Article Review: Stupid Tutoring Systems, Intelligent Humans

Citation

Baker, R. S. (2016). Stupid Tutoring Systems, Intelligent Humans.International Journal of Artificial Intelligence in Education, 1-15.

Google Scholar Link

Summary

Purpose of the research

In this article Ryan Baker discusses the potential of educational data mining (EDM) for driving human decision-making surrounding learning. He suggests an alternate paradigm that focuses on intelligence amplification (amplification of the teacher or learner) rather than artificial intelligence or intelligent tutoring systems.

Research questions

Although not explicitly stated, the overarching researcher’s question is what is the future and potential for EDM and LA?   

Methods

The author uses a literature review and insight to present a alternative paradigm to EDM and LA.

Important note: I reviewed this article due to the significance of Dr. Bakers standing in the EDM and LA research community. Although this article does not provide any new data or evidence, I consider Dr. Baker’s opinion and perspective to be increasingly important. His perspective on the direction for LA research, coming from a deep understanding of EDM is significant and important. 

Subjects (describe subject – including number of)

N/A

Results

Dr. Baker suggests that despite the promise of intelligent tutoring systems to create complex learning interactions, “the learning tutoring systems used at scale today are much simpler” and are not delivering on the larger promise (p. 1). He states that, “… there is a disconnect between the vision of what intelligent tutoring systems could be, and what they are; a disconnect the most impressive examples of what intelligent tutors can do, and what current systems used at scale do” (p. 2).  To expand our understanding, he explains that most tutoring systems today rely on simple methods to extract student behavior such as having students get a concept right three times in a row.

Because the development of rich intelligent tutoring systems have been slow and less that desirable, he poses an alternative. He states that, “Perhaps we do not need intelligent tutoring systems. Perhaps instead what we need, what we are already developing, is stupid tutoring systems. Tutors that do not, themselves, behave very intelligently. But tutors that are designed intelligently, and that leverage human intelligence” (p. 3).

He goes onto site several examples of this type of ecosystem where learning tools give instructors ongoing distillation of student activities allowing the teacher take real time actions to improve the quality of the classroom experience. For example, “Reasoning Mind, uses reports in real-time, obtaining information that a student is struggling with a specific concept right now, and engaging in proactive remediation” (p. 7).

He also mentions ASSISTments stating that, “teachers often read the reports of the previous night’s homework before class, and re-work their planned lecture base on data about what questions students struggle with” (p. 7).

Discussion

Implications

I believe that this alternative paradigm is vital and important for the overall direction of Technology Enhanced Learning (TEL). I concur with Dr. Baker that hybrid systems that enhance teacher intelligence through identification and reports is important. As he states, automated interventions (e.g., ALEKS) are time-consuming and expensive to author. In addition they are brittle, for example “an encouraging message may not be encouraging the 12^th time” (p. 8). Students adapt quickly and will figure out how to defeat the system.

This is why I believe that our work on Math 160 is relevant. We are not attempting to create an automated tutoring system. In fact, we are keeping the human (Ben, Gabby, and Jessica) at the center of the intervention. We are attempting to use our technical system to identify study behaviors that are not engaged, or less than desirable. Instead of forcing students to follow a similar learning path (e.g., getting three right in a row), one that can then become burdensome or tiresome, we are detecting behaviors and allowing intelligent humans to do what they do best, adaptation.

As Dr. Baker suggested, “Humans are flexible and intelligent. Humans can’t sift through large amounts of information quickly, which is why then need data mining and reporting to inform them. But once informed, a human can respond effectively” (p. 9).  The unique ability that humans possess for adaptation and change should be leveraged to enhance learning, in my estimation.  We are looking for the sweet spot where human ingenuity and computer computation work synergistically to enhance learning and teaching.

Contribution and future research

Future research focus

Future research should focus on a learning ecosystem approach, where we leverage what humans do best (adaptation, ingenuity) and we leverage what computers do best (analysis, pattern recognition), bringing those strengths together to enhance learning.

Implications for the technology enhanced learning (TEL) environment

(see discussion above)