ASPPH logo




Innovations in Pedagogy Using Artificial Intelligence & Natural Language Processing

Exploring the Possibilities: Ask Jetson

ASPPH introduced Ask Jetson to attendees at the 2017 ASPPH Annual Meeting. Intended as a showcase for Artificial Intelligence (AI) and Natural Language Processing (NLP), Ask Jetson provides answers to a variety of questions, from the latest CDC flu surveillance data, to where the closest coffee shop could be found. It even proctors sample Certified in Public Health (CPH) practice exams!

We created the Ask Jetson project during an ASPPH Innovation Lab Autonomy Happy Hour. ASPPH Innovation Lab events provide a venue for staff to experiment with new and emerging technologies and discover ways to apply those technologies to further ASPPH’s mission and work.  The Ask Jetson app leverages Amazon’s Alexa skills–built with Lambda functions–to automate our cloud infrastructure, surface relevant information in a timely and accessible manner, and proactively alert us of actionable intelligence.

While AI is a useful back-end tool for ASPPH’s IT enterprise, the impact and application for ASPPH members are profound.  AI is poised to advance interactive and adaptive learning environments by:

Natural language as a user interface will fundamentally change how both teachers and learners interact with digital content and provide new opportunities in pedagogy.

This article introduces some of the more advanced AI technologies and the implications for public health educators.

AI in the Higher-Ed Classroom

AI is bringing education into the cognitive era. The educational landscape as well as the learning experience are being transformed through new levels of personalization. Cognitive devices that can now understand, learn, and reason allow educators to cut across all learner skill levels, gaining insights into learning styles, preferences, and aptitude of individual students. The results are holistic learning paths, for every learner, through their lifelong education journey.

By individualizing a student’s path to content mastery through adaptive learning or competency-based education delivery using AI, educators will be able to assess learning outcomes at a more granular level, including assessment of skills such as systems thinking, collaboration, and problem solving in the context of deep, authentic subject-area knowledge.  Educators will be able to diagnose in near-real-time, the learning needs or course trouble spots faster and more in-depth, and can provide targeted interventions to improve student success and reduce overall costs to students and institutions.  AI will also offer course designers the opportunity to use more game-based environments for learning and assessment, where learning is situated in complex information and decision-making situations.

Similar to assignments and assessments in courses, AI has several potential roles:

  1. To serve as an ongoing means of monitoring what students know. This allows instructors and students to tailor teaching and learning to problematic areas.
  2. To serve as the principal means by which students learn new information. In many subjects, most in-depth learning happens through assignments in which students manipulate, derive, or construct knowledge.
  3. To serve as a key component of grading. Grading has multiple goals, from certifying students’ accomplishments to providing motivation for desired behaviors by students.
  4. To serve as a summative assessment of students, teachers, institutions, and courses. Summative assessment has many high-stakes goals, such as student certification and school accreditation.

Beyond Jeopardy: IBM Watson

One of the most famous examples of artificial intelligence in pop culture is IBM Watson and its 2011 win against two of Jeopardy’s greatest champions. Six years after its widely-hyped TV appearance, Watson has quietly matured into a real-world working stiff, and is incorporated into about 17 different industries, including healthcare, and education. IBM, along with other AI vendors, is already targeting the education space with cognitive solutions. IBM Watson offers three products that are most relevant to our members:

  1. Engage during class: Provide insightful information on how a class is doing overall and gives teachers a tool that encourages meaningful interactions with their students during class.
  2. Get student insights: Give teachers a centralized place to add and see helpful information about each student.
  3. Track academic progress: Work with an easy-to-use tracker to help teachers evaluate student progress compared to learning standards.
  1. Understand the class: Teachers can begin the school year with an understanding of their class by accessing information about their students from a single source.
  2. Get actionable insights: Teachers can optimize their time and impact throughout the year using actionable, on-demand insights about their students.
  3. Target learning experiences: Teachers can craft targeted learning experiences on-the-fly from content they trust.

Big Data & AI

How do big data and AI intersect? Well, one could argue that big data is the other side of the same AI coin. Data has always been an important aspect of business. Whether to make decisions or to analyze the past, data is required.  Big data can be described as huge chunks of structured and unstructured data. Being able to correlate these divergent datasets comes in handy when business leaders need to make crucial decisions for expansion, cost savings, or competitive advantage. Leaders typically rely on a group of trained professionals to extract useful meaning out of Big Data Analytics, which can impact the business in a positive manner. The insights incurred by a Big Data professional are extremely beneficial for businesses to make significant, informed and strategic decisions.

In the simplest of terms, big data helps AI learn and become even smarter.  Data is the input by which every cognitive engine algorithm is formulated, and subsequently, how its predictions are validated. When a cognitive engine acquires a critical mass of data and achieves a high reliability of predictability, it can in turn identify the data points that are most likely to violate and therefor require modification to its algorithm. This perpetual feed-back loop is what can propel AI to truly human-level cognitive ability.

In education, intelligence provided by large datasets can help identify algorithms for the best learning pathways and most suitable content to improve learning outcomes. The use of big data has become well established in business, entertainment, science, technology, and engineering. Whereas big data is beginning to be utilized for decision making in higher education as well, practical applications in higher education instruction remain rare. Beyond the potential to enhance student outcomes through just-in-time, diagnostic data that is formative for learning and instruction, the evolution of higher education practice overall could be substantially enhanced through more research and experimentation.

The toughest challenge in data-intensive research and analysis in higher education is to find the means to extract knowledge from the extremely rich data sets being generated today and to distill this into usable information for students, instructors, and the public.

Surfacing Actionable Data: Salesforce Einstein

Data science is an increasingly in demand, yet rare, skill. Data scientists fetch premium salaries and are an expensive resource to keep on staff. Here too, AI can help lower the cost for these skills and make data science accessible to more organizations.

Salesforce Einstein is AI built into the core of the Salesforce Platform, where it powers the world’s smartest CRM. It delivers advanced AI capabilities to sales, service, and marketing — and enables anyone to use clicks or code to build AI-powered apps that get smarter with every interaction. While traditionally a for-profit sales tool, Salesforce has recently diversified to include non-profit- and education-focused offerings with its initiative.

With this tool, anyone in any role and industry can use AI to perform better. Einstein is like having one’s own data scientist dedicated to bringing AI to every customer relationship. It learns from all available data — CRM data, email, calendar, social, ERP, and Internet of Things (IoT) — and delivers predictions and recommendations in context of one’s goals. In some cases, it even automates scheduled tasks, so smarter decisions can be acted upon with confidence and more attention focused on the customers at every touchpoint.

A Treasure-trove of Data: MOOCs & AI

Massive open online courses (MOOCs) are a great source of actionable learning data. A MOOC is a model for delivering learning content online to any person who wants to take a course, with no limit on attendance. The term has been used to define an assortment of methods, rationales and approaches for offering and delivering online learning experiences on a very large scale. The delivery has been through different services and centralized platforms with varying degrees of success. The services include an aggregation of blog sites, social media feeds and learning management systems (LMSs). MOOCs have been designed to support professional development, academic scholarship, university curricula, community outreach, and corporate training applications.

MOOCs illustrate the many types of big data that can be collected in learning environments. Large amounts of data can be gathered not only across many learners (broad between-learner data) but also about individual learner experiences (deep within-learner data). Data from MOOCs includes longitudinal data (dozens of courses from individual students over many years), rich social interactions (e.g., videos of group problem-solving over videoconference), and detailed data about specific activities (e.g., watching various segments of a video, individual actions in an educational game, or individual actions in problem solving). The depth of the data is determined not only by the raw amount of data on a learner but also by the availability of contextual information.

AI can offer individualized learning at MOOC scale. While MOOC are a relatively new phenomenon, two quite different philosophical and practical approaches have emerged (xMOOCs and cMOOCs) that may require different approaches to and application of AI:

Meaningful Conversations

Language is the most natural interface for human-to-human interactions, but only recently has the technology evolved as a practical interface for human-to-computer interactions.  As anyone who has experienced the frustration of an automated phone attendant knows, computers are not the most skilled at understanding the nuances of human speech. This is where natural language processing (NLP) comes in.

NLP learns with every interaction to listen better and to respond better. Cognitive technologies have broadened the power of existing IT platforms to automate tasks traditionally performed by humans such as call routing, accessing account status, performing transactions, and responding to queries. Such technologies help organizations improve the quality of services, reduce response time for customers, and reduce costs.  When combined with conversational language, cognitive technologies can provide a highly personal and engaging experience on a large scale. Conversational user interfaces, either text-based (bots) or auditory (virtual assistants), are now common as front-line support, help, and tutorial resources.

A bot, at its most basic, is a piece of software that performs an automated task, be it finding an awesome GIF, ordering toilet paper, or downloading a file. Bots are great at making sense out of lots of different types of information (schedules, meeting notes, documents, notifications from other business applications), and making all that data more useful by allowing people to interact with it like they would in a conversation with a person.

Bots range from the obvious—bots for recognizing good work, posting photos, translating text—to the utterly frivolous, like playing poker. There’s one to notify you each time your institution is mentioned somewhere online, streamlining that whole “wasting time on the Internet” thing. They absolutely can save you time.

Imagine an Office Hours bot, that has learned and responds to the most commonly asked course-related questions gathered over millions of interactions with students. Talking to students/faculties, seeking their help, or simply chatting with them over mundane tasks provides opportunities for insight and inspiration while saving the professor valuable time.

Virtual assistants take the interactive experience of a bot even further. Interacting with a virtual assistant is now an every-day experience and you are likely to have engaged with at least one of the common engines from Amazon, Apple, and Microsoft.

Alexa, Siri, & Cortana:

  Alexa Siri Cortana
Manufacturer Amazon Apple Microsoft
Origins Alexa is the name Amazon gave the digital assistant living inside its Echohome device, which started selling widely in June 2015. The first virtual personal assistant to enter the mainstream, debuting on the iPhone in 2011. Cortana was introduced in 2014, named after the AI character in the hit Xbox game “Halo.”
Functions Alexa began with limited uses, most often responding from the in-home speaker to requests for things like weather and news. Alexa can also find a radio station from TuneIn. It’s an Amazon pitchwoman through and through, so it will take commands like remembering your shopping list. Only widely opened to third-party apps in 2016 with the arrival of iOS 10. That’s allowed developers outside Apple to give it capabilities such as hailing an Uber if you simply ask for one, which so far is the pinnacle of achievement for most consumer-facing AI. Microsoft calls Cortana a “digital agent.” It can handle basics like controlling calendars, getting weather and taking dictation on an email.
Alexa isn’t in your phone; it’s stuck inside the Echo or the newer Dot device. Amazon just made it available in its tablets. Criticized early on for poor speech recognition because Apple was behind rivals in “neural network” technology—that’s nerd-speak for computers that act more like human brains—however Apple improved this technology in 2016. Like all AI-powered assistants, Cortana is only as smart as its programming, and Cortana gets much of its information from Microsoft’s Bing. It does not use Google, the leading search engine.
Upgrades In 2016, Alexa connected with Amazon’s TV device to control streaming video. It also made the leap from simply taking dictation around shopping lists to actually placing the orders. Alexa also has a program to control the home environment. Tinkerers have modified the device to perform functions like starting their Tesla cars remotely or automating their cloud infrastructure (Ask Jetson). Siri is getting smarter with Apple TV. A viewer can ask Siri to pull up a livestream from inside Apple TV apps. It also is embedded into Apple CarPlay for hands-free control of key systems like navigation, music and messaging. And Siri is gaining control of homes through Apple HomeKit. Microsoft has built Cortana into its Edge web browser, which means it’s there to help complete online tasks such as making reservations or looking for discounts while shopping. Cortana also works with Google Android and Apple iOS devices. Microsoft wants Cortana to be everywhere, including of course its own ecosystem, from the Xbox to Skype to LinkedIn.

What’s Next?

Hopefully, this article on advanced AI technologies and the implications for educators has provided you with some food for thought.  While these technologies are new and ever-evolving, ASPPH members will soon see practical examples in pedagogy as more and more educators incorporate AI into the classroom. Are you using AI at your institution?  If so, let us know.