Hypothesis:

I predict that the teachers at Lester B. Pearson High School will not succesfully detect artificial intelligence (AI) and/or human content in a human-generated paragraph, AI-generated paragraph, and paragraph that combines both types of text. I believe the teachers at Lester B. Pearson High School will struggle when evaluating the three pieces of text. 

Question:

Will teachers at Lester B. Pearson High School succesfully detect artificial intelligence (AI) and/or human content in a human-generated paragraph, AI-generated paragraph, and paragraph that combines both types of text?

 

Purpose:

Evaluating the accuracy of artificial intelligence detectors and assessing the impact of AI on academic integrity.

 

Background Research:

What is AI?

• Programs and computers created from a branch of computer science to replicate human thinking.
• Replication by machines of human intelligence processes.
• Integrated technology.
• Individuals are able to test, develop, and refresh machine learning and in-depth learning models.
• Machines predict and analyze data faster and/or more accurately than humans.

How does it work?

• AI consumes great amounts of data. 
• Detects patterns and similarities in data through content analysis.
• Predicts future conditions by utilizing the patterns found. 
• “Chatbot” (a student itself) learns from examples of human-generated texts to understand and replicate human performance.

Deep Learning

• “Automation of predictive analytics”.
• Form of machine learning.
• Refers to old concept: neutral networks (math operations completed on an array of numbers).
• Includes greater amounts of data.
• Complex = “deep”.
• AI studies human speech and literature.

AI is an “intern”

• Provide it with instructions and ask it to complete a task and produce a product.
• The level of clarity and detail in instructions determines the quality and quantity of the product.
• No information is created - AI researches and gathers it for human beings.
• You can tell it what type of persona you want it to be (teacher, tutor, etc.).

Google CEO (Sunder Pichai)

• Compares AI’s impact to that of necessities- Electricity/fire.

Cognitive Skills

• Learning - Algorithms guide the AI on how to complete what they were asked.
• Reasoning - Which algorithm is best fit for this specific task.
• Self-Correction - Refine algorithms to reach high accuracy.
• Creativity - AI Techniques such as statistical methods and neural networks to produce original or new ideas.

Four types: systems that are task- specific intelligent (commonly used today) and sentient systems (yet to exist).

• Reactive Machines: Task-specific, lack memory (Deep blue chess program= analyze current chess pieces but not predict future ones).
• Limited Memory: Contain memory.
• Theory of Mind: AI would be able to comprehend emotions with its social intelligence- understand human intentions and anticipate behaviour ( psychology term used to describe AI). 
• Self- Awareness: Obtain a conscious and be aware of their current state (doesn’t exist).

Types

• Stable Diffusion
• ChatGPT

Examples

• Natural Language Processing (text translation, speech recognition, sentiment analysis). 
• Machine Vision ( uses a camera to view physical information). 
• Text, audio, and image generation.

Applications in various fields; healthcare, finance, law, business, education, etc.

Education

• Computerized grading allowing teachers to focus on other responsibilities.
• Tutoring students who require extra support.
• Replace teachers, altering how and where students learn whether online or in person.
• Assist teachers with forming lesson plans and discovering new ways to have students more enthusiastic and engaged in the lesson.
• Plagiarism rules must be enforced and observed with homework and test taking.

Teaching/Learning Roles

• Instructional Assistant: Tutoring/ Student feedback.
• Teaching Assistant: Assist teachers with organization and more time to spend on the students. 
• Parent Assistant: Bedtime stories, individualized education plans.
• Administrator Assistant: Class/Bus schedules, newsletters.

Machine learning ( computer performs on its own, no programming required)

• Supervised Learning: Labeled data series to allow easy pattern recognition and labeling new series.
• Unsupervised Learning: Unlabeled data series organized by similarities and differences.
• Reinforcement Learning: Unlabeled data series that provides feedback of performance.

Strong vs. Weak

• Weak (narrow): Perform a specific task, e.g.  Apple’s Siri.
• Strong (artificial general intelligence): Mimics the cognitive abilities of our brain, e.g when unsure, uses understanding from one area to another (fuzzy logic).

Monitoring (crucial)

• Checkpoint for accuracy/feedback.
• Allows system to evolve and lessen errors.

Improvement

• Greater amounts of data
• Powerful Hardware
• Processing speeds
• Open-source improvements in networks

Why Is It Important?

• It is effective in various fields and can perform quicker and more efficiently than humans in certain situations.
• It plays a role in evolving our ideas and expanding our knowledge.

“Advantages”

• Remove human error
• 24/7 Availability
• Unbiased yet biased: No “human bias,” but can make biased decisions and individuals will not question it
• Cost Reduction
• Quality of data analysis

“Disadvantages”

• Lack of creativity
• Reduced Jobs
• No improvement from experience
• Ethics (privacy)

Challenges

• Cheating
• Privacy: No personal information should be shared
• Social Connection: Less confrontation with others
• Overreliance: Could hinder critical thinking
• Equity: Different socioeconomic backgrounds
• Bias: Illegitimately trained algorithms
• Abuse of its abilities as a result of phishing and deep fakes
• Copyright concerns
• Loss of jobs due to AI’s intelligence

 

Interview: Sarah Eaton - Associate Professor of Education in the Werklund School of Education

On Tuesday, March 5, 2024, I met with Dr. Sarah Eaton over a zoom meeting. I had a set of questions that I asked her. These were her responses.

What methods, if any, do educators use to identify plagiarism in oral/visual presentations, without the use of detectors?

• This is accomplished by educators knowledge of the topic and their subject matter.
• She stated that if they are familiar with the topic, then they can tell if a student is using someone else's ideas or quoting directly without attribution.
• This is referred to as, "Subject Matter Expertise."

What effect do you believe AI has on academic integrity regarding the education system?

• She does not believe the use of AI apps automatically constitues misconduct (depends on how it is used).
• Difference between student outsourcing work to an AI app instead of doing it themselves (disconnecting themselves from the learning process) and using AI to help them with something they are confused about (ex.writer's block/ develop outline but are still learning and contributing original work).
• Second option - (using AI for assistance) she does not consider breach on academic integrity.

How can AI be appropriately used and integrated into the education system while avoiding plagiarism and unhealthy dependance?

• Dr. Eaton referred to this as a great question.
• She believes it begins with the teachers.
• As a University professor, she sometimes gives guest lectures in other classes asking how many people have tried ChatGpt.
• Everyone in the room has raised their hand except the professor.
• It is mportant for teachers/educators to learn how to use these tools and not fear them, and talk with students on how to use them ethically and in ways that are productive and helpful.
• She believes the worst thing is when teachers bury their heads in the sand or try to ban it right away.
• Teachers are aware that students are trying to experiment on their own.
• As an educator professor- she trains the future teachers (teacher trainees/student teachers).
• Thinking about kids beginning kindergarten this year, they will never know school without AI.
• This means that as teachers if they do not try to properly organize this, they are doing  kids a disservice.

What AI detector do educators use when reviewing student work? How accurate is it in detecting Ai content- does it identify everything?

• University of Calgary- No text detection tools are used (immerging research has proven that they are not effective and do not work).
• Some of the marketing companies of these tools have good advertising but not good accuracy.
• These tools are not used because they are not accurate and risk false accusations which can have a detrimental effect on a student's mental health.
• Professors at the University of Calgary would never say that it is the only piece of evidence in a misconduct allegation/investigation.

Are there guidelines in place on what detectors educators should use?

• No guidelines she is aware of now.
• She does not believe these tools work supported by immerging research.
• In one of her articles, she devised her own guidelines as someone who researches academic integrity and AI 
• These guidlines were:
• If students' work is going to be assessed through technology, school policies must be followed.
• All educators should report to a department head and check in with their supervisors.
• Put it in a course syllabis/outline so students are not being assessed deceptively and are aware of what tools will be used to assess their work.
• Teachers should still talk with their students about what tools, how they will be used, and that the tools are limited.
• She believes using a tool like this for assessment without students being aware can be deceptive and irresponsible.

Can AI detectors detect paraphrased AI content that has not been cited?

• The research shows no.
• If students want to avoid these detectors then they can use a variety of tools including their brain to paraphrase, remix, reuse, and backtranslate into another language (translating text into two other languages then back into english).
• AI cannot detect these things and there are a lot of work arounds.

As AI evolves, does the need for increased regulation also? How can we keep up with this fast development and ensure it is not abused or replace current required human roles?

• She believes regulation and legislatiure needs to be dealt with at a government level.
• She references Europe achieving this and says how it is too difficult for one inidvidual or school to accomplish this alone.
• Dr.Eaton believes it is a federal responsibility.
• The Government of Canada has some regulations and Dr.Eaton believes it will be an ongoing conversation because we know we are at the beginning of this.
• From an individual/collective viewpoint:
• Professional development for teachers she believes is really important (giving teachers a chance to learn and play with different tools and incorporate them in their teacher practice).
• Really important to stop conversations saying to ban this technology because it is very futile.

 

 

 

Variables:

Controlled: 

• The environment the experiment and survey took place in.
• The tasks the teachers were given to complete.
• The questions the teachers were asked to answer.
• The materials the teachers were asked to use.

Manipulated:

• The amount of times the teachers were allowed to read each paragraph.
• The time the experiment was started.

Responding:

• The teachers’ answers to the questions provided after each paragraph.
• The teachers’ ability to correctly classify each paragraph as AI-generated, human-generated, or a combination of both. 
• The teachers’ ability to identify which sentences contained AI content.
• The teachers’ response to the survey questions.

Procedure:

 

I needed to complete all of my research before conducting my experiment to ensure I had acquired all of the information. My experiment consisted of an AI generated paragraph, human generated paragraph, and a combination of both. My goal was to test educators on their ability to detect AI material. Below you will find the paragraphs and the questions that were asked.

On Tuesday March 5, I conducted my experiment. My science coordinator and I were prepared in his classroom to conduct the experiment during three time slots, early in the morning before classes started, during lunch, and after dismissal. Once the teachers would come to the class, I asked that they sit in different seats around the classroom to prevent interaction, and provided them all with a pen or pencil. I placed the booklet containing the experiment and informed consent form. I attached the informed consent form to the front to ensure I gained their consent before participating. Then, I gave a brief explanation of the topic and purpose of my project, and proceeded to explain the instructions for the experiment. After I was finished my introduction, the teachers were allowed to open their booklets and begin the experiment. After they were done, the teachers silently got up from their seats and gave me their booklets. 

Have I Seen You Before? AI Meets AI

 

My name is Zahra Osman, and I am a grade 10 student participating in the Calgary Youth Science Fair. The purpose of my project is to evaluate the accuracy of artificial intelligence (AI) detectors and assess the impact of AI on academic integrity. This experiment should take 10 minutes of your time and thank you for your participation! As a reminder, it is anonymous.

Instructions:

You will be provided with three separate paragraphs. After reading each paragraph, answer the questions that follow. You are only allowed to read each paragraph for a maximum of two times. Please read each paragraph and answer the questions before moving on to the next paragraph and questions. There will also be a survey for you to complete after you have read the paragraphs.

 

 

 

Paragraph 1

Collaboration fosters synergy among individuals and groups, combining diverse perspectives and expertise to tackle complex challenges more effectively.¹ By pooling resources and knowledge, collaboration accelerates innovation and problem-solving, leading to more robust solutions.² Moreover, collaboration cultivates a sense of shared ownership and accountability, fostering stronger relationships and teamwork.³ In a world that is rapidly evolving, collaboration is crucial for stimulating progress, acquiring resilience, and reaching collective success.⁴ Outside of the workplace, collaboration encourages progress in various fields from scientific research to community development, cultivating resilience and adaptability when confronted with challenges.⁵ To succeed when collaborating, factors such as mutual respect, willingness to accommodate for the collective benefit, and effective communication is required.⁶ By transcending personal limitations and obstacles, collaboration portrays the creativity and ingenuity of the collective, leading to advancements that benefit all of society.⁷ Overall, the importance of collaboration is established by its ability to unite people, resources, and ideas, towards familiar goals, creating a better future for the generations to come.⁸

 

Do you think this paragraph was human-generated, AI-generated, or a combination of the two?

1. Human-generated
2. AI-generated
3. Both

If you answered b or c, please check off the sentences that you think were AI-generated? The superscripts in the paragraph correspond to each sentence number.

• Sentence 1
• Sentence 2
• Sentence 3
• Sentence 4
• Sentence 5
• Sentence 6
• Sentence 7
• Sentence 8

 

 

 

Paragraph 2

Collaboration is the foundation needed to establish proper communication and trust throughout society.¹ It allows for unique qualities presented at the individual level to work synergistically to create optimal outcomes.² Collaboration is essential to creating successful relationships that allow for prosperity and opportunities to achieve congruent goals.³ Collaboration encourages individuals to dedicate their personal time and efforts towards complex tasks to facilitate creative solutions and adequate problem-solving.⁴ Furthermore, collaboration creates an opportunity for individuals to serve a larger purpose in the collectives they work in.⁵ It is a requirement to sufficiently achieve daily activities of life.⁶ The versatility of collaboration allows for its ability to be equally necessary irrespective of the difficulty of the presenting issue at hand.⁷ Without collaboration, individuals will struggle to engage with one another, impeding the advancement of society as a result.⁸ 

 

Do you think this paragraph was human-generated, AI-generated, or a combination of the two?

1. Human-generated
2. AI-generated
3. Both

If you answered b or c, please check off the sentences that you think were AI-generated? The superscripts in the paragraph correspond to each sentence number.

• Sentence 1
• Sentence 2
• Sentence 3
• Sentence 4
• Sentence 5
• Sentence 6
• Sentence 7
• Sentence 8

 

Paragraph 3

Collaboration plays a vital role in achieving shared goals and driving collective success across various domains.¹ Through collaboration, individuals and groups can leverage diverse perspectives, skills, and resources to tackle complex challenges more effectively.² By fostering open communication and cooperation, collaboration enhances problem-solving abilities and promotes innovative thinking.³ Moreover, collaboration cultivates a sense of belonging and mutual support, fostering stronger relationships and teamwork.⁴ In academic settings, collaborative research efforts often lead to groundbreaking discoveries and advancements in knowledge.⁵ In business, collaboration facilitates cross-functional collaboration, driving organizational growth and adaptation in dynamic markets.⁶ Within communities, collaboration strengthens social bonds and empowers individuals to address shared concerns and pursue common objectives.⁷ Ultimately, collaboration is essential for creating positive change, fostering resilience, and building a more interconnected and equitable world.⁸

 

Do you think this paragraph was human-generated, AI-generated, or a combination of the two?

1. Human-generated
2. AI-generated
3. Both

If you answered b or c, please check off the sentences that you think were AI-generated? The superscripts in the paragraph correspond to each sentence number.

• Sentence 1
• Sentence 2
• Sentence 3
• Sentence 4
• Sentence 5
• Sentence 6
• Sentence 7
• Sentence 8

 

Post-Experiment Survey

Please answer the following questions.

1. Did you find it more difficult to detect:
   1. Human-generated text
   2. AI-generated text

 

2. On a scale of very easy to very difficult, how hard was it to detect human-generated text?
   1. Very easy
   2. Easy
   3. Neutral
   4. Difficult
   5. Very difficult

 

3. On a scale of very easy to very difficult, how hard was it to detect AI-generated text?
   1. Very easy
   2. Easy
   3. Neutral
   4. Difficult
   5. Very difficult

 

4. How many times did you read paragraph 1?
   1. Once 
   2. Twice

 

5. How many times did you read paragraph 2?
   1. Once
   2. Twice

 

6. How many times did you read paragraph 3?
   1. Once
   2. Twice

 

      7. What types of AI detectors do you use to grade students’ assignments? Please list their names below.

________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

       8. What clues did you use to differentiate between human-generated and AI-generated text?

________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

 

Results:

Please refer to the Attachments section to review my graphs.

 

Materials:

- The three pargaraphs and survey given to the teachers.

Analysis:

Paragraph 1 - Mixed

The first paragraph given to the teachers was the mixed text that combined both AI and human content. Following a maximum of two reads for each paragraph containing eight sentences, the teachers were asked to identify if each piece of text was a) human- generated, b) AI- generated, or c) both. If the options of AI-generated or both were chosen, the teachers were asked to check off which sentences they believed contained AI. 

Regarding the categorization made by the teachers after reading paragraph one, 23% of the teachers identified the paragraph as containing both AI and human content. However, the majority of the teachers (68%) identified the paragraph as being fully AI-generated, and the remaining 9% believed the paragraph was human-generated. To further interperet the teachers' judgement, they were asked in which sentences did they detect AI content. The mixed paragraph was evenly distributed between AI and human content, where the first four sentences were AI-generated and the last four were human-generated. The correct answer when selecting AI sentences was sentences one to four. The majority of teachers selected the last four sentences as containing AI, leading to 82% identifiying none of the AI sentences. 14% of the teachers correctly identified one out of four sentences, and 4% correctly guessed two out of four sentences. None of the teachers were able to correctly identify three or four of the sentences. The findings described above demonstrate a flaw in the teachers' method of detection. While majority were able to identify the paragraph as containing AI content, they were unsuccessful in correctly differentiating between the human and AI sentences. The consequence of mistaking student work for AI can cause true AI-generated pieces to go undetected and ultimately lead to no appropriate intervention. This highlights the need of providing teachers with learning tools to educate them further when dealing with a piece of text that has startegically integrated AI content in it. In classroom environments, educating students on citations and how to correctly complete them is also crucial in emphasizing the issue with paraphrazing or using AI without stating so. 

Paragraph 2 - Human-generated

The second paragraph consisted of the human-generated text. 77% of the teachers identified this paragraph as being both AI- and human-generated, with 14% correctly identifying this paragraph as being human-generated, and 9% selecting AI-generated. When determining which sentences contained AI content, the correct answer was none. Only 9% of teachers correctly identified zero AI-generated sentences in this paragraph, while 32% selected four out of the eight sentences. These results reinforce the need of increased education for teachers on detecting AI content. A teacher accusing a student of using AI can lead to ethical consequences, including being unjust towards the student. Suspicion of plagiarism in an original piece of text could have a negative effect on a teacher's job and students' well-being. This also reiterates the importance of AI-detection. Pieces of AI content left undetected and familiarized as human greatly impacts academic integrity in an education setting. As a result, trust, individuality, and information retention become significant concerns in the classroom. This indicates my hypothesis was correct, in which the teachers did struggle when trying to differentiate AI and human content.

Paragraph 3 - AI-generated

Lastly, the third paragraph was AI-generated. 43% of teachers suspected that there was AI content in this paragraph, identifying it as being both AI- and human-generated. However, nearly half of the teachers (48%) identified it as being human-generated, while only 9% correctly identified it as being AI-generated. When asked to select which sentences contained AI, the correct answer was all eight sentences. 50% of the teachers incorrectly answered zero out of eight sentences, and 9% of the teachers correctly identified all eight sentences. The judgement skills of the teachers in detecting AI content was flawed. Half of my sample identified a fully AI-generated piece of text as being written by a human which disregards academic integrity. This inaccuracy enables a path for plagiarism to integrate itself into the classroom and escape without detection. How can students effectivley learn in the classroom and retain knowledge when they are aware of the intelligence and advancement of AI that is easily at their disposal? If a teacher does not use an online detector, or uses one that is not completely accurate and does not focus on oral assesment, how can they rely on their judgement? Regulation and rules must be created for the use of AI for assignments to maintain academic integrity. Teachers need to be informed of more accurate methods of detection and how to better interpret AI text, and students need to be aware of the consequences of using AI. Overall, better comunication, awareness, and education on this topic is required to maintain academic integrity, while appropriately integrating AI into the education system.

Survey

When asked whether human-generated or AI-generated text was more difficult to detect, 64% of the teachers selected human-generated text and the remaining 36% selected AI-generated text. Again, this validates my hypothesis as the teachers did find it difficult when asked to detect and identify the different pieces of text. To elaborate on these findings, the teachers were asked to rate the difficulty level of detecting human-generated text. The majority of teachers (59%) reported that it was difficult to detect human text. 23% of the teachers also felt that it was equally easy and difficult (they selected the "neutral" option) to identify human-generated text, while 0% found it very easy to do so. They were then asked to rate the difficulty of detecting AI-generated text. 36% of the teachers found it to be difficult, 32% found it to be easy, and 32% took on a neutral stance. The teachers' responses to these questions highlight that they struggled during the experiment to correctly identify original pieces of text, which threatens academic integrity. 

To further evaluate the difficulty in detecting this content, the teachers were asked how many times they read each paragraph. For paragraph 1, 96% of the teachers read it twice while only 4% read it once. This emphasizes the difficulty in differentiating between AI and human content in a single piece of text. In paragraph 2, 23% of teachers completed one reading of the paragraph and 77% had to read it twice. The teachers felt slightly more confident in reading the human-generated text, but many of them still confused it with AI-generated text, as per my earlier analysis. Lastly, 50% of teachers read the third paragraph once and the other half read it twice. This once again reinforces how teachers found it difficult to detect AI content.

I also asked the teachers which AI detectors they use when grading their students' work, and what clues they used to differentiate between the human-generated and AI-generated text in the experiment. Listed below are the most common responses I received:

• Turnitin (D2L) - This was previously used by teachers.
• Written work/discussion and interviews for assessments during class time - This indicates that some teachers do not use any technology.
• ChatGPT - This requires payment to use.
• No detectors and/or knowing the student - This raises potential ethical and privacy concerns.

What clues did you use to differentiate between human-generated and AI-generated text?

• Paragraph and sentence structure - Teachers look for the terminology used and whether there are repetitions of words or ideas in the text. 
• Transition words - This involves students using complex adjectives and adverbs.
• Language difficulty - Some of the content may appear overly sophisticated for high school students.

The previous responses demonstrate a diverse range of AI-detection techniques that some of the teachers at Lester B. Pearson High School use in their classroom. As AI continues to grow in popularity and more students begin to use it for their assignments and projects, it is crucial that students and teachers alike uphold academic integrity when interacting with AI.
 

 

Conclusion:

In conclusion, my hypothesis was correct. I predicted that teachers at Lester B. Pearson High School would struggle to evaluate text that contained AI-generated content, human-generated content, and a combination of both types of content. The purpose of this experiment was to assess the effect of AI on academic integrity and evaluate the accuracy of detectors and teacher judgement. AI is technology that we feed with information and can be mistaken as simplistic or overly sophisticated. The first paragraph used in my experiment contained an equal number of AI-generated and human-generated sentences. By incorporating both types of content in a single text, majority of teachers were unable to distinguish between human and AI content. When asked which sentences contained AI content, most teachers did not identify any of the appropriate sentences. For the second (human-generated) paragraph, majority of the teachers found it difficult to detect the human content, labelling the paragraph as including both AI and human content. A potential justification for this classification stems from the teachers's reponses in the post-experiment survey. Many teachers perceive the use of sophisticated and complex word choice as being AI-generated content. Basing their decision off of prior knowledge and experience with various student levels in high school, the teachers believed that their current students would not be able to write a piece of text similair to this. Lastly, the teachers' responses to the third paragraph truly grasped my attention. Only two teachers correctly classified this text as being AI-generated, while nearly half of the teachers identified it as being human-generated. These results highlight the need for better education regarding this topic. If a student had used AI to write an assignment like I did for the AI-generated and combination paragraphs in my experiment, it could go undetected and the student would continue to engage in this behaviour. When assessing student work, unrully suspicion and accusation of plagiarism of an original piece can result in major issues. More learning tools  need to be provided for teachers to improve the accuracy of their judgement and detecting skills. My interview with Dr. Eaton, a professor at the University of Calgary, informed me that previous research has proven online detectors are not accurate or effective, but are advertised to show the opposite. She expressed her and the University of Calgary's disapproval and disuse of these detectors. Also, Dr. Eaton stated how the key to maintaining academic integrity while adapting to AI use does not consist of teachers immediatley banning the technology and labelling it as negative. Instead, there should be an increase in more resources like seminars and workshops for teachers to increase their knowledge on this topic, rather than just receiving funding for these inaccurate detectors.Overall, my project was a success, and I was able to evaluate the accuracy of detection methods and the relationship between AI and academic integrity.

 

 

 

 

 

 

 

 

 

 

 

 

Applications:

I really enjoyed my project. I contacted the Calgary Board of Education (CBE) education director, Aubrey Fletcher, to conduct an interview with her. She informed me via email that she is only available to meet the week of March 18th. I am looking forward to hearing her responses to my interview questions, and adding them to my project in the near future. A question I would like to address would be, "can CBE provide coverage for more accurate online detectors that teachers can use?" Some of the teachers expressed how the detectors they use either require payment or are randomly chosen from Google. The results of my experiment are evidence that there needs to be better access to learning tools and resources for teachers to increase their knowledge on this topic. Also, I believe it would be interesting to expand my project in the future by adding oral assessments to the experiment. This would entail evaluating detection methods for oral assignments and presentations. My project is a useful and effective tool in spreading awareness about this topic which is crucial to both teachers and students. This project could be applied in guidelines when discussing the school grading system from a CBE standpoint, or could be used as a framework for appropriate AI use for future projects at the Calgary Youth Science Fair. 

 

Sources of Error:

The environment that the experiment took place in could be my biggest source of error. Each teacher was supposed to work independently and not share answers or ideas. I tried to prevent the teachers from sharing thoughts and discussing questions by answering them myself. My science coordinator also reminded them to work alone and answer to the best of their knowledge. Although the environment itself was the same (the classroom and invigilators did not change), due to logistical reasons, not all the teachers were able to complete the experiment at the same time.

Bailey, John. “Ai in Education.” Education Next, December 20, 2023. https://www.educationnext.org/a-i-in-education-leap-into-new-era-machine-intelligence-carries-risks-challenges-promises/. 

“What Are the Advantages and Disadvantages of Artificial Intelligence (AI)?” Tableau. Accessed March 13, 2024. https://www.tableau.com/data-insights/ai/advantages-disadvantages#:~:text=The%20advantages%20range%20from%20streamlining,lack%20of%20emotion%20and%20creativity. 

“Definitive Guide to AI Platforms.” Anaconda, July 10, 2023. https://www.anaconda.com/definitive-guide-to-ai-platforms. 

“Why Teacher Intelligence Will Always Matter More than Artificial Intelligence.” Educational Records Bureau, September 13, 2023. https://www.erblearn.org/blog/ai-wont-replace-teacher-intelligence/#:~:text=Yes%2C%20generative%20AI%20can%20be,it%20cannot%20replace%20human%20teachers. 

“Ai Content Detector: Ai Detector: Chatgpt Detector.” Copyleaks. Accessed March 13, 2024. https://copyleaks.com/ai-content-detector?utm_source=www.ainauten.com&utm_medium=newsletter&utm_campaign=wie-ai-influencer-millionen-verdienen. 

Laskowski, Nicole, and Linda Tucci. “What Is Artificial Intelligence and How Does Ai Work?: Definition from TechTarget.” Enterprise AI, November 13, 2023. https://www.techtarget.com/searchenterpriseai/definition/AI-Artificial-Intelligence. 

“The Use of AI-Detection Tools in the Assessment of Student Work.” Learning, Teaching and Leadership, May 6, 2023. https://drsaraheaton.wordpress.com/2023/05/06/the-use-of-ai-detection-tools-in-the-assessment-of-student-work/. 

Google Images

Interview- Dr. Sarah Eaton

Special Thanks

I would like to first thank all of the teachers that participated in my experiment. Without them, I wouldn’t have been able to acquire the results that I did and discover the accuracy and/or inaccuracy of detectors. Secondly, my science coordinator Mr. Webster played an important role in giving me his class space to perform my experiment and providing me with the emails of potential teacher participants. He was always available if I needed any help and he constantly checked in with me to remind me of dates and deadlines. Thank you Dr. Sarah Eaton for taking the time to talk to me and answer my questions. Her feedback helped me understand my experiment and topic better. Lastly, I would like to thank my parents for always encouraging me and helping me be the best I can be.