Artificial intelligence has come a long way in recent years, but how much can it change the educational landscape?
Artificial intelligence (AI) has recently been propelled as the next technological phenomenon and its application and possibilities in different fields are vast and varied. However, when it comes to its use in teaching and general educational activity AI still has a lot to learn.
Understanding Current Artificial Intelligence
When one hears the term “artificial intelligence” one often thinks of its closest approximations to science fiction where machines or computer systems are aware of themselves and can think or behave just as a human would.
Although the goal of researchers and scientists is to make a machine think the same way a human would, AI and its current possibilities are more linked to data learning. A process called “machine learning”.
As such, Machine Learning is the computational method where a system calculates possible probabilities in huge data sets. With that calculation of probabilities, the system can “predict” what might happen in the environment in which it is applied.
We can see the application of machine learning in things like technology for auto-completion of texts, real-time translation of complex languages, image recognition, autonomous cars or in board games and video games.
In fact, the most popular cases where artificial intelligence has “surpassed” humans are often related to video games—such as Dota 2—, or board and strategy games—such as Shogi or Go—.
But recognizing images or predicting Shogi’s moves is not remotely the same as adapting to new academic challenges.
Artificial Intelligence in Education
The data analysis part of today’s AI is especially important for environments where students need to make informed educational decisions.
Such is the case of Georgia State University in the United States, where recently technologies based on automatic learning have been used to improve the graduation rate.
Its application is clear: It analyses students’ grades and historical information and predicts which of them might be most likely to drop out of school so that action can be taken before that happens.
Another real application of artificial intelligence in the educational field is the development and application of technological tools that can support teachers in activities such as assigning tasks according to the academic profile of each student.
Thinking of classrooms where students come from different parts of the world, we also work on Microsoft PowerPoint plug-ins that help translate slides into several languages in real time.
The current objective of academic artificial intelligence is the creation of support tools that can better understand the student and provide efficient information for the teacher. However, the teacher and other figures, besides being benefited by AI, must also act as its main regulators or promoters, since without the correct intervention of human criteria and initiative, AI cannot yet take a leading role in the educational environment.
Even though its possibilities are almost infinite, artificial intelligence still has several obstacles to overcome before thinking about a real educational revolution.
The limitations of a disruptive technology
When it comes to making a “smart machine” it can be understood through two methods. In the first method, engineers provide specific knowledge to systems or give orders to systems to complete a specific action. This method is generally called “knowledge-based”. In the second, the system or “machine” has the ability to learn for itself and create its own rules for completing an action. This method is what is being done in modern artificial intelligence processes.
Humans can “jump” from one method to another and, if necessary, we can instantly integrate them into our daily lives. But in order for artificial intelligence to do this, it would have to work with two types of system and reasoning at the same time: One that is deductive and through standard procedures can make a decision and another that is inductive, that can recognize patterns in the information it is processing and can predict what will happen and act in prevention.
Today’s artificial intelligence cannot understand both types of reasoning at the same time. They are good at one, but not both, as is the case with the human brain.
This leads us to the main barrier that current artificial intelligence encounters with respect to an activity such as education, it does not “think” like a human and therefore it is impossible for it to understand the processes with which humans learn or the situations that make them not learn.
For example, in the MIT Cognimates project, where children aged 7-10 learn the basics of machine learning programming, it was found that the children who had the best results in programming were not those who spent the most time programming – a merely logical-mechanical activity – but those who shared their methods and experiences in programming with their peers.
This type of learning, focused on communication and socialization, is one of the ways in which we humans learn the most and above all it is proof that our brain can reason deductively and inductively at the same time even at very early ages. Girls and boys knew the basics of programming, obtaining a result (deductive reasoning) and shared their experiences to reach an improvement in their processes (inductive reasoning).
Although AI is far from the human capacities and from “thinking” on its own as science fiction portrays it, it does have capacities that could represent a problem for society as we know it and therefore it is important to know how to regulate it.
The importance of ethics in artificial intelligence
One of the most famous uses of today’s artificial intelligence is its application in autonomous cars. Cars that have the ability to drive themselves thanks to risk prediction processes and that technically have the capacity to offer their services in transport apps.
However, even though the car could take a person from point A to point B without a problem, it does not have the capacity to know the risks that may exist within itself. In other words, autonomous car cannot calculate or deal with the risks of misconduct among its own passengers.
Human behavior is so unpredictable that it would be irresponsible to have an autonomous car without a regulatory figure inside (such as a driver) to do the job and take responsibility for actions that only human judgment can address.
The same thing is happening again in the educational environment. In the specific case of Georgia State University, there has been concern on part of researchers about whether or not making an in-depth analysis of student data is an invasion of privacy, or whether artificial intelligence is racially biased in saying that students who come from diverse communities are more likely to drop out.
That is why Georgia State University has placed special emphasis on hiring school counselors who can better interpret the information collected by artificial data analysis. Again, human judgment is needed even with intelligent machines supporting.
Although it is certain that machines and their intelligence will continue to grow – at present a processor is 100 times more powerful than it was in 2012 – it is also certain that humanity is complex and that learning and teaching are human activities plagued with particularities that artificial intelligence does not yet understand.
But could an AI replace a teacher’s job? Probably not, since even with its remarkable progress in recent years, it is far from achieving anything remotely like it. And even if AI had the capacity to do so, it’s not the obligation of humans to let that happen.
After all, it is humanity that guides its own future and it is not the responsibility of machines or artificial intelligence to do so.
Sources and references:
Barshay, J., & Aslanian, S. (2019, August 6). Predictive analytics are boosting college graduation rates, but do they also invade privacy and reinforce racial inequities? Retrieved October 21, 2019, from The Hechinger Report website: https://hechingerreport.org/predictive-analytics-boosting-college-graduation-rates-also-invade-privacy-and-reinforce-racial-inequities/
Preston, C. (2019, September 26). Will AI really transform education? Retrieved October 21, 2019, from The Hechinger Report website: https://hechingerreport.org/will-ai-really-transform-education/
Seabrook, J. (2019, October 14). Can a Machine Learn to Write for The New Yorker? | The New Yorker. Retrieved October 21, 2019, from The New Yorker website: https://www.newyorker.com/magazine/2019/10/14/can-a-machine-learn-to-write-for-the-new-yorker
Thompson, D. (2018, September 28). Can Artificial Intelligence Be Smarter Than a Person? Retrieved October 21, 2019, from The Atlantic website: https://www.theatlantic.com/ideas/archive/2018/09/can-artificial-intelligence-be-smarter-than-a-human-being/571498/
