AI in Education: Building Skills for the Future of Work

In 2013, two researchers from Oxford, Carl Benedikt Frey and Michael A. Osborne, predicted that within the next decade or two, nearly half of the jobs in the U.S. are at risk of being automated. The researchers compiled 702 occupations in the U.S. and got a group of machine learning experts to estimate how susceptible these occupations were to computerization (Frey & Osborne, 2013).

A later research by the McKinsey Global Institute used a different approach and arrived at an estimate close to Frey and Osborne’s findings. The McKinsey researchers projected that around 50% of current work activities have the potential to be automated (Manyika et al., 2019). This research however employed a task-based approach by considering how many parts of a job can be automated instead of just looking at the job as a whole.

One thing is evident from both these research and it is that we are seeing a new emerging gatekeeper to the economy (Mazumdar, 2020) – Artificial Intelligence (AI). In a not so distant future, AI will be the technology used to decide who gets a loan or a job.

If the purpose of education is to develop, prepare, and equip students for a rapidly transforming world so they can thrive in society (Sloan, 2012), we need to respond to the changes brought about by AI technology to prepare our children for the future. What does the future of learning with AI look like and what skills might we inculcate in our students today to enable and prepare them for the future of work? This article aims to investigate and explore these questions, but first, let’s take stock of the current state of AI and determine the phases involved.

The Four Waves of AI

According to Kai-Fu Lee, the revolution of AI will happen over four waves – “internet AI, business AI, perception AI, and autonomous AI” (Lee, 2018). The third wave of perception AI is already upon us but it is the fourth wave – autonomous AI – that will have the biggest impact on the world.

First Wave: Internet AI

The first wave AI has already been around since almost 15 years ago. Most of us will recognize this first wave of Internet AI in the form of recommendation engines – all those suggestions you get just because you browsed a particular item on an online shopping or media site. That’s the algorithms of internet AI at work through machine learning. The trend we are seeing with a rise in AI-driven internet companies like ByteDance (owner of Tik Tok) only serves to illustrate the opportunities opened up by first wave AI.

Second Wave: Business AI

Then comes the second wave AI – business AI. Businesses have been collecting and labeling data for a long time. With every action from issuing loans to recording health diagnostics, data is being recorded, labeled, and categorized which then enables business AI algorithms to study all these data points to find correlations and make predictions. Industries with large amounts of “structured data”, meaning data that has been labeled, classified, hence searchable, have a high likelihood of being automated. Take for instance deep-learning algorithms doing a much more efficient job of deciding the credit-worthiness of a loan applicant than your banker. There is also an opportunity in the future for business AI to make more accurate medical diagnoses than your human doctor. Medical diagnostics relies heavily on data collection through identifying symptoms and looking at medical history and then predicting and finding the corresponding diagnosis based on the data collected.

Third Wave: Perception AI

Meanwhile, third wave AI – perception AI, is about giving eyes and ears to machines, bringing the online and offline world closer. Smart devices and sensors are helping to turn our physical world into digital data that can then be interpreted by machines. Deep-learning algorithms in perception AI can analyze and recognize photos, videos, and audio files, deriving meaning the way our brain does. Kai-Fu Lee posits that as perception AI matures, we will get to a point where the boundary between online and offline gets so blurred that the two environments become blended. He calls these new blended environments OMO: online-merge-offline. Perception AI also opens up new opportunities in education, from how instruction takes place to how homework, tests and assessments are done. With perception AI, more personalized learning will be possible to cater to each students’ learning pace and abilities.

Fourth Wave: Autonomous AI

Finally, comes the fourth wave AI – autonomous AI. When the powers of the preceding three waves come together, giving sensory powers to structured data, that’s when we begin to see the powers of autonomous AI revolutionizing our daily lives. There is a distinction between automation and autonomous. Automation affords the ability to repeat actions and tasks of routine jobs but not make judgments and improvisations depending on dynamic outcomes. Autonomous driving technology is an example of fourth wave AI and this technology is still in its infancy stage. In order to reach maturity and the level of safety required for wide acceptance, autonomous AI will need to be fed more data, and I mean a lot more data – data from billions of hours of driving, from perceiving millions of objects on the roads and under different driving conditions. The more it learns, the better this technology gets.

The Future of Learning with AI

Now that we have looked at the development of AI in general, let’s narrow down the scope to look at how AI will be impacting the field of education specifically.

What powers AI is a combination of the ability for machines to perceive and understand the world and the algorithms to process and make sense of that knowledge. Luckin, Holmes, Griffiths, and Forcier posits that there are three key models at the foundation of AI in education (AIEd):

1. The pedagogical model – covers the teaching approach and is concerned with finding the most effective teaching methods, for example assessments to measure learning progress and action-triggered feedback to improve learning.
2. The domain model – covers knowledge of the subject being learned, for example
3. The learner model – covers knowledge about the students, for example the learning history and emotional state of the student.

Another dimension being researched that will enrich the above models are models that consider the social, emotional, and meta-cognitive aspects of learning (Luckin et al., 2016). Taking into account the social-emotional state of learners as well as the awareness of their own thinking and the ability to regulate and control the thinking, this provides for a potentially very powerful accelerator to AIEd. These models have given rise to a few key trends in AIEd.

Personalized learning

A very common problem faced in traditional classrooms is that struggling students are often left falling behind, unable to catch up with their peers. At the other end of the spectrum are the advanced learners who get bored from the lack of challenge after mastering lessons ahead of their peers. Adaptive learning solutions using AI technology enables the creation of personalized learning paths tailored for individual learner’s pace and abilities. With artificial intelligence, algorithms can be developed to detect mastery of learning goals at different learning milestones. Data from sentiment analysis, eye to mouse movements can also be collected during the learning process to inform us about the student’s learning disposition, confidence level, mindset, and cognitive ability (Bughin et al., 2017).

Autonomous learning

The concept of autonomous learning (Pant, 2020) takes personalized learning to the next level. Autonomous learners have the freedom and independence to direct their own learning path by first identifying their strengths and personal interests, then personalizing their own learning program based on those factors. They then monitor their learning progress and performance. A day in the life of an autonomous learner might look like this: A student who is learning about chemistry puts on a mixed-reality smartglasses and “arrives” at his “virtual lab” where his AI tutor is waiting to walk him through the lesson modules with live simulated experiments. The experiments are conducted in the safety of a virtual environment while allowing the student to experience the different chemical reactions in real-time. This draws upon the concept of using “intelligent virtual reality to support learning in authentic environments” (Luckin et al., 2016). The student proceeds to meet with his peers in a virtual environment where discussions take place and they are given problems to solve as a group. Using AI technology, the virtual tutor is able to assess their solutions and understanding of the topic. The learning carries on even after the student arrives home, as his AI language tutor practices his Spanish speaking skills with him. In response to those who wonder what is to happen to the human teacher as autonomous learning with AI matures, Pant (2020) succinctly puts it this way, “Technology will not replace teachers, but teachers who use technology effectively to develop autonomous learners will replace those teachers who cannot”.

Reskilling Revolution

In early 2020, The World Economic Forum launched Reskilling Revolution. Stemming from the expected displacement of jobs due to technological change, Reskilling Revolution has attracted support from a global community of leaders from government, industry, civil society, and academia with one clear aim – “to provide one billion people with better education, skills, and jobs by 2030”.

In the World Economic Forum’s publication on Building a common language for skills at work, a list of 2025 skills trend projection was made at the global, country, and industry level. Below are some highlights from the emerging skills results:

Global Top 5 skills of 2025

1. Analytical thinking and innovation
2. Active learning and learning strategies
3. Complex problem-solving
4. Critical thinking and analysis
5. Creativity, originality, and initiative

Emerging Jobs in the Data and AI Cluster

1. Artificial Intelligence Specialist
2. Data Scientists
3. Data Engineer
4. Big Data Developer
5. Data Analyst

Top 5 Skills in the Data and AI Cluster

1. Data Science
2. Data Storage Technologies
3. Development Tools
4. Artificial Intelligence
5. Software Development Lifecycle (SDLC)

Based on the findings above, it is evident that conceptual and technical skills aside, the development of life skills in learners is equally as important to prepare them for success in a future where humans and robots will coexist. Another notable trend with the emergence of AI is the rise of data literacy, the ability to derive meaningful information from data (Bryla, 2018).

As educators, having an understanding of the emerging skills in AI will help us rethink how we shape learning for our students. The International Society for Technology in Education (ISTE) provides a great collection of resources including case studies, research, webinars, and curriculum to help educators integrate artificial intelligence (AI) and related skills in K-12 education.

References

1. Frey, C. B., Osborne, M. A. (2013). The future of employment: How susceptible are jobs to computerisation? Oxford Martin Programme on Technology and Employment. Retrieved February 06, 2021, from https://www.oxfordmartin.ox.ac.uk/downloads/academic/future-of-employment.pdf
   
2. Manyika, J., Lund, S., Chui, M., Bughin, J., Woetzel, J., Batra, P., . . . Sanghvi, S. (2019, May 11). Jobs lost, jobs gained: What the future of work will mean for jobs, skills, and wages. Retrieved February 06, 2021, from https://www.mckinsey.com/featured-insights/future-of-work/jobs-lost-jobs-gained-what-the-future-of-work-will-mean-for-jobs-skills-and-wages
   
3. Mazumdar, M. (2020, October). How bad data keeps us from good AI. Retrieved February 06, 2021, from https://www.ted.com/talks/mainak_mazumdar_how_bad_data_keeps_us_from_good_ai
   
4. Sloan, W. M. (2012, July). What is the purpose of education? Association for Supervision and Curriculum Development (ASCD). Retrieved February 07, 2021, from http://www.ascd.org/publications/newsletters/education-update/jul12/vol54/num07/What-Is-the-Purpose-of-Education%C2%A2.aspx
   
5. Lee, K. (2018). AI superpowers: China, Silicon Valley, and the new world order. Houghton Mifflin Harcourt.
   
6. Luckin, R., Holmes, W., Griffiths, M. & Forcier, L. B. (2016). Intelligence Unleashed. An argument for AI in Education. Pearson. Retrieved February 07, 2021, from https://static.googleusercontent.com/media/edu.google.com/en//pdfs/Intelligence-Unleashed-Publication.pdf
   
7. Bughin, J., Hazan, E., Ramaswamy, S., Chui, M., Allas, T., Dahlström, P., . . . Trench, M. (2017). Artificial intelligence: The next digital frontier? (Discussion paper). McKinsey Global Institute. Retrieved February 07, 2021, from https://www.mckinsey.com/~/media/mckinsey/industries/advanced%20electronics/our%20insights/how%20artificial%20intelligence%20can%20deliver%20real%20value%20to%20companies/mgi-artificial-intelligence-discussion-paper.ashx
   
8. Pant, M. M. (2020, January 24). The future of education is the AI powered autonomous learner [Blog post]. Retrieved February 07, 2021, from https://mmpant.com/2020/01/24/the-future-of-education-is-the-ai-powered-autonomous-learner/
   
9. World Economic Forum. (2021, January 22). The reskilling revolution. Retrieved February 07, 2021, from http://www.reskillingrevolution2030.org/
   
10. World Economic Forum. (2021, January). Building a common language for skills at work: A global taxonomy (Publication). Retrieved February 07, 2021, from https://www.reskillingrevolution2030.org/reskillingrevolution/wp-content/uploads/2021/01/Skills-Taxonomy_Final.pdf
    
11. Bryla, M. (2018, September 19). Data literacy : A critical  skill for the 21st century [Blog post]. Retrieved February 07, 2021, from https://www.tableau.com/about/blog/2018/9/data-literacy-critical-skill-21st-century-94221
    
12. International Society for Technology in Education (ISTE). Artificial intelligence in education. Retrieved February 07, 2021, from https://iste.org/learn/AI-in-education