Machine Intelligence Workshop @ NIPS 2016



Recent years have seen the success of machine learning systems, in particular deep learning architectures, on specific challenges such as image classification and playing Go. Nevertheless, machines still fail on hallmarks of human intelligence such as the flexibility to quickly switch between a number of different tasks, the ability to creatively combine previously acquired skills in order to perform a more complex goal, the capacity to learn a new skill from just a few examples, or the use of communication and interaction to extend one's knowledge in order to accomplish new goals.

This workshop aims to stimulate theoretical and practical advances in the development of machines endowed with human-like general-purpose intelligence, focusing in particular on benchmarks to train and evaluate progress in machine intelligence. The workshop will feature invited talks by top researchers from machine learning, AI, cognitive science and NLP, who will discuss with the audience their ideas about what are the most pressing issues we face in developing true AI and the best methods to measure genuine progress. We are moreover calling for position statements from interested researchers to complement the workshop program (see below).

The workshop will also introduce CommAI-env, a new environment for the development of communication-based AI, to the research community, encouraging discussion on how to make it the ultimate benchmark for machine intelligence. CommAI-env is an interactive playground where systems can only succeed if they possess the hallmarks of intelligence we listed above. A prototype of the environment is already available, so that researchers interested in submitting position statements to the workshop can experiment with it and take it into account in their proposals.

Slides from the Talks

Nursing Turing’s Child Machine: Towards Communication-based Artificial Intelligence, Marco Baroni
CommAI-env, Allan Jabri
Human-like dialogue: Key challenges for AI, Raquel Fernandez
Learning how to Learn Learning Algorithms: Recursive Self-Improvement, Jürgen Schmidhuber
Finding a Jack-of-All-Trades: An Examination of Transfer Learning in Text Comprehension, Rudolf Kadlec, Ondrej Bajgar, Jan Kleindienst
Consolidating the Search for General AI, Marek Rosa, Jan Feyereisl
Virtual Embodiment: A Scalable Long-Term Strategy for Artificial Intelligence Research, Douwe Kiela, Luana Bulat, Anita L. Vero, Stephen Clark
General Intelligence through playing games?, Julian Togelius
Minimally Naturalistic AI, Steven Hansen
Remarks on the CommAI-env, Gemma Boleda

Invited Speakers

Fernando Diaz (MSR)
Emmanuel Dupoux (LSCP, Paris)
Raquel Fernandez (ILLC, Amsterdam)
Brenden Lake (NYU)
Jürgen Schmidhuber (IDSIA, Lugano)
Arthur Szlam (FAIR)
Julian Togelius (NYU)


Marco Baroni
Allan Jabri
Armand Joulin
Germán Kruszewski
Angeliki Lazaridou
Tomas Mikolov
Klemen Simonic


Time Description
08:40-09:00 Marco Baroni: Introduction
09:00-09:20 Marco Baroni: A roadmap for communication-based AI
09:20-09:30 Allan Jabri: The commAI-env environment for communication-based AI
09:30-09:55 Raquel Fernandez: Human-like dialogue: Key challenges for AI
09:55-10:20 Jürgen Schmidhuber: Learning incrementally to become a general problem solver
10:20-10:30 Rudolf Kadlec, Ondrej Bajgar, Jan Kleindienst: From particular to general: A preliminary case study of transfer learning in reading comprehension
10:30-11:00 Coffee break
11:00-11:10 Marek Rosa, Jan Feyereisl: Consolidating the search for general AI
11:10-11:20 Alex Peysakhovich: Gaining insights from game theory about the emergence of communication
11:20-11:30 Tomo Lazovich, Matthew C. Graham, Troy M. Lau, Joshua C. Poore: Socially constructed machine intelligence
11:30-11:40 Douwe Kiela, Luana Bulat, Anita L. Vero, Stephen Clark: Virtual embodiment: A scalable long-term strategy for Artificial Intelligence research
11:40-12:30 Panel on basic requirements for machine intelligence: Angeliki Lazaridou (moderator), Katja Hofmann, Brenden Lake, Jürgen Schmidhuber, Arthur Szlam, Jan Feyereisl, Rudolf Kadlec, Armand Joulin
12:30-14:00 Lunch
14:00-14:25 Brenden Lake: Building machines that learn and think like people
14:25-14:50 Fernando Diaz: Measuring Performance of Intelligent Systems
14:50-15:00 Jon Gauthier, Igor Mordatch: A paradigm for situated and goal-driven language learning
15:00-15:30 Coffee break
15:30-15:55 Arthur Szlam: In praise of fake AI
15:55-16:20 Emmanuel Dupoux: An evolutionary perspective on machine intelligence
16:20-16:45 Julian Togelius: Are video games the perfect environments for developing artificial general intelligence? Which kind of general intelligence?
16:45-16:55 Steven Hansen: Minimally naturalistic Artificial Intelligence
16:55-17:05 Gemma Boleda: Remarks on the CommAI-env
17:05-18:00 Panel on CommAI-env and other environments for the development of AI: Germán Kruszewski (moderator), Julian Togelius, Tomas Mikolov, Emmanuel Dupoux, Raquel Fernandez, Alex Peysakhovich, Gemma Boleda, Igor Mordatch

The schedule may change prior to the start of the workshop.

CommAI-env: An Environment for Communication-Based AI

Scientists at FAIR are organizing a challenge for intelligent machines learning to communicate in natural language. The competition consists of language tasks that get incrementally harder. A prototype of CommAI-env, the interactive environment supporting the competition, is available here.

Join our Facebook page, for general discussion about the project:

For more information on the ideas motivating the environment and the planned competition, see:

A Roadmap towards Machine Intelligence