The aim of the workshop is to bring together AI researchers with complementary skills and background and to foster a discussion aimed at cross-fertilizing different AI sectors and to provide concrete means for autonomous reasoning agents design and evaluation.

In a computer-aided problem solving process, there is always a substantial human inter-vention enabling the encoding of a problem in a machine-understandable model, that can be solved automatically through a problem solving technique/algorithm. The human inter-vention is essential for identifying problem components, common-sense and hidden knowledge in the problem description.

In a long-term vision, next-generation artificial cognitive systems and robots will be au-tonomous end-to-end solvers that perform the whole problem-solving process without any human intervention. Starting from a (possibly multi-modal) problem description, an end-to-end problem solver should automatically understand the problem, identify its compo-nents, devise a model, select a solving technique, and find a solution. Such autonomous in-telligent agents should be pro-active and problem-solving driven; deep understanding and deep reasoning, not necessarily based on big-data, will be a crucial ingredient for their design.

In this context, it would be important to identify specific challenges, to assess the level of autonomy achieved, the effectiveness of end-to-end solvers, and to ease the dissemination of AI results to a general audience. This ambitious goal requires an unprecedented integration of AI area and techniques such as Natural Language Processing, Machine Learning, Constraint-based reasoning, Logic, Planning, Cased-based reasoning, Human-Machine Interaction, and Cognitive Science, and could represent an important step forward reducing the fragmentation of modern AI.

The workshop aims at providing some initial insights on the following research questions:

1. To what extent AI techniques can remove the human intervention and let the computer explore the whole problem-solving path autonomously?
2. How far is a next generation of AI engines going beyond the pure question answering by embedding deep reasoning and understanding?
3. Whether existing techniques such as Natural Language Processing, Vision, and Machine learning should be extended to extract and use focused knowledge driven by the problem solving process in a limited domain, instead of general knowledge?
4. How to address reasoning aspects typical of human problem solving such as metaphorical reasoning, abstraction, creativity, and intuition?
5. How research results from Artificial Intelligence, Natural Language Processing, Vision, Cognitive Science, Mathematics, Education can be put together to cope with the inherent complexity of the faced problems?
6. Is it possible to define new challenges and crafting real competitions, to push forward the limit of what computers can do autonomously in problem solving, and to measure the level of achieved intelligence?
7. On the road to autonomy, which is the role played by human-computer interaction in the problem solving activity along the line of AI collaborators?
8. As in the case of Computational Thinking, what would be the implication of "AI Thinking" in the educational setting? Can AI thinking equip students of different disciplines with new insights in modelling and problem solving?
9. Is this the time to go "Beyond the Turing Test"? What about the risks for humanity related to these next-generation intelligent agents?