We’ve explored three real-life examples to see what types of artificial intelligence are changing the education industry and in what way.
This article is neither a philosophical essay on the role of AI in the contemporary world nor a horror describing that AI will soon replace us all. Here, we analyze real-life examples from the education industry to see the different types of artificial intelligence in action and evaluate the effect of adopting AI in education.
Jill Watson, a virtual teaching assistant
While delivering a massive open online course, Georgia Institute of Technology found it challenging to provide high-quality learning assistance to the course students. With about 500 students enrolled, a teaching assistant wasn’t able to answer the heaps of messages that the students sent. And without personalized assistance, many students soon lost the feeling of involvement and dropped out of the course. To provide personal attention at scale and prevent students dropping out, Georgia Tech decided to introduce a virtual teaching assistant.
Jill Watson (that’s the assistant’s name) is a chatbot intended to reply to a variety of predictable questions (for example, about the formatting of the assignments and the possibility to resubmit the assignments). Jill was trained on a comprehensive database consisting of the students’ questions about the course, introduction emails and the corresponding answers that the teaching staff had provided.
Initially, the relevance of Jill’s answers was checked by a human. Soon, Jill started to automatically reply to the students’ introductions and repeated questions without any backup. When Jill receives a message, ‘she’ maps it to the relevant question-answer pair from the training database and retrieves an associated answer.
AI type used: Being a chatbot, Jill represents interactive AI – ‘she’ automates communication and without compromising on interactivity.
Third Space Learning, an online learning platform
While giving one-to-one math lessons to 3,500 pupils weekly, Third Space Learning was looking to improve the learners’ engagement and identify best practices in teaching. To achieve that, they have applied AI to analyze the recorded lessons and identify the patterns in the teachers’ and pupils’ behavior. For example, AI can identify if a pupil is showing signs that correspond to the ‘losing interest’ pattern. In the future, Third Space Learning plans to provide their tutors with real-time AI-powered feedback during each lesson. For example, if a tutor talks too fast, AI will advise them to slow down.
AI type used: Third Space Learning’s AI (with both its current and future functionality) looks exactly like analytic AI, which is focused on revealing patterns in data and producing recommendations based on the findings.
Duolingo, a language-learning platform
Among the three use cases that we are considering, Duolingo appears to be an absolute champion in terms of the number of challenges solved with the help of AI. For example, when many users felt so discouraged with being offered too simple learning materials that they dropped out of the course immediately, Duolingo introduced an AI-powered placement test. Being computer-adaptive, the test adjusts the questions to the previously given answers, generating a simpler question if a user made a mistake and a more complex question if the user answered correctly. The complexity of the words and the grammar used also influence the test configuration.
Besides, Duolingo uses AI to optimize and personalize lessons. For that, they have developed a ‘half-life regression model’, which analyzes the error patterns that millions of language learners make while practicing newly learned words, to predict how soon a user will forget a word. The model also takes into account words’ complexity. These insights allow identifying the right time when a user should practice the word. Duolingo says that they have seen a 12% boost in user engagement after putting the model in production.
With the same purpose of boosting user engagement, Duolingo tried bots to help learners practice the language. Available 24/7, the bots readily communicated with the users, as well as shared their feedback on a better version of the user’s answer. Besides, the bots contained a ‘Help me reply’ button for those who experienced difficulties with finding the right word or applying the right grammar rule. Though currently unavailable, the bots will reappear (at least the official message from Duolingo’s help center leaves no doubt about this).
AI type used: Analytic AI (the placement test and the prediction model), interactive AI (bots).
Two sentence-long findings
The examples we considered show that AI positively affects the education industry, allowing its adopters to solve such challenges as bringing personal attention at scale, improving students’ performance and engagement, identifying teaching best practices, and reducing teachers’ workload. And as we see, to solve these challenges, the industry players resort to analytic and interactive AI.
About the Author
Irene Mikhailouskaya (Makaranka) is a Data Analytics Researcher at ScienceSoft, an IT consulting and software development company headquartered in McKinney, Texas. With a focus on business intelligence, big data and data science, Irene explores trends, technologies, key challenges and solutions in the world of data analytics.
This is a companion discussion topic for the original entry at https://blog.datasciencedojo.com/ai-transforming-education-examples/