Deep Learning
The definition of deep learning can be described not by what is, but what is not. Deep learning is not instructionist in nature, which focuses learning isolated facts about the world and procedures for solving problems. Deep learning promotes the application of facts and procedures in order to develop a deeper conceptual understanding. Deep learning does not assume the learner is a clean slate, and builds upon existing knowledge, experiences, and possible misconceptions. It does not ask learners to accept a given learning process as is, but instead asks them to understand and reflect on it as it applies to them personally. It does not simply focus on teaching, but instead on the development of learning. It does not look at knowledge as a static mental structure but a process involving a person, her environment, and the situations where her knowledge is being applied.
Human Deep Learning vs. Computer Deep Learning
By 2000, there were still no studies that could show that computers aided student performance in any way [1]. However, more recent research on the expansion of learning sciences and educational technology shows the success of computer-assisted learning is improving. The terms AI, machine learning, and deep learning get used quite synonymously, but should not be confused. Artificial intelligence (AI) can be thought of as the most narrow view of intelligence. It is instructionist in nature, operating only with facts and assigned procedures. However, add machine learning to AI, and the result is something much more complex. Machine learning takes facts and procedures, gives them decision trees, reinforcement logic, etc. and is able to apply them to specific situations. Lastly, deep learning solves real world problems using neural networks. Machines can break down subsets of machine learning tasks and assess their application, effectiveness and accuracy. The results get better with more data, larger models and more computation happening. This is not so different from how humans using deep learning.
Sociocultural Learning
A large part of the learning process is sociocultural. Almost all learning outside a classroom environment occurs in some kind of complex and collaborative environment. More and more online tools offer a means to support group work and learning. For example, Google Docs facilitates collaboration with documents that can be worked on from any remote location, from any device type, and by an infinite amount of group members. Tracking changes, group chat, and commenting options allow for reflection of the learning process. Other online learning tools, like Edmodo, are tailored towards children. Children are known to think different than adults; they are able to retain information better in order to generalize it to a broader range of contexts. Edmodo encourages dialogue both inside and outside of the classroom and features learning communities. Other Internet-based learning tools, such as Treehouse and Coursera, allow learners to articulate their knowledge by intermittently giving quizzes during course lectures. This gives learners an opportunity to apply developing knowledge in a visual or verbal way.
The Future of AI and Deep Learning
According to Sawyer, AI can support learners by taking more of a facilitating role in helping them have the kind of experiences that lead to deep learning. This can be accomplished if AI does the following:
- Represents abstract knowledge in concrete form
- Contains tools that allow learners to articulate their developing knowledge in a visual and verbal way
- Allow learners to hone their developing knowledge via UI tools, in a complex process of design that supports simultaneous articulation, reflection, and learning
- Supports reflection in a combination of visual and verbal modes
- Include networks of learners that can share and combine their developing understandings and benefit from the power of collaborative learning
References
Copeland, M. (2017, February 09). The Difference Between AI, Machine Learning, and Deep Learning? | NVIDIA Blog. Retrieved June 21, 2017, from https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
Sawyer, R. K. (2008). Optimising Learning Implications of Learning Sciences Research. Innovating to Learn, Learning to Innovate, 45-65. doi:10.1787/9789264047983-4-en