[Xuedong Huang] Empowering the Digital Transformation of Education with Artificial Intelligence

During the peak of the COVID-19 pandemic, I witnessed the ways in which artificial intelligence (AI) could transform the future of education for both teachers and students. 

In this video, a young student independently read out loud into a web cam from a teacher-supplied assignment and received feedback on ways to improve their reading fluency. It's the type of remote learning scenario that has become common during the pandemic. 

But, what's uncommon in this interaction is how the assignment is assessed and how the student receives feedback, both of which occur in an application powered by artificial intelligence. 

AI saves teachers time as it automatically assesses the speed and accuracy of the student’s speech against the reading assignment, identifying specific errors like mispronunciations and omissions. 

Teachers are in control as they can review the assessment before it’s shared with the student. 

Students benefit too. Augmenting education with AI can help students who need more personalized attention, who have learning disabilities, or who want additional instruction to get the support they want. 

Educators often have classes of more than 20 students and can take advantage of AI to help automate tasks, thus allowing them to focus more deeply on other aspects of classroom instruction. 

It's a future where education is enhanced by artificial intelligence to provide a more personalized experience that is available anywhere, anytime and in any language.

In 2020, the United States and many other countries unexpectedly shifted to mandatory remote learning during the ascent of the COVID-19 pandemic. 

The shift was difficult for students, educators and families alike, as the support required to ensure a quality and equitable education experience was not universally available. 

Educators were forced to quickly find solutions. As a result, today across many countries, a byproduct of mandatory remote learning has been significant increases in students with laptops, improved access to the internet, and familiarity with videotelephony applications like Microsoft Teams and Zoom. 

The following are just three examples of AI augmented tools that educators have utilized to make remote learning better. 

First, the example I referenced of the student practicing reading is enabled through Reading Progress in Microsoft Teams. Educators can upload single or multiple assignments that are easily accessible and customizable for individual students’ reading level and progress. 

Students can read at their own pace and in a comfortable setting, removing any stigma, stress, or distraction associated with reading out loud. The app has streamlined a typically time-intensive process by using AI to automate speech recognition and assess the speed and accuracy of the student’s recorded reading. 

Errors such as mispronunciations, repetitions, phrasing, intonation, and omissions are automatically identified and used to customize future instruction for each student. Teachers control the experience and can review and share assessments when appropriate. 

Another great example is GE Chinese School in Bellevue, Washington. They used Azure Cognitive Services for Speech to create a targeted language learning application to help children learn Chinese more effectively. 

With real-time instructive feedback to students, AI helps improve pronunciation and enhance long-term memory for improved language learning outcome. 

Secondly, assistance with reading and writing, especially for people with learning disorders such as dyslexia and dysgraphia, is enhanced and enabled through Immersive Reader, the main feature in Microsoft Learning Tools. 

Students can have content read aloud to them and teachers can customize the delivery of assignments for each student by changing text size and background color, breaking words into syllables, highlighting one or more lines of text, and more. 

Educators report that struggling readers are reading more quickly and accurately, and stronger readers can focus on inferencing and higher-order thinking. Immersive Reader uses AI to read text out loud, translate between languages for non-native speakers, and extract text from images like PDFs.  

Thirdly, improving accessibility helps to create a more inclusive education environment. Speech and language AI are used in real-time to automatically generate captions or subtitles translated to another language. These capabilities are broadly available today in Microsoft PowerPoint’s Slide Show and in Microsoft Teams calls to improve accessibility, inclusiveness, and learning.

Fortunately, the time is right for the use of AI in education as the availability of massive amounts of relevant, quality data, access to unlimited cloud computing power, and advancements in deep learning modeling have resulted in unprecedented improvements in AI quality. 

State-of-the-art AI like Microsoft Cognitive Services demonstrate such high quality for tasks like speech recognition, language understanding and translation, object recognition and image captioning. 

AI has surpassed human capability in some of these open research benchmarks over the past decade. 

AI-augmented education is not a replacement for in-person or one-to-one interactions with teachers. Remote learning during the pandemic has demonstrated that students need in-person and direct interaction with teachers and peers to stave away feelings of isolation. 

In the example I shared in the beginning, the Reading Progress feature in Microsoft Teams brings the student and teacher together in a natural interaction that is enabled by technology. 

It allows the student to complete the assignment on their own time and assists the teacher by providing time-saving AI-based proofing tools to help lessen higher workloads. 

Teachers and parents can provide the holistic support that comes from understanding the full context of a student’s needs and academic experiences. 

AI can be a valuable collaborative partner in those efforts.   

It’s important to note that new digital education experiences won’t be accessible to those who need them most without closing the digital divide. People with lower incomes or who live in rural areas continue to struggle with access. 

John Roese, Global Chief Technology Officer from Dell Technologies outlined the global access situation well in their column from Davos 2021, “COVID-19 exposed the digital divide. Here's how we can close it.”
AI-augmented instruction on its own is not a panacea for improving education. 

Rather, it's a part of a broader digital transformation of education that has been accelerated by the COVID-19 pandemic and that will achieve its full potential through a partnership among educators, technology developers, and governments. 

Governments and companies need to continue making fast, reliable internet available to everyone, especially those in poor or rural areas. Developers can create new AI augmented education experiences using readily available AI from cloud providers. 

As the global community continues to focus on access and development, the future is bright for AI-enhanced education.

By Dr. Xuedong Huang

►Doctor Xuedong Huang is a leading computer scientist and technology executive in AI and Spoken Language Processing. He is currently a Microsoft Technical Fellow and Chief Technology Officer overseeing Microsoft Azure AI engineering and research, covering Microsoft’s AI API services in Speech, Computer Vision, Natural Language, and Decision. 

Since 2020, he has been serving as one of Prize Council members of VinFuture Prize - the first global science and technology prize from Vietnam, and one of the world's largest annual prizes in science and technology.