AI & learning outcomes
In 2016 Lee Se-dol, the world's top-rated Go player lost three games in a row to AlphaGo, an Artificial Intelligence machine programmed to play the ancient Chinese board game.
Our former champion did not take the defeat well.
Following the loss, Lee Se-dol vowed never to play Go professionally again. "Even if I become the No. 1, there is an entity that cannot be defeated," said Lee. "So what's the point?"
Lee's story is a cautionary tale highlighting what some futurists have argued for years; it's only a matter of time before artificial intelligence (AI) makes the human brain look as dumb as a wet sponge.
Others believe that future AI capabilities are over-hyped. These team-human thinkers acknowledge that AI will play a central role in 21st-century digital industries, but they're confident that the thinking machines will always be under our control.
So let's keep our fingers crossed on that one.
In the meantime, instead of trying to subjugate the human race as an army of meat slaves, AI is doing great things, including helping us learn faster and smarter.
AI is improving the classroom environment for students and teachers. Integrating learning analytics and adaptive learning platforms into the classroom is helping teachers provide individual learning plans, including customised learning deliveries tailored to different learning styles.
Moreover, responsive data platforms provide real-time information to teachers and administrators, enabling schools to identify and address attendance issues earlier, increasing student retention and engagement.
And then there are the friendly robot teaching assistants who are more than happy to do the repetitive and time-consuming admin work, giving their human counterparts more time and energy to work with students and drive better learning outcomes.
Let's take a closer look at all these developments in more detail.
Adaptive learning and AI.
Adaptive learning is an educational approach that uses the latest technology to deliver customised resources, activities, and learning objectives to match the unique needs of each learner.
The first example of modern(ish) adaptive learning was BF. Skinner's teaching machine. Designed and built in the 1950s, it focused on incremental skill building. Skinner's device adapted the flow of questions based on previous answers, providing students with immediate feedback and allowing them to learn at their own pace.
The principles behind adaptive learning haven't really changed in the last 70 years. Skinner machines and methodology were based on the following:
- Dynamic and intelligent adjustment of content based on previous performance
- Self-paced study
- Optimal, individualised learning pathways
- Instant feedback and remediation
These still apply to 21st-century adaptive learning platforms. The only thing that has changed (and changed quite dramatically) is the technology.
Powered by AI and data analytics, adaptive learning 3.0 can provide the kind of individual learning that Skinner could only dream of. For example, AI-powered adaptive platforms can analyse the relationship between content, learning objectives, personality types, and knowledge level to create a personalised learning pathway that is...well...actually personalised.
This more efficient and effective learning experience enables:
- Real-time and complex pathway adaptations based on learner performance and behaviour
- Instant, data-driven, personalised feedback for each individual student
- Predictive analysis of future performance and improvement curves, setting realistic targets for each student.
- A massive reduction in learning times and teaching hours wasted on repetitive admin work and classroom planning.
The great leap forward
Adaptive learning platforms harness AI to evaluate student performance. They adjust the delivery and difficulty of educational content based on individual learning styles and competency levels. These intelligent platforms adapt to the user's needs, creating a learning environment where everybody has the same chance to reach their full academic potential.
Smarter chatbots in schools and colleges
These more intelligent bots have a wide range of potential applications.
Georgia State University adopted AI chatbots to deal with the 'summer melt' – students who were accepted but never actually enrolled. The school's AI-enhanced chatbot, Pounce, reached out to students, performed vital data collection on enrollment obstacles, and responded to thousands of questions via messages sent directly to smartphone devices.
The Loyola University of Chicago piloted its AI-powered digital assistant/chatbot LUie in the fall semester of 2019. LUie provided instant answers to questions that would have previously required a phone call or some serious googling. LUie sent out vital updates to students and applicants during the COVID-19 lockdowns. And when re-enrollment started up again, LUie's ability to instantly check students' GPA scores and other important information reduced some online application timescales to minutes instead of hours.
Intelligent tutoring systems
Intelligent tutoring systems (ITS) are super smart chatbots that can answer questions and provide personalised instructions and feedback to users without needing a human teacher. In other words, these are chatbots that can teach. Although calling them a chatbot is a bit like calling the latest iPhone a 'mobile phone'; it just doesn't do it justice.
Korbit AI is one of the best examples of what these intelligent tutoring systems can do. Korbit's AI and data-powered interface looks just like a chatbot, but scroll down the feed, and you'll see a range of interactive learning resources, followed by a link to a quiz. Students can attempt to answer the question or ask Korbit for assistance. If the answer is wrong, Korbit's AI brain really kicks in. It searches for appropriate educational resources to help students fill the knowledge gap. The resource might be a cheat sheet breaking down a mathematical formula, a written clarification of a logic problem, or some multiple-choice options to prompt learners toward the correct answer.
Korbit is smart enough to teach college-level data science and is now being used by over 20,000 learners worldwide. Korbit is helping them learn in a way that is faster, more efficient, and far more personalised. And its AI-driven approach can guarantee a superior teaching experience to other online learning platforms, including Massive Open Online Courses (MOOCs).
A controlled study by researchers at Cornell University found a significant increase in the learning outcomes with students on the Korbit platform compared to those using a MOOC. In some cases, the increase was up to 2.5x higher.
The study titled A New Era: Intelligent Tutoring Systems Will Transform Online Learning for Millions, reads:
"The results demonstrate the tremendous impact that can be achieved with a personalised, active learning AI-powered system. Making this technology and learning experience available to millions of learners worldwide will represent a significant leap towards the democratisation of education."
Replaced by robots?
Elon Musk recently unveiled a prototype of Tesla's humanoid robot Optimus. Musk is now aiming to mass-produce the bots and expects them to be on the market within the next five years.
Musk stated that the Tesla Bots were "designed to replace people in repetitive, boring, and dangerous tasks and are, of course, intended to be friendly."
That's comforting to know, especially given some of Musk's more pessimistic views on the future of AI and Robotics. He's previously suggested that Terminator 2 could end up looking like a documentary. And that's terrible news. Arnold Schwarzenegger turns 76 on his next birthday. He won't be 'coming back' to save us this time.
But whatever does or - hopefully - doesn't happen when Tesla bots start marching off the production line, thinking robots are part of the future of work. And they're part of the future of education.
They've already arrived in classrooms.
So is this the beginning of the end for human educators?
No. And it probably won't be for a very, very long time.
Instead of replacing 'real' teachers, these robots operate as classroom assistants, providing basic support and supplementing teaching plans and lessons.
The introduction of intelligent robots in education is about integration, not replacement.
NAO is a humanoid robot designed by the French firm Aldebaran Robotics. It's been used as a teaching resource for kids with autism since 2013. NAO's software is programmed to improve students' social interaction and verbal communication skills. And in China, kindergarten teachers get extra assistance from a small robot named Keeko. Keeko tells stories, poses logic problems, and reacts with positive facial expressions when students respond with the correct answer.
Then there's 'Yuki.' He 'works' at The Philipps University of Marburg in Germany. Yuki marks papers, invigilates tests and exams, and can answer basic questions and queries.
'Yuki' was a welcome addition to the university's teaching faculty.
"Yuki is a big help," says Professor Jürgen Handke, the first person at the higher education institute to work alongside the bot. "He frees up my time to plan lessons, research, and work more closely with my students. I'm extremely confident that he won't be taking my job anytime soon."
Yuki receives regular upgrades and updates, so he's constantly evolving. Yuki recently started giving his own lecture, albeit a very basic introduction to first-year topics.
Still, does this mean Yuki and his fellow are only a few generations from outsmarting their human co-workers and taking over further education for their own nefarious ends?
Again, the answer is 'unlikely.' And it comes down to how we - and AI - process language.
Why computers can't speak 'human'
AI is capable of performing unbelievably complex calculations in milliseconds. And it will get them right every single time. Guaranteed. However, when it comes to interpreting the nuances of human speech, AI is still pretty dumb.
In his seminal research paper on AI, machine learning, and language, Douglas Hofstadter, a Pulitzer Prize-winning author, translator, and cognitive scientist, wrote, "Language and translation is an incredibly subtle art that constantly draws on one's creative imagination."
"Until computers start thinking like humans, which is still a long way off, they'll never be able to fully grasp our layered forms of language, like sarcasm, irony, or humour. Moreover, they'll continue to be confused by ambiguous adjectives like "deep", which can be a relatively mundane description of a body of water or an expression of intellectual profundity."
In other words - and people always have different words to use - AI robots can perform calculations that would melt the human brain, but they need to figure out where to start when tasked with interpreting the ambiguities that define Hamlet's most studied monologues.
Better essay writing through technology
AI can't deliver a thought-provoking lecture on Hamlet, but it can help humans write thought-provoking essays about any of Shakespeare's plays.
Grammarly is probably the most well-known writing assistant. And at the risk of giving some of our content writing secrets away, Grammarly is awesome. It's like having a friendly English grammar teacher watching over your shoulder as you type out each sentence. Users learn as they write, receiving real-time feedback instead of waiting for a teacher to read and mark their work—apps like Grammarly equal less time waiting and more time learning.
Grammarly is ideal for people who prefer a more active style of learning. These are people who learn by doing, not listening. Active learners can often become disengaged by the more academic learning approach on courses and programs requiring lots of essay writing.
Purists would say that essay writing apps are a form of soft cheating and suck away students' brain power by outsourcing critical thinking tasks to AI. But, in this case, the purists are wrong. Like all good pieces of edtech, apps like Grammarly speed up the teaching and learning process.
A student who wants to learn how to introduce a subordinate clause into a sentence - like this one - could waste an hour locating a copy of 'The Elements of Style' in the university library. Or they could waste a different hour scanning grammar websites for the one piece of information they're looking for.
Or they could play around on Grammarly for a few minutes. Grammarly shows you how to construct and arrange all kinds of sentences instantly. Grammarly and other AI writing tools mean students work smarter, not harder.
Indonesia has a big home tutoring culture, where parents pay for private tutors to help their children with homework. However, geographical distances and costs mean that not all children have the same opportunities. Indonesian EdTech startup CoLearn provides the ideal solution for students and parents aiming for academic excellence.
That's where CoLearn comes in.
CoLearn is an online learning platform with over 300,000 video lessons. They include AI integration features that make learning more efficient, accessible, and productive. CoLearn's Ask feature allows students to scan and upload photos of homework problems. CoLearn then searches its vast catalogue for a suitable lesson or learning resource.
"Homework and private tutor lessons are expensive and time-consuming,"
says Abhay Saboo, co-founder and CEO of CoLearn.
"Our AI can be used in tandem with existing tutoring infrastructure to provide more streamlined and efficient academic assistance. Our research suggests that our technology saves students and teachers up to four hours a day. And we're scaling up all the time. At the start, CoLearn could answer a few hundred questions in a day. Now, it's answering over one hundred thousand."