Abstract
This document contains the outcome of the first Human behaviour and machine intelligence (HUMAINT) workshop that took place 5-6 March 2018 in Barcelona, Spain. The workshop was organized in the context of a new research programme at the Centre for Advanced Studies, Joint Research Centre of the European Commission, which focuses on studying the potential impact of artificial intelligence on human behaviour.
The workshop gathered an interdisciplinary group of experts to establish the state of the art research in the field and a list of future research challenges to be addressed on the topic of human and machine intelligence, algorithm’s potential impact on human cognitive capabilities and decision making, and evaluation and regulation needs.
The document is made of short position statements and identification of challenges provided by each expert, and incorporates the result of the discussions carried out during the workshop. In the conclusion section, we provide a list of emerging research topics and strategies to be addressed in the near future.
1 Human behaviour and machine intelligence in the digital transformation project HUMAINT: objectives and
workshop
Emilia Gómez
Centre for Advanced Studies, Joint Research Centre, European Commission Universitat Pompeu Fabra
Over the last few years, thanks to an increase in data availability and computing power, deep learning techniques have been applied to different research problems related to computer vision, natural language processing, music processing or bioinformatics. Some of these models are said to surpass human-level performance (e.g. image recognition (He et al., 2015) and model highly abstract human concepts such as emotion (Kim et al.,
- or culture. The practical exploitation of such algorithms brings up a discussion on the impact of these algorithms into the ways human behave:
- On one side, machine intelligence provides cognitive assistance and complement humans to interpret data more efficiently and discover hidden knowledge in large data resources.
- On the other side, these algorithms may also affect the way we perform some cognitive tasks and thus affect autonomy and decision making. This is especially relevant when algorithms perform tasks in a high level of abstraction and when they may contradict and influence human interpretations. The goal of the Human behaviour and Machine Intelligence (HUMAINT) project, carried out at the Centre for Advanced Studies, Joint Research Centre of the European Commission, is to (1) provide a scientific understanding of machine vs human intelligence; (2) analyse the influence of current algorithms on human behaviour and (3) investigate to what extent these findings should influence the European regulatory framework.
This document summarizes the results of the first HUMAINT workshop, which took place in Barcelona on 5-6 March 2018, and brought together key researchers from complementary disciplines and backgrounds. The workshop was defined with two main goals:
- Build an interdisciplinary roadmap on human vs machine intelligence, potential algorithm’s impact on human cognitive capabilities and decision making, and evaluation and regulation needs. We intend to study the state of the art, identify future research challenges, and reach a consensus on practical way to address these challenges.
- Build a community of researchers for the HUMAINT project to collaborate with.
1.1 Workshop details
The workshop was structured in a set of short position presentations followed by panel discussions. Presentation slides can be found at the workshop web page https://ec.europa.eu/jrc/communities/community/event/humaint-kick-workshop.
DAY 1
9.00-9.30: Welcome (Jutta Thielen-del-Pozo, Vanesa Daza, Emilia Gómez)
(1) Human vs machine intelligence
9:30-12:00: Presentations
- Joan Serrà. Unintuitive properties of deep neural networks.
- Gustavo Deco. Whole brain modelling and applications.
- Karina Vold. Extended minds and machines.
- Rubén Moreno-Bote. Slow and fast biases in decision making. 12:00-13:00: Panel discussion. Presentation and moderator: Ramón López de Mántaras
13:00-14:00: Lunch
(2) Algorithms’ impact on human behaviour
14:00-16:30: Presentations
- Henk Scholten. Digital transformation and governance of societies.
- Nicole Dewandre. Artificial intelligence: an interesting leverage point to rethink humans' relations to machines…and to themselves.
- Carlos Castillo. Algorithmic bias.
- Fabien Giraldin. Experience Design in the Machine Learning Era. 16:30-17:30: Panel discussion. Presentation and moderator: Verónica Dahl
DAY 2
(3) Evaluation and regulation of algorithms
9:00 - 11:30: Presentations
- Alessandro Annoni. Digital transformation and artificial intelligence: the policy-oriented perspective.
- Martha Larson. Reality, requirements, regulation: points of intersection with the machine learning pipeline.
- Anders Jonsson. Benchmarks and performance measures in artificial intelligence.
- Ansgar Koene. The IEEE P7003 Standard for Algorithmic Bias Considerations.
- Heike Schweitzer. Algorithmic decision-making-in need of (which) regulation? 11:30 - 12:30: Panel discussion. Presentation and moderator: Xavier Serra.
12:30-14:00: Lunch
(4) Application domains and new paradigms
14:00 - 15:00: Presentations
-
Sergi Jordà. Enhancing or Mimicking Human [Musical] Creativity? The Bright and the Dark Sides of the Moon.
-
Miguel Ángel González-Ballester. Machine learning in healthcare and computer-assisted treatment.
-
Fabien Gouyon. The influence of machine intelligence on the music industry.
-
Blagoj Delipetrev. HumanAI.
-
Luc Steels. Will AI lead to digital immortality? 15:00 - 16:00: Panel discussion. Presentation and moderator: Perfecto Herrera.
16:00 - 17:00: Wrap-up session. Moderator: Emilia Gómez.
1.2 Goals and structure of the report
The goal of this report is to provide an interdisciplinary state of the art overview on the interaction between human and machine intelligence and identify which are the research challenges related to this interaction and the practical ways to address them.
In order to do so, the document is structured in five parts. The first four parts are related to the four sessions of our workshop. Each part contains a series of original contributions and position statements provided by workshop presenters. They are complementary to their presentation slides and contain their statements with respect to a set of questions proposed beforehand. In addition, each part includes a summary of the workshop discussions provided by panel moderators. In part V, the report incorporates input from other scholars involved in our discussions that were not able to present at the workshop.
The report finishes with some general conclusions of this interdisciplinary discussion and of the HUMAINT project and some directions for future research within the HUMAINT project and beyond.