Educational AI

Educational AI: Benefits & Risks

by Jeremy Deering

Note: This may not be reprinted without my permission. I am including this paper to demonstrate my theoretical knowledge and research within the educational gamification field.

Looking to AI and the future, educators should use educational technology to help their students. Clearly there are certain benefits and risks with educational artificial intelligence (AI). This essay will evaluate several writers, who have explained the benefits of AI and the risks of AI.

Benefits of AI

First, there is the theme that AI will be the innovative wave of the future. By using artificial intelligence, educators hope to grow learning. "According to eLearning Industry, upwards of 47% of learning management tools will be enabled with AI capabilities in the next three years." (Karandish, 2021, p. 1). So this trend of innovation represents a turning point.

During the pandemic time period, there is an incentive for educational technology to grow because of the need for remote learning. Basically EdTech is increasingly important for both the K-12 schools and the higher education at colleges and universities. "AI has the power to optimize both learning and teaching." (Karandish, 2021, p. 1). It is this innovation point which is beneficial

Using AI, the students experience benefits. For example, AI may improve communication between the students and their teachers. In this case, the goal of AI is to simplify education by freeing up the time schedules of both the students and the teachers. Ideally time saved via AI will be used for higher level thinking processes. For example, there are multiple levels with Bloom's Taxonomy. Perhaps the AI can help with the knowledge and comprehension levels. Then the teacher can work with the students on higher levels such as application, analysis, etc. For example, with an AI reading tutor; the AI could change the difficulty level automatically. So the AI creates personalized instruction. In that case, the AI scores a reading assessment based on the five components of reading such as phonemic awareness, phonics, fluency, vocabulary, and comprehension.

Second, there are personalization benefits with AI. For instance, students can have a "personalized approach to learning programs based on their own unique experiences and preferences." (Karandish, 2021, p. 1). In a way, this can be accomplished by analyzing how the students were learning in the past. So the students' weaknesses are examined, and the AI will help provide a roadmap to achieve future success. In fact AI helps by "providing access to the right courses."(Hubert, 2021, p. 1). With AI advice, the student has an AI learning expert.

Also AI helps individual needs. For example, "Every student is different from the other." (Maurya, 2021, p. 3). And AI "helps to adapt to the student’s speed of learning, weaknesses, abilities, interests in specific courses, future goals and more." (Maurya, 2021, p. 3). This means that the personalization factor of AI makes a twenty person class into twenty different learning paths. For example, there is a reading tutor AI. If ten students have more difficulty with vocabulary, then the AI can pinpoint that challenge. On the other hand, if the other ten students need more help with phonics, then the reading tutor AI can focus on that.

Plus AI tutoring gives more advantages. "AI tools can help students sharpen their skills and improve weak spots outside of the classroom." (Karandish, 2021, p. 2). The main advantage of this is the idea of tutoring on demand. Speedy responses matter when it comes to educational questions; therefore, AI tutoring can help. This means that the "benefits of AI in education drive many schools to rethink their approaches." (Hubert, 2021, p. 1). So AI tutoring augments teaching because it helps to promote positive changes in the curriculum. "Advances in AI are giving teachers a better understanding of how their students learn and allowing them to customize the curriculum accordingly." (LiveTiles, 2021, p. 3). After all, if AI tutoring helps to reveal the students' knowledge deficits, then changing the curriculum may help to avoid those deficits in the first place.

With AI technology, the mechanics of tutoring are also changed. For example, ITS (Intelligent Tutoring Systems) "can effectively challenge and support the learner using different algorithms." (LiveTiles, 2021, p. 3). This is interesting because if tutoring involves flashcards, then intelligent AI tutoring is like the flashcards rearranging themselves to become more effective. So this AI tutoring represents more efficient educational training.

Third, AI can help with repetitive tasks. For instance, there is "the ability to delegate manual tasks." (Hubert, 2021, p. 1). Examining AI applications, one sees that the questions facing teachers are often quite similar. As a case in point, an article mentions that teachers are hounded "with repetitive questions on a daily basis. AI can help students find answers to their most commonly asked questions." (Karandish, 2021, p. 2). In a way the AI can function like a mechanical FAQ system.

Another advantage for students is that AI systems have universal access. This means that, if the students know how to use it and if the students have the correct computer, then they will be able to use the AI system whenever they like. Basically online AI is a product of the Internet-based classroom instruction. In some cases AI can help with equity by "making classrooms available to students around the world, including those who speak different languages or have hearing or vision problems." (Hubert, 2021, p. 1).

Fourth, AI helps with speed. "Artificial intelligence can help students find answers to their questions within seconds." (Hubert, 2021, p. 2). Not only is the learning faster, but it also saves time between learning sessions because the EdTech is Internet-based. For example, AI "makes it possible for students to learn at any time and from anywhere." (Hubert, 2021, p. 2).

Now, there are also some AI benefits for educators. For example, there is the issue of time management. In fact, "educators want to spend more time educating students one-on-one, diving into research and continuing their own education." (Karandish, 2021, p. 2). Additionally there is also a personalization process for the instructors. Essentially the AI shows the strengths and weaknesses of the instruction, so the teachers may improve their efforts. I interpret this to mean that the AI is like an automated self-improvement device, due to its feedback mechanism.

Automated learning also helps the teachers save time. "AI-powered chatbots can answer a variety of generic and repetitive questions students typically ask without involving a faculty member." (Karandish, 2021, p. 3). Ideally the time, which is saved, can be better spent on planning lessons, researching knowledge, or improving the overall course. AI apps "can learn from each teachers’ experience to automate a variety of tasks." (Hubert, 2021, p. 3). And that saves time too.

Even those planning tasks can be helped with AI: "teachers spend 31% of their time planning lessons, grading tests and doing administrative work." (Karandish, 2021, p. 3). Here the goal is that, by having support automation tools, the AI can help with student engagement. In other words, the more mundane tasks are helped with AI tools. My reaction to this is that the teacher had better be sure that the AI is doing an adequate job. Those support automation tools have to be understood by all the students, or else there will be confusion as the class goes forward.

Fifth, AI helps with data analysis. For example, "While AI-powered apps help teachers automate tasks, they also collect data for further analysis." (Hubert, 2021, p. 3). This data is needed to improve the class and its teaching methods. "Artificial Intelligence in education has helped schools and other educational institutes to optimize their tasks" (Maurya, 2021, p. 3). Without the appropriate data-gathering, this task optimization cannot happen. In fact data is like a two way street since there can be survey data of how the students perceive their teachers. In fact "AI improves the process of education for students via enhancing interaction with teachers." (University of Technology, 2021, p. 1). And the data, conceived with AI, helps that interaction.

Sixth, AI can help the teachers with grading. For instance, "during the grading process, the program analyzes each test to provide teachers with valuable feedback about students’ academic progress and knowledge." (Hubert, 2021, p. 3). That means the teachers will have the data, needed to help the students to improve their future performances. In fact "Educational data can be accessed whenever and wherever." (Maurya, 2021, p. 3). Therefore the grading, feedback, and improvement aspects of teaching are likewise possible whenever and wherever.

Sometimes AI can grade essays. For example, "software that can instantly grade student essays is a significant benefit. Every graded essay adds to a central database to which future essays are compared." (LiveTiles, 2021, p. 3). This will save the time, which is devoted to routine tasks.

Besides grades, AI can help with the issue of student retention. "Schools are using AI to do everything from identifying potential dropouts to answering student questions online using AI-powered bots." (Mire, 2019, p. 2). This is really great because it will help to keep the students in school. Also there can be student satisfaction surveys, which are meant to be a mechanism of feedback. Anything that can help the students to become successful in schools is really a blessing for the school system. Whether it is K-12 or higher learning, student retention is a win-win for both students and teachers.

Seventh, there is more time for creative teaching. As a matter of fact, "Now the educators find more time to come up with creative ways of learning that can help students understand better." (Maurya, 2021, p. 4). This is interesting because it represents how AI is changing the qualitative aspect of what can be done via EdTech. Consequently those changes are not only quantitative results; they are positive quality changes in education.

AI can help with educational games. For instance, AI "is being used in various simulation and gaming technologies already that can play a major role in this regard." (Samantha, 2020, p. 3). Over time the gamification of education will make learning fun.

Eighth, AI can help with educational access. "Whether deaf, blind or with any other disabilities, there are AI tools created for their education." (Maurya, 2021, p. 5). This is important because access represents a quality of life issue. "Now they can have access to high end learning without all the struggle they usually have to face in normal day to day life." (Maurya, 2021, p. 5). It is interesting that AI can do a lot of good in many areas of broadening access.

Ninth, AI helps with the study of educational groups. For instance, intelligent moderation "allows human tutors, moderators and teachers to analyze the data produced by large groups with the assistance of AI techniques like machine learning." (LiveTiles, 2021, p. 3). Of course, if this were done by a single teacher, then this would take longer. So AI intelligent moderation saves the time it takes to study the overall group of students. "In turn, educators can be more efficient in the classroom." (LiveTiles, 2021, p. 3).

Tenth, AI helps with self-paced learning. As an illustration of this concept, "artificial intelligence-based applications allow learners to study during free time, spending 10 or 15 minutes on a task." (Lynch, 2022, p. 2). This is good because it gives the learners more direction and control over their learning. Additionally "learners can get feedback from teachers in a real-time mode." (Lynch, 2022, p. 2). This matters because the feedback will provide the foundation for the next level of the students' self-paced learning.

Lastly, AI helps the students to overcome their weaknesses. For example, " the Classra app could notify the educator if most learners chose incorrect answers to a question. The teacher can then focus on the areas of weakness that are identified." (Lynch, 2022, p. 2). Of course this idea of helping to overcome wrong answers is a major part of AI.

Risks of AI

? In contrast to the benefits of AI, there are certain risks from artificial intelligence. Several writers have shared their warnings, concerning the drawbacks of AI.

First, overemphasizing computer AI risks overshadowing the contributions of human teachers. In fact "the key driver of AI applications is cost-reduction, which means reducing the number of teachers, as this is the main cost in education." (Bates, 2018, p. 3). Of course the promise of cost-reduction must not be at the expense of harming people. For example, AI "solutions frequently incorporate false and/or unsupported educational ideas reflecting the biases of their developers." (Bates, 2018, p. 2). So this means that the student-teacher interaction must continue to be emphasized.

When developers are biased against teachers, there should be new developers brought in. And these new developers should try to use AI in harmony with human teachers, so that the situation is a win-win. Otherwise it should not be a win-lose proposition, where AI wins and teachers lose. We should remember that, "Teachers are the most important part of a society." (Adil, 2021, p. 3).

Second, AI recommendations must have relevant criteria. When reviewing the students' works, the AI can make recommendations. These results come from how the AI is programmed. Yet the recommendations should not stifle creativity. "This means replacing criteria such as the number of hits, or likes, with more educational criteria, such as clarity and reliability." (Bates, 2018, p. 2). Ideally, AI will "create more immersive learning experiences in which students use their voices and critical thinking skills to access information." (Brown, 2018, p. 1). But there must be a clear understanding of how the AI algorithm works, so the technology is actually relevant for K-12 schools and colleges.

Sometimes emotions are needed in recommendations. This represents the human feeling, which builds connections within the classroom. "Lack of emotional intelligence is also one of the negative effects of artificial intelligence in education." (Adil, 2021, p. 4). Understanding how someone else feels is important when understanding how to interact. So the classroom interactions "critically depend on emotion and meaningful human connections to be optimally beneficial." (Bates, 2018, p. 2).

Third, educational AI has privacy dangers. With AI there are "a host of potential data privacy risks." (Brown, 2018, p. 1). For example, many common apps have privacy issues: "Voice-activated, artificial intelligence devices such as Alexa and Google Home are becoming a part of classrooms." (Brown, 2018, p. 1). Since this technology is being adopted so rapidly, there is a grave danger that its strategic issues are not being fully analyzed.

Schools with AI need privacy policy strategies. Obviously, there are rules such as the Family Educational Rights and Privacy Act, the Children’s Online Privacy Protection Act, and the Children’s Internet Protection Act, which are meant to address these privacy concerns. Often teachers "don’t really understand what those laws are and how noncompliance can impact their students, themselves and the district." (Brown, 2018, p. 1). This is not a small matter; think about how lawsuits can damage a school district. First, there is the monetary issue of money damages. Second, there is a loss of trust between students, teachers, administrators, and the community when there are legal issues. Third, the fact that these laws exist shows just how much privacy should be respected. In this case, it is very important to have legal advisors study how AI is used within the classroom.

Voice activation is another privacy matter. This is one of "a host of potential data privacy risks." (Brown, 2018, p. 1). One teacher mentioned in "The privacy risks of AI use at school" article how the AI devices are "wide open for student voices to be recorded and used to develop algorithms. These devices also circumvent any school filtering and firewall." (Brown, 2018, p. 1). And there must be school district policies, which address this voice-activated AI risk.

Student names are at risk with AI identification. There are FERPA policies. "The Department of Education is very clear that student names or any kind of identifying information, cannot be divulged." (Brown, 2018, p. 1). Now, part of this problem is if an app has data-sharing policies. This means that some information may be shared across a company's related apps. Or else it could be shared with third party companies. It is very important to read an AI company's privacy policies, so those specific concerns are addressed.

Fourth, AI can reinforce discriminatory bias. As a matter of fact, there is the "potential to reproduce or even amplify existing biases.” (Wood, 2021, p. 1). And that means diversity must be a part of AI adoption. Part of the reason for bias is how machine learning works: Generally "the software [is used] to observe a very large set of examples of past users." (Wood, 2021, p. 2). So, if those past users were from a sample population, then members of that population may be better served by that AI. In contrast, the users from the alternative populations, which were not factored within the machine learning's data gathering, will not be as well-served by this AI. Really the AI is more a reflection of the bias of its creators. Sometimes that bias can be unintentional; nonetheless, it is a risk of AI.

Also AI recommendations can be biased. "Algorithmic bias is discrimination against one group over another due to the recommendations or predictions of a computer program." (Wood, 2021, p. 2). This is a subtle form of discrimination because its consequences may take awhile to be fully understood. In other words, if a biased recommendation leads to a poor decision, then that biased decision will have further consequences.

With this knowledge there is a key concern: What is particularly alarming is that those biased recommendations, poor decisions from them, and the results of the AI apps will lead to more data. The problem is that the new data will be tainted by the biased factors which created it. Yet, with machine learning, the new biased results will then be interpreted as the updated model information. Essentially the bias reinforces itself. Unfortunately, unless a careful audit is undertaken, it may be very difficult to fully understand just how pervasive the bias is. Consequently any systemic bias, reinforced via AI recommendations, is very damaging for an educational system.

Fifth, AI may create a problem by replacing too much human input. For example, it is reported that, "One of the main negative effects of artificial intelligence is it decreases human interaction in education." (Adil, 2021, p. 2). Interestingly, human interaction is needed for higher level learning. Some examples of this are: "deep knowledge about life... experiences, and... ways of living." (Adil, 2021, p. 3). Basically there is a risk that AI becomes too much of a substitute for teachers. And then the highly valuable knowledge potential of teachers is lost. This would be unfortunate. So we should ensure that the human element of AI, student, and teacher interaction is not forgotten.

Sixth, there is the risk that educational AI can be expensive. Usually "content is neither easy nor cheap." (Bates, 2018, p. 2). Now, that is the software content. Likewise the hardware is also cost-prohibitive for many: "The implementation and maintenance of robotic machines and artificial intelligence in educational institutes are very expensive." (Adil, 2021, p. 3). In the field of business technology, there is an innovation cycle: Often being an early adopter is the most expensive purchase cost. So that means the newest technology is almost always the most expensive. Yet, in a competitive world, there is usually the need to keep up with the latest technology.

Besides purchase costs, AI has maintenance fees. For instance, "If some error or bug occurs in artificial intelligence, it will be impossible for the educational institutes to prevent it." (Adil, 2021, p. 3). What that means is the error will not be spotted until it has affected the system. And the repair of that error will become a maintenance expense. This is interesting too because it shows that an educational system which depends upon AI apps for routine, boring tasks may have to learn again how to do those tasks if the AI system goes down.

Seventh, AI can be addicting. For instance, "students get addicted to the usage of artificial intelligence." (Adil, 2021, p. 4). Perhaps, if AI makes certain parts of the educational process easier, then that will create a cycle of dependence. In that situation, the students need the AI as a crutch. In fact one educator reports how, "They only want to use artificial intelligence technology for doing any kind of work."(Adil, 2021, p. 4). Sometimes addictive behavior can happen when certain emotional needs are not being met. "In fact, the key lesson from all AI developments is that we will need to pay increased attention to the affective and emotional aspects of life in a robot-heavy society." (Bates, 2018, p. 3). Ideally AI should be a helping aid, which augments education; it should not be an addictive replacement for human connections.

Eighth, AI risks being a substitute for creativity. AI is no more creative than someone who programs it. If the students are working mostly with AI software, then that may lead to knowledge teaching on the lower end of Bloom's Taxonomy. "The trick though is to recognize exactly what kind of applications these new AI developments are good for, and what they cannot do well." (Bates, 2018,?p. 3). When creativity is needed, the students should not be reliant upon AI.

Ninth, AI models are limited. For example, "AI cannot always account for constant change and adaptation like humans can since there is no thought process and experience." (Roe, 2020, p. 7). In reality, successful schools are "being very clear about the purpose of AI applications in education and being wide awake to the unintended consequences." (Bates, 2018, p. 3). Anyone who thinks that an AI model will cure all the world's problems is being overly optimistic about the ability of AI models.

Tenth, AI may increase the digital divide. In fact "disadvantaged populations might get excluded from AI-powered education, resulting in a digital divide." (Artiba, 2021, p. 2). This is bad for America because it will harm the economy in the long run. If some students do not have equal access, then they will be shut out of the technological jobs in the future. That is why educators should "close the educational gap between economically rich and poor students." (Artiba, 2021, p. 3).

In contrast, the best way to overcome this divide is to "create a social platform where students can interact with each other, teachers and parents." (LiveTiles, 2021, p. 5). Overall humans must solve the digital divide because the AI will not do it by itself. Towards this goal educators must strive to close the divide. So that educational technology may brighten the future for everyone!

Reference List

Adil, Muhammad. (2021, November 15). Top 10 Negative Effects of Artificial Intelligence in Education. Tech Stonz. https://techstonz.com/negative-effects-artificial-intelligence-education/

Artiba. (2021, June 11). Top 5 Challenges of Adopting AI in Education. Artiba (Artificial Intelligence Board of America). https://www.artiba.org/blog/top-5-challenges-of-adopting-ai-in-education

Bates, Tony. (2018, March 2). Assessing the dangers of AI applications in education. Online Learning and Distance Education Resources. https://www.tonybates.ca/2018/03/02/assessing-the-dangers-of-ai-applications-in-education/

Brown, Emily Ann. (2018, September 13). The privacy risks of AI use at school. District Administration. https://districtadministration.com/the-privacy-risks-of-ai-use-at-school/

Hubert, Kyle. (2021, March 30). The Benefits of AI in Education. Capacity. https://capacity.com/the-benefits-of-ai-in-education/

Karandish, David. (2021, June 23). 7 Benefits of AI in Education. THE Journal. https://thejournal.com/articles/2021/06/23/7-benefits-of-ai-in-education.aspx

LiveTiles. (2021, February 5). 15 Pros and 6 Cons of Artificial Intelligence in the Classroom. LiveTiles. https://livetilesglobal.com/pros-cons-artificial-intelligence-classroom/

Lynch, Matthew. (2022, January 6). Benefits of Artificial Intelligence for Educators and Learners. The Tech Edvocate. https://www.thetechedvocate.org/benefits-of-artificial-intelligence-for-educators-and-learners/

Maurya, Sushma. (2021, August 23). 7 Benefits of AI in Education. Digital Sushma. https://digitalsushma.com/benefits-of-ai-in-education/

Mire, Sam. (2019, 29 September). What Benefits Will AI Bring To Education? 19 Experts Share Their Insights. Disruptor Daily. https://www.disruptordaily.com/ai-benefits-education/

Roe, David. (2020, April 30). A Look at the Downsides of Artificial Intelligence. Reworked. https://www.reworked.co/information-management/a-look-at-the-downsides-of-artificial-intelligence/

Samantha, John. (2020, December 11). Major Benefits of Artificial Intelligence in Education. Witty Sparks.?https://wittysparks.com/major-benefits-of-artificial-intelligence-in-education/

University of Technology. (2021, September 30). Benefits of AI in education. University of Technology. https://www.universityoftechnology.edu.in/benefits-of-ai-in-education/

Wood, Maria. (2021, May). What Are the Risks of Algorithmic Bias in Higher Education? Every Learner Everywhere. https://www.everylearnereverywhere.org/blog/what-are-the-risks-of-algorithmic-bias-in-higher-education/