Responsible AI for Human Support (RAI4HS)
by Dr Daniele DI MITRI (email@example.com)
Many concepts of Artificial Intelligence (AI) such as ‘machine learning’ or ‘artificial neural networks’ are inspired by human intelligence. Most of the applications people use every day are powered by human data such as movie streaming apps, speech recognition, machine translation or music recommendations. But can we design AI tools that can support humans in their decision and augment human abilities in day-to-day life? In this seminar, we will focus on different ways in which AI can support human decisions and activities, with a particular focus on education and human learning. We will look at the opportunities that AI brings to humans as well as the connected risks. We will delve into the assumptions, models and principles which need to be embedded in the AI system to make them fair, trustable, explainable and understandable for the user.
The participants of this seminar will learn about:
- Peculiarities of AI systems for human support
- Various roles that AI can have for humans with a strong focus on education
- Design components of Human-AI systems
- Connections to human psychology, communication and learning
- User modelling and user adaptation
- Roles that humans can have in shaping and influencing AI systems
- Multimodal and multi-sensor data collection and analysis
- Role of feedback in AI applications
- Components for Responsible AI systems including Ethics, Privacy and Trust
- Algorithmic transparency and explainability
- Research directions and opportunities in the field
This is a “design seminar” that uses Problem Based Learning. The participants will divide into couples and conduct a research assignment that will last the entire duration of the seminar. The assignment consists of the design of an Artificial Intelligence system applied in the field of education and learning. The assessment will be based on a final presentation and a final report to be submitted by each team.
Due to Corona, the seminar will take place via Moodle and Zoom Sessions. The seminar will require mandatory participation in one online meeting (Zoom session) per week for the duration of 1 or 2 hours. In the online meetings, the participants of the seminar will be able to attend different lectures, group-discussion, Q&As and brainstorming sessions.
This seminar is offered by the Educational Technology group (www.edutec.science) of the DIPF, the Leibniz Institute for Research and Information in Education. The whole seminar will be in English.
The final assessment would be based on:
- a group-project work to be carried throughout the seminar
- final presentation
- final scientific report
- active participation in the seminar
VIRTUAL MASTERCLASS HUMAN A.I. 10th September 2020
Subscribe for free! https://www.ou.nl/-/aanmeldformulier-masterclass-humanai
Do you want to learn how to apply principles of Responsible AI in the design of AI systems that interact with humans? Many concepts of Artificial Intelligence (AI) such as ‘machine learning’ or ‘artificial neural networks’ are inspired by human intelligence. Most of the applications people use every day are powered by human data such as movie streaming apps, speech recognition and translation or music recommendations. But how to best design AI tools which can support humans in their decisions and augment human abilities in day to day life? In the Human AI masterclass, we will focus on different ways in which AI can support human decisions and activities.
COSTS You can participate for free, and online
* 09:45 - 10:00 Welcome words and preparation
* 10:00 - 10:15 Introduction presentation
* 10:15 - 10:30 Round of presentation
* 10:30 - 11:15 Lecture 1 - Artificial intelligent systems for human support
* 11:15 - 12:00 Break-out session 1 - designing an AI system
* 12:00 - 13:00 Lunch break
* 13:00 - 13:45 Lecture 2 - User modelling and user adaptation
* 13:45 - 14:30 Break-out session 2
* 14:30 - 15:15 Lecture 3 - Responsible AI and ethics
* 15:15 - 16:00 Break-out session 3
* 16:00 - 16:45 Final presentations of the projects
* 16:45 - 17:00 Conclusions & evaluation
INTRODUCTION PRESENTATION (10:00 - 10:15)
We explain the motivations behind the course, in particular why talking about AI interfacing with Humans is relevant in this context.
ROUND OF PRESENTATION (10:15 - 10:30)
Participants present themselves in 30 seconds LECTURE 1 - ARTIFICIAL INTELLIGENT SYSTEMS
FOR HUMAN SUPPORT (10:30 - 11:15)
Lecture 1 gives an overview of the existing AI
systems for human support including emotion
detection, speech recognition and recommendation
engines. Potential connected risks in these
areas are introduced.
BREAK-OUT SESSION 1 (11:15 - 12:00 )
The participants gather in groups to decide on the functionality of their AI tool by picking from a pool of ideas of HumanAI applications or by coming up with another interesting one.
BREAK-OUT SESSION 2 (13:45 - 14:30)
Participants continue working on the selected functionality, this time trying to the aim of applying user-modelling and HCI principles to make sure their AI tool fully models and understands
the user correctly.
LECTURE 3 - RESPONSIBLE AI AND ETHICS (14:30 - 15:15)
Lecture 3 touches on three main pillarls constituting the Responsible AI: Ethics & Privacy, Fairness and Explainability.
BREAK-OUT SESSION 3 (15:15 - 16:00)
In the last breakout session, the participants try to improve the design of their AI tool/ functionality having in mind the Responsible AI principles.
FINAL PRESENTATION (16:00 - 16:45)
Each group presents their AI-tool functionality explaining the decisions taken.
- Dr. Daniele Di Mitri
- Dr. Stefano Bromuri
- Professor Dr. Roland Klemke
- Ekaterina Muravyeva, MSc