The project will have a significant contribution to the promotion of Second Life amongst researchers and practitioners and will also generate important results in various disciplines. Below we provide the list of key areas targeted by the project and outline the most significant contribution to each of those areas in more detail.

Second Life

In its current form Second Life features a large variety of virtual environments, many of which represent sophisticated and visually stunning designs. One of the problems we believe Second Life is currently facing is a lack of dynamics in those environments.

Our research suggests that having virtual AI characters interacting with human users would make Second Life more engaging and under certain conditions will increase the immersion of the users and even the likelihood of them reaching the state called “the flow” (the state of optimal experience which many psychologists believe being the key driving force for the humans in enjoying performing a certain task).

While it may be difficult for Second Life to compete with other Virtual Worlds platforms available on the market in terms of rendering quality, in our opinion Second Life has one of the best toolkits for creating interactive objects and virtual characters interacting with those objects. One of the intended contributions of the project is the creation of an agent toolkit for easy population of a given piece of land with virtual AI characters as well as facilities for training those characters by human experts. Having such a tool available will position Second Life as a leading platform offering artificial intelligence facilities in Virtual Worlds.

As the project is conducted by two non profit research organizations one of our key goals is to inform the community about the results of the project through publishing in leading journals and conferences. During the period of 2009-2011 we intend to produce 15-20 high quality scientific publications, which will provide a great possibility to promote Second Life amongst various research communities. We have recently submitted a proposal to demonstrate the first prototype at the AAMAS 2009 conference, which is the top event for AI researchers attended by thousands of key people working on Multiagent Systems.

Altogether, in its current form Second Life doesn’t offer facilities for incorporating artificial intelligence. A few attempts to have automated bots in Second Life can not be considered being proper AI and are usually focused on gaining unfair advantage over other users in exploiting of the Second Life economy. This project will not only provide a toolset for quick and easy development of AI driven characters, but will also highlight the positive effect AI may have on a Virtual World.

Finally, the Virtual Institutions technology that our research is based upon is not only useful for development of intelligent AI controlled agents, but also for structuring and regulating the interactions of humans users. There is currently no secure mechanism present in Second Life  (as well as in any other Virtual Worlds platform known to us) for structuring such user interactions. Gartner predicts that many commercial activities will migrate into non gaming Virtual Worlds and for Second Life to be the key technology in this respect it is important to improve the security of such activities. Having Virtual Institutions in place will prepare Second Life for the aforementioned challenges and will help to minimize the number of disputes arising between its users.

Artificial Intelligence

We expect our project to have a significant impact on research in Artificial Intelligence and Multiagent Systems (MAS) in particular. The majority of MAS researchers are either developing agent algorithms for form-based (2D) applications or apply their research to robotic agents. Only a few researchers use artificial intelligence for the agents immersed in 3D Virtual Worlds. One of the aims of the project is to highlight the high potential of 3D Virtual Worlds and Second Life in particular in the further development of artificial intelligence.

In contrast to robotic agents in 3D Virtual Worlds there is no need to solve pure engineering problems like object recognition, embodiment dissimilarity, believable locomotion, etc. that have become a primary concern of AI researchers working in robotics. Having the agent fully immersed in the environment and having a sophisticated environment formalization established via the application of Virtual Institutions technology offers unique possibilities to advance the intelligence of autonomous agents through such mechanisms as imitation learning and implicit training.

The ultimate goal of Artificial Intelligence is the development of intelligent agents. Having a general purpose agent development platform (which is one of the technological contributions of the project) will bring more AI researchers into Second Life.

Virtual Worlds

Non gaming Virtual Worlds are often referred to as Web 3.0 by many researchers and developers. Gartner predicts that 80% of the Internet users will be actively participating in such environments by 2011. It is believed that many online tasks that are currently being performed through form based interfaces will migrate into Virtual Worlds.

Unfortunately, up until now there wasn’t many practical applications of non gaming Virtual Worlds available on the market. We believe our project to be one of the pioneers highlighting the importance of non gaming Virtual Worlds.

Furthermore, the Virtual Institutions technology establishes a mechanism of creating Virtual Worlds with normative regulation of participant interactions. The development of the concept of Normative Virtual Worlds and the corresponding technology is an important research contribution of our project to the field of non gaming Virtual Worlds.

Education

Through our experience of working on this project we have learned that having virtual AI characters present in the environment is much more engaging than simply learning about the ancient site from 3D models, static pictures and text. We have also discovered that when it comes to recreating the history there is a lot of available data concerning the key events, key people and architecturally significant buildings. However, when it comes to the life of ordinary people there is not enough knowledge readily available. The existing history books are normally focused on the key events and not that much on ordinary people. We have also found it to be quite difficult to rely on the opinion of a single subject matter expert, simply because it is hard to find an expert that would know all the details about lives of ordinary people in a particular epoch. We discovered that while having multiple experts meeting together in a Virtual World, going through a scenario and reaching an agreement we were able to progress much faster. The experts themselves where positively surprised with the quality of experience Second Life provides for such meetings.

Although we haven’t yet conducted the experiments involving history students, we believe that their learning experience will also become much richer while interacting with virtual humans. Every new generation develops its own new educational practices. Some research studies conducted amongst the present day teenagers suggest that youngsters may gain more knowledge about history from a blockbuster than from a history book. One of the drawbacks of studying history in such a way is that students in this case absorb information in rather passive manner. In contrast, Virtual Worlds populated with virtual humans offer an engaging dynamic environment stimulating active student involvement into the education process.

Having students interacting in a Virtual World with AI controlled agents brings a number of important benefits into the education process. First of all, this form of education tends to be much more engaging than reading a history book. On the other hand, making a student use the textbook for the completion of the assigned task may also become a very good motivation for the student to read the book.

It is possible that the task assigned to a student only involves competition with autonomous agents (not other class mates). Moreover, such agents may be capable of adjusting the complexity of the task so that each individual student will be able to solve it. Eliminating direct competition between the classmates and adjusting the task complexity to individual needs may help students to gain self confidence and maintain a high level of interest in the subject.

When the students are completing their assignments while being fully immersed in a 3D Virtual World it is also much easier to document their progress and provide them with constructive feedback than it is in an ordinary classroom.

The aforementioned hypotheses, of course,  require further scientific evidence. One of the aims of the project is to collect such evidence and share the outcomes with the education community.