#artint en Educació Superior #artinthe (1/3)

Començo avui una petita sèrie de tres entrades al blog sobre la intel·ligència artificial (i la Internet of Things) en educació superior, que he preparar al meu mesoblog Molecularity Report. Ho faig en anglès perquè així és més fàcilment reutilitzable.


Image CC-BY-SA from A_Mind_Map_on_ICT_and_Pedagogy.jpg at Wikipedia.

Academic conferences that promote cutting-edge research in this field include Artificial Intelligence in Education (AIED), Intelligent Tutoring Systems (ITS), Educational Data Mining (EDM) and Learning Analytics & Knowledge (LAK) [8]

Systematic review of research on AI applications in HE – where are the educators? [1] enumeartes four areas: profiling and prediction, assessment and evaluation, adaptive systems and personalization, and itelligent tutoring systems. Talks about intelligent tutors, expert systems and chatbots. Names Institute for Ethical AI in Education.

Conjecture: every aspect of learning can in principle be precisely described and a machine can be made to simulate it. Ai refers to machine learning, natural language processing, data mining, neural networks, algorithms, … AI is used for classification and profiling. Also recognizing patterns, make predictions, apply newly discovered patters… The concept of agent is central to AI. Virtual agents can act as teachers, facilitators of students’ peers. THere are virtual or remote labs.

Another paper reviewed refers to personal tutors, intelligent support for collaborative learning, and interlligent virtual reality. Learning is a social exercise (thus… social networks should be useful!)

Perspectives for AI tools are learner-facing, reacher-facing, and system-facing.

Useful in admission, counselling and library services, besides assessment, feedback and tutoring.

Focusing on teaching and learning, on finds teaching course content, recommeding/providing personalisez content, supporting teachers in learning and teaching desing, using academic data to monitor and guide students, and supporting representation ofknowledge using concept maps.

One may think of a hybrid reccommender system to help with reaching strategies. Flexible and interative, personalized learing opportunities fhould relieve teachers from burdens like grading, so they may focus on their real task: empathic hyman teaching.

Problem: educational surveillance is questionable. Ai systems require control y humans. We should go beyond the tools, and talk about learning.

International Journal of AI in Education (IJIAED) [2] carries a fair number of papers.

UNESCO report on challenges and opportunities in AIEd for sustainable development: preparing teachers for AI-powered educations, quality and invlusive data systems, ethics and transparencet. [3]

AI in Education: compendium of promising initiatives (Mobile Learning Week 2019) [4]

Driving reader engagement.
Special needsd, wearables.
Educaiton beyond the classroom
Intellignt tutoring system
Computer games
Augmented intelligence
Increase of movitavion and engagement
Digital Intelligence Virtual Assintan
Support of Human Intelligence.
Symbolic AI and machine learning (OERs)
Digital inclusion
Humanizing altorithmic societies
Experimenting with AI in the classroom

How to recognize AI snake oil [5]

Perception is one of a few areas in which AI has make rapid progress. Ehical concers because of high accuracy. Automatic judgement is fa from perfect, but improving. However, reasonable people can disagree about the correct dicision; ethical concers in part because some error is inevitable. Predicting social outcomes is fundamentally dubious: predicting job performance, for example; ethical concerns amplified by inaccuracy.

And note: regression analysis is a hundred years old! Regarding social outcomes, there is a lack of exaplinability.

7 Types Of Artificial Intelligence [10]

Will AI replace university lecturers? Not if we make it clear why humans matter [6]

Forget robo-lecturers whirring away in front of whiteboards: AI teaching will mostly happen online, in 24/7 virtual classrooms. AI machines will learn to teach by ferreting out complex patterns in student behaviour – what you click, how long you watch, what mistakes you make, even what time of day you work best. This will then be linked to students’ “success”, which might be measured by exam marks, student satisfaction or employability.

The AI tutor will design personalised learning plans that optimise each student’s outcome. Should one student watch their lecture at breakfast time, or in the evening? Where should their first test pop up in a busy schedule? How much preparation will they need to understand a certain concept? While a skeleton crew of humans would be needed initially to design curriculums (the creative bit) and film lectures (CGI is still too expensive), AI tutors could do the rest.

Consuming content is not equivalent to learning.

How AI and Data Could Personalize Higher Education [7]

With a personalized learning experience, every student would enjoy a completely unique educational approach that’s fully tailored to his or her individual abilities and needs. This could directly increase students’ motivation and reduce their likelihood of dropping out. It could also offer professors a better understanding of each student’s learning process, which could enable them to teach more effectively.

Chatbots at universities increase student motivation (e.g., Univ. Murcia) and adress questions 24/24. Answering students’ questions result in a large volume of big data that would be obtained regarding students’ concerns and areas of interest. This data could be analyzed to help enable universities to create innovative new services and programs to further improve students’ educational experiences.

The most crucial point to address (in AI at unviersities) is the way in which educational institutions can best prepare students for the new technology-based world and the many disruptive technologies that will change the way people work.

It is essential that students understand that over time, more repetitive and routine tasks will be automated and performed by artificial intelligence, automation, and robots. However, there will always be roles requiring creative skills, cognitive skills, and emotional intelligence skills

University leaders and administrators should become proactive in initiating pilot programs to test the use of AI in various ways, while critically considering the results and the ethical obligations that must be met along the way

Letting Artificial Intelligence in Education Out of the Box: Educational Cobots and Smart Classrooms [9]

Themes of future research could include creating systems that provide social interaction (an extension of the Computer Supported Collaborative Learning that already exists); exploring new modalities for interfaces between learners, teachers and supporting technologies and investigating the application of the Internet of Things in education.

Embedding AIED into smart classrooms would produce real time data streams from lots of sensors in the classroom and focused on the learners would require huge amounts of Educational Data Mining and would lead to new models of how learners behave in the wider environment of the classroom and not just with the instructional packages that we have developed to date

Interesting quotes:

Teaching is a creative, insightful, collaborative, soul-enriching human activity [6]
The best results will come from combining the strengths of AI and human abilities [7]
We must never underestimate the value of human interaction and critical thinking in the field of education [7]
Deception is an integral part of AI and robotics. In some ways, AI is the science of illusion. [a]

[1] https://educationaltechnologyjournal.springeropen.com/articles/10.1186/s41239-019-0171-0
[2] https://www.springer.com/journal/40593
[3] https://unesdoc.unesco.org/ark:/48223/pf0000366994
[4] https://iite.unesco.org/publications/ai-in-ed-compendium-of-promising-initiatives-mlw-2019/
[5] https://www.cs.princeton.edu/~arvindn/talks/MIT-STS-AI-snakeoil.pdf
[6] https://www.theguardian.com/education/2019/sep/06/will-ai-replace-university-lecturers-not-if-we-make-it-clear-why-humans-matter
[7] https://hbr.org/2019/10/how-ai-and-data-could-personalize-higher-education
[8] https://en.wikipedia.org/wiki/Educational_technology#Artificial_intelligence
[9] https://link.springer.com/article/10.1007/s40593-016-0095-y
[10] https://www.forbes.com/sites/cognitiveworld/2019/06/19/7-types-of-artificial-intelligence

[a] https://link.springer.com/article/10.1007/s10462-007-9048-z