made
the statement that PRI previously
education was more of an exchange
questions and answers between Socrates
and his students in Greece or something
or today it's limited to Elite schools
where you can have a tutor that you have
your conversation with and but you're
arguing that the AI can then
reinvigorate and make that available
globally that same kind of dialogic
approach to education yes interesting
interesting uh do you know of any
organizations that are already
practicing that or is that just
something no I I I try to practice that
in my classes um like I mean um when uh
CHT came came around uh and I'm teaching
AI for undergrad course uh uh I supposed
to teach about these kind of things so
my first assignment to them is that you
sh DBT and and try to ask ask questions
and uh that they cannot answer or push
to the limits yes so that was very nice
experience because stud had to really
figure out what kind of question they
ask and keep uh pushing and and figure
out what is the limitation of these
systems um and again uh this semester I
was teaching course about computer
vision and Chad gbt came with another
version where uh you can now put an
image and and and answer a question so I
since that's a core of my course I
challenge them also to to see what are
the limit of this system put an image
start to ask question and see what kind
of answer they giving and what kind of
images they cannot answer or cannot
inere about um so I find fascinating to
use this kind Tools in classrooms um but
I'm I don't have a clear idea about how
this can be a part of a systematic way
of learning um I know for example um in
the last few years people have
introduced other Technologies like
YouTube um and in have this inverted
classroom kind of uh Concepts yes but
now how that evolve now with AI because
again when you look at YouTube uh
learning I mean everything's in YouTube
Right lots of lecture about anything uh
but again it's linear system you
basically it's like the same as reading
a book except they have many many
choices yes and so AI is is to different
from that and and um but for a student
who doesn't know what to look for uh
that's becoming a problem and that's the
role of universities when or schools
when you put a curriculum and and and
and lineup of topics and things like
that and and that's the Dilemma how can
we Bridge these two things these two
paradigms of teaching yes no no I I can
I'm thinking about how I might go back
to myself to the class and say you know
I don't know we're going to learn about
the uh Spanish Civil War or or the art
of balasz and then you put students in
groups and each they ask questions in Ai
and they try to compile compile some
kind of understanding and then the
professor orchestrates some kind of
overall integration of what they
discovered I can imagine all kinds of
experiments and models along that that
line if I may add that please probably
some of you already know it but I think
it's interesting I come from the
publishing industry and that's a typ
topic of discussion for at least the
last 15 years which is the Gutenberg
parenthesis uh this idea that so the
Gutenberg parenthesis parth okay yeah
the basic idea is that the way we learn
the way we understand the way we uh
model our world was changed like 500
years ago when the printed machine came
MH before that we learned in many
different ways in totally different ways
and it seems that the parenthesis is
what part of the theory is that is being
closed now yes uh with different
Technologies and now with AI and I think
is related to what you were saying is
kind of this new way of maybe going back
up in one side of this Q&A and and
discussions and but on the other way
taking advantage of the technology that
we have now uh and and is is is is you
know there are papers that show that you
know we learn we model our brain in
terms of how we learn to read and write
and this may change in the future so I
find it really fascinating how we may be
entering maybe in a different world uh
in
the in the future yeah a little bit of
Back to the Future or something along
those
I I don't know if I share your
perspective that students are going to
be sitting there with AI really engaged
for any kind of um significant amount of
time maybe in we're really bad at
predicting how technology plays out so
it's just really a wondering for me but
I I wonder if in 10 years we'll be
looking back at this moment and saying
how did we think that that was going to
change fundamentally the way that
students learn about things um that they
don't necessarily love right there
there's some topics that students don't
love and the reasons why they end up
learning those things is because they
have a loving relationship with their
teacher and with the people around them
many years ago when I was running my
Organization for Refugee youth one of
the biggest challenges that we had was
um the SAT test uh to get into college
and it felt very unfair
because all the studies show that test
scores are really closely related you
probably know this to income and that's
like the the biggest indicator of
success on on the test and that's
because a lot of these kids have private
tutors um and so here comes here comes
the College Board the organization that
creates the SAT and they partner up with
KH Academy and they had this big promise
to the world that we're going to change
the whole dynamic of this inequity and
all of these kids are going to be able
to have their own private tutor on this
KH Academy thing and it's all it's using
AI in different kind of ways a little
bit more uh instruct instructivist which
is like answer a question get another
question based on that and chat gbt is a
little bit different from that um but
what they ended up finding out and they
learned pretty quickly is that um
students from low income backgrounds
don't want to sit around on a computer
and just talk with it all day and I saw
this firsthand in my Academy it it was
like pulling teeth uh for the lack of a
better analogy with our kids they just
didn't want to do it and so then you
wonder why are the wealthy kids why are
they doing well it's because they're
being encouraged by these tutors they
have these relationships it's the social
Enterprise of learning that is there and
so I I think that um that's one
challenge that's there and KH Academy is
continuing to make all these uh promises
there's the I think it's called conmigo
if you go on the website another Google
search if you go on their website you'll
see things like have your own tutor have
your own pilot and it's actually not
that and I think it's I think it's a
little bit disingenuous and dishonest
because it's again I think following the
Playbook of what the College Board did
on the SAT test cuz it's not going to be
like having a tutor and I think that we
need to be realistic about what that
means but one one other thing that's
coming to mind about possible challenges
is everyone familiar with the term
hallucination here I think we're
starting to learn that word is basically
when chat gbt just makes something up
and sometimes they're really wild I I
was just doing a quick search on my
phone yesterday I think if you have an
Android WhatsApp has um uh has that
function there so I've been playing with
it a lot and and I just asked it what
did the Harvard president say about the
uh divestment movement of of for South
Africa in the 1980s and it said
something completely wild that the
president compared them to the clu clux
Clan I said I knew that that wasn't true
because I've been in this divestment
movement for a long time and I've never
heard that that wasn't true um and so
then I guess you wonder why are these
hallucinations happening if you're
someone who's learning code and it gives
you a false code and you're you're a
novice you're not going to know that
it's not true if you know what you're
doing you know this is not right he even
does basic arithmetic wrong as well um
and i' I've experienced that when I was
trying to use that to kind of uh surpass
certain uh cognition that I was not
interested in doing um and so I I I
think these are some of the the issues
that we have and so well-resourced
schools are going to have a lot more
time to teach young people about how to
discern how to make decisions about
these how to you know do what we call uh
prompt engineering and students at
schools that are less funded that have
less ability to do that kind of guidance
um those students are going to be you
know the the the object of these
Technologies in ways that we cannot
predict um and so that that's what
that's some of the things that I'm
thinking about right now that feel a
little bit precarious I think no that's
very good um so I'm gonna we'll open it
up to some questions the audience so be
thinking if you have a question um so
yes you you've again nailed
again there there exists these biases
and inequities um in the educational
system without AI right and I um and I I
experienced it when I was I went to
California public school um and I
remember being at my friend's house and
their friend's father would come in and
say you guys should be studying the
smart kids are studying right now and
then I moved my kids to um the greater
Washington DC area where there's the
highest concentration of advanced
degrees everyone there works for the
World bank and they have a PhD or so um
of course that has an influence on the
kids and so my problem was and that my
my kids wouldn't study is they wouldn't
stop studying because of all of their
friends were studying completely I had
to go upstairs and say stop studying
come down and watch Game of Thrones with
me you know um come play fun so um and
it speaks exactly to that and it you
know it doesn't it it wasn't
intellectual capability or potential it
was the environment in which they f
found themselves and um you know that
that environment motivated them by their
peers and by the system to be um excel
in education in ways that in my
environment we weren't encouraged to do
it's what non chomsy calls the luxury of
leisure yes right that money that
provides those opportunities to happen
that's right yeah and then the question
then again AI we've heard some
possibilities that it will minimize or
or lower these barriers because one of
the things about the technology unlike
say an autom or something is if you can
get access to the internet which is not
Universal yet but you can have access
then to these models so in that sense it
could democratize access in way other
Technologies in the past haven't but
then again how they're used the
environment in which they're used is
consequential yeah I I think I think
we'll find with AI what we found in many
other Technologies right when I think 10
years ago when mukes came out those
massive online courses they said this is
going to democratize learning and and I
don't really think that it did um and so
so there's a lot of these kind of
platitudes that are happening right now
and um yeah I I lost the plot in my
thinking but we'll come back to it well
we're we're we're spread we're traveling
a lot space so does anyone have any
questions we're yes go right ahead so so
I don't have a microphone but I'll speak
loud just to propel a little bit the
conserv so I I get your point regarding
usal assistance to support
personaliz now I think it's it's also
important to keep in the ra that this
we're at the very early
stages um I do believe honestly that in
a few years every student is going to
have a
c theep you go into not only the areas
of opportunity from an academic
perspective for a particular student but
also how the student learns
I think at that point maybe I'm but I
truly think that we're about to the
disruption and the driver is going to be
vir assist
that um so I guess kind of I I I just
wanted to stimulate
controvers I believe there are going to
be other ethical aspects that will
we'll be discussing in a few years today
we're talking maybe about um can we use
from a privacy perspective the data to
truly know how you learn that's a bit
controversy but that's going to change
when we consider privacy the kids today
it's a different concept of privacy so
that's number one and and and when we
talk about the future we'll see and
there's a book called AI 241 I don't
know if anyone has seen it there's
and there are issues like
How Deeply personally involved the
student could get with these assists
they become their
friends
boyfriends the person that you tell your
secret and that's a different I think
it's a way deeper eal
implication discussing today wanted to a
little bit of okay so for the for the
people that couldn't hear um the
question was uh broad but uh was the
idea of personalization in education
through the virtual Assistance or AI
agents um and that opens up
opportunities that we similar to what
we've discussed but there was also then
ethical questions around the information
needed to personalize that there's
questions of privacy around that and
then also questions about what kind of
relationships uh students are going to
create with these virtual assistance and
what the implications of that so as if
we didn't have enough ethical concerns
to discuss let's uh toss that out to our
panel if I may answer one of your your
comments I I found it really
interesting
um using the systems that you mentioned
is going to force us to be very critical
with how we understand those systems
themselves so let me put you a very a
very simple example uh one of the
companies I built in in in
2018 uh did um use Ai and statistics to
understand how people read how people
were reading books and listening to
audio books and things like this so we
had these models and these things one of
the things we always had to explain to
our customers and in order for them to
put into into the privacy policy is that
uh nobody understood and maybe now this
is obvious but I can tell you in 2018 it
wasn't obvious is that if I know what
you read I'm going to understand your
sexual preferences your religious
beliefs I'm going to understand so much
about you and it wasn't taken as as U
personal information uh so for many of
some of our customers I wouldn't say
many but some person were like well
we're just we're just seeing how people
read books no no no no uh even by
knowing where you stop reading I can
infer what are your interests and this
was just a small very small boutique
company for very small very very
specific uh education related uh uh
stuff uh what we can have now in terms
of and this is one of my my fears is
when you have a huge model that is going
to be used as the operating system for
Access or for for for for many people to
access that information uh any small
mistake there small uh Cave there can
affect uh lots of people and that's
gonna that's I think that's going to be
one of the challenges there that that
you and I are going to have in the in
the near future so that's very
interesting and already now it's uh
Amazon supposedly with their data about
you're purchasing they can predict
things like women's fertility or they
can even predict the way you will behave
even though you don't believe you behave
that way right
you so um we're gonna this will be we're
coming to the end here we have about
five more minutes so let's just follow
this line and and uh either one of you
if you want to take a shot at that
question about the virtual agents and
something else to say on that yeah I
just I want to follow up on that I mean
obviously there's no answer this is the
open question open uh I just want to um
take this into the next level which is
the role of Technology have been in the
last um since invention of the brand to
the invention of the internet uh to now
ai is is giving access to knowledge
right that's basically uh it's so 30
years ago if you are living in a world
third world country don't have access to
knowledge other than local library the
internet allow us to have access uh so
it's opportunity an AI will allow more
access and that lead us to the
fundamental question of what is
education what when you go to school
what are you looking for if you have
Global access to knowledge um why going
to school and that go back to your point
which is basically about the SAT exam
it's it's really about um uh motivation
and and uh uh coaching and uh uh
teaching students um who doesn't want to
learn basically who doesn't see the
motivation why you learn so that's why
they go to school and and we kind of
push them through that um and this is
fundamental thing that AI cannot
solve
so not the problem of AI I mean it's a
problem of how we motivate kids to learn
or even ourself to learn
basically because that's going to change
maybe I don't know I think for person
who are motivated AI would be a great
opportunity because it give them access
to Global Knowledge but for for for kids
who are not not motivated enough it's
our job as Educators to to to give them
these skills to how to motivate them and
how to make teach them how to be
perseverance and and how to do critical
uh thinking these are the basic things
that we need to teach our kids because
the knowledge is is already now
accessible and and we can get aners
themselves so um I think AI will help us
really re understand the role of
Education all together in the future can
I give a counter example really fast yes
we might be so I've seen examples of AI
being used for autistic kids to to
motivate them very
specifically um and they are successful
yeah
and there's a lack of access to
institutions
for this
ISF maybe yes yes amazing but I'm I
agree with you the role of the teacher
has to keep involving coaching
supporting and okay well and that it
comes back to the the much of the
ethical questions here is that we don't
know how people are going to use the
technology we don't know necessarily
what the technology will do um we are
still at the very beginning uh and one
thing about technology it especially
digital Technologies they are
commercialized they are ubiquitous in
society and we only find out the
unintended consequences the ethical and
moral considerations often after the
fact after they're already ubiquitous so
this will be a ation that will continue
here in this forum and in the world
going forward I'd like to thank all of
the panelists and everyone who attended
this session um and I believe is it
break time or is it okay and we have a
break and uh if you have questions for
the panelists I I guess they'll stick
around okay thank you all thank you did
a really great
job makes a big difference one the facil
really good job app how you I will take
that to heart thank
you it was you guys I don't know how
they decided to put I had no
idea I think AI decided cuz they said
that they decided all the topics so
maybe they did that as
well I know it's AIT
weird I'm in Cal in San Diego San
Diego yeah man able to live in San Diego
for a long time right now you're in in
DC in DC not a bad place to be no no I
said if be
theost I hear you I hear
you man that that was great my brother I
appreciate you man
this
Ver Más Videos Relacionados
GEF Madrid 2024: Welcome remarks
AI (Artificial Intelligence) Literacy | Holistic Competencies
GEF Madrid 2024: Academic-Corporate Collaboration in AI
Can AI Agents be Ethical? (Ethics of Artificial intelligence in Medical Imaging)
What Comes After AI?
Anonymous Charity: Artificial Intelligence and The Planned Destruction of Humanity. May 6th, 2019
5.0 / 5 (0 votes)