The keynote today started off with a conversation about metacognition among students, with emphasis on young learners, 3rd and 5th graders. In their case, the learning material was the Spanish language for English speakers. They found that children both 3rd and 5th graders were good at identifying things that they knew or didn’t know. Next the 5th graders studied topics where they experienced more difficulty, whereas 3rd graders didn’t seem to have a reason for choosing what to study. The idea is that 3rd graders use some other metric for deciding what to study, possibly they are basing their study choices on what topics are fun to study. Whereas 5th graders and adults, spend their time studying the topics that they don’t know.
Later, Vincent Aleven had a different finding, which is that for problem solving requiring multiple steps, young students are not skilled at identifying problems that they didn’t know and their meta-cognitive skills are not as tuned for these types of problems. Both of these findings make for an interesting conversation, and as with all research, when you find the answer to one question, you make two more.
Panel Talk – 10 Years On, is AIED Mainstream:
Turns out, no, not so much. The talk was nicely organized with a brief overview of all the venues which are using ITS software in their institutions. Next we had a wonderful list of grand challenges that have advanced learning contained as major focuses that each respective community would like to address. Afterwards there was discussion about expanding AIED from Artificial Intelligence in Education to something more encompassing, which I feel is the correct move. Proof of which can be seen by the broad range of paper titles. This may also explain why my own paper was accepted, considering it is very loosely, if at all related to AI, yet is quite relevant to the work being done by community.
Lastly, what I feel is the most important issue was addressed, which was how do we expand the presence of AIED and the technologies encompassed by the conference presentations truly into the mainstream. The goal to me seems to find a way to monetize and commercialize personal learning. Notably members of the audience put forth that perhaps public schools, with their limited resources, numerous rules and red tape, particularly with respect to resource accessibility, may not be the most ideal consumer for a commercial ITS package. It was pointed out that company training, and military use are two other potential options, where efficiency and cost are more in line with the product that we are capable of providing. Regardless monetizing and commercializing personalized learning and proving their success on a wide range, larger than select school systems – though that is impressive, will be necessary to quickly let the world know that our community is here. The keyword in that sentence being quickly, perhaps we are more like the AI community than we’d like, the AI crash of 90′s comes to mind, in which perhaps a quick recognition by the world is not, in the long term, in our best interest. Nevertheless, people and the world are calling for ITS and personalized learning, from Microsoft, to the National Academy of Engineering. So the iron is most certainly hot, and if the AIED community is capable of striking, now would certainly seem like a good time. At the very least, I would like to see someone stand tall and take a swing.
Serious Games and Intelligent Tutoring Systems:
In the afternoon I stayed for an interesting talk from one of the students from NC State, James (Jim) Thomas. In this particular case he had built a substantially large system for identifying plans of students in an open ended environment. In this case, students have X number of tasks to complete within some 3D environment and his system, Annie, in a domain independent way develops its own understanding of the world and can hint the user on what to do next according to the various pre and post conditions of the tasks remaining. Although that is the focus of the work, another interest is to show that the game he developed as the test bed for his system, also provides learning gains to students which at this point has yet to happen. Luckily, from our experiences at Charlotte developing serious games, I was able to provide some meaningful and helpful suggestions. Sadly, many of those suggestions are ideas written in a yet to be accepted paper that Mike has put together. From these experiences it is still clear to me that people developing educational games, could still use a best practices advice paper for designing a game with the idea of maximizing ones potential for obtaining learning gains. After your domain is selected, some of those guidelines include:
1. Sit down with a set of students who are the intended audience and tutor them on the subject with pen, paper and white board. Paying particular attention to the areas which cause confusion in the students.
2. Next consider the interaction or simulation which captures exactly the explanation that, you as the tutor, explained to your audience.
3. Third, and this is potentially the most important, write, re-write, design and consider deeply what are the pre and post test questions you will use to test your learning gains. The test questions are your means for measuring, and just like when measuring distances, you want a device that always shows the same length for the same distances. A ruler made of yogurt isn’t particularly helpful. Your pre and post test questions are the basis for the game design.
4. Lastly, the mechanic for your game should be as closely matched to the concept your are trying to teach as is possible to make. If you are making a game to teach chess, applying the rules of chess should be the focus of the game and the way the player advances. A how to play chess game doesn’t need any open world 3D exploratory environment. Anything that isn’t precisely the concept that you wish to teach is only a distraction. In addition, you’re trying to teach how to play chess, not play chess, so just a chess game is not sufficient. An appropriate game would be where players see a series of different moves and have to select the appropriate applicable rule. For example, you get a piece and you have to choose from multiple options which movement behavior is correct or incorrect.
My thanks to James Thomas for the interesting and fun conversation.
As per request, here is a photo. From the conference, Ryan Baker gives his talk on Towards Predicting Future Transfer of Learning.
For more images and just general information about my trip and time in New Zealand, you can read my separate blog entry here.