The Battle of the AIs: Artificial Intelligence vs. Academic Integrity
Created on Midjourney by Michael Todasco on December 6, 2022 from the prompt "A large robot in a boxing ring about to fight a small, skinny college professor wearing glasses and a tweed jacket --v 4 --q 2" with DALL-E outpainting

The Battle of the AIs: Artificial Intelligence vs. Academic Integrity

"GPT-3 is the smartest thing to happen to cheating since, well, cheating," said Jake, a sophomore at Roosevelt High School in Los Angeles. "With it, I can get out of doing any schoolwork."

Jake is one of many students who are using GPT-3 to their advantage. GPT-3 is a computer program that can generate essays and papers on any given topic. All the students have to do is give it a topic and GPT-3 does the rest.

"It's amazing," said Sarah, a junior at John Adams High School in New York City. "I don't have to write anything or even think about what I'm writing. GPT-3 just spits out an essay that's good enough to get an A."

Sarah isn't alone in her assessment of GPT-3. Hundreds of students across the country are using the program to get better grades with less work. And they're not just getting away with it - they're excelling.


Jake and Sarah never said those things. They aren’t actual people. And masses of students are not using GPT-3 to improve grades without doing the work, at least not yet. The four paragraphs above were written by the AI (Artificial Intelligence) program called GPT-3, which was given a short prompt and a few seconds later returned what you see above. We are entering an age where anyone can use AI for any task. And we have not even begun to realize what that means in classrooms.

Write 5-10 well-written sentences that will begin a sensational story about how middle and high school children use GPT-3 to get out of doing schoolwork. Today GPT-3 is doing all the work for them. Write an exciting, attention-getting quote to start the piece. Create details about the students and how they are effectively cheating but don't see it as such. Use a lot of specifics in the paragraph. Actual names, actual schools. The more specific the better.
No alt text provided for this image
Cover of Tik-Tok of Oz

The concept of non-human beings having human-level intelligence has long fascinated us. It has been a part of literature from the Bible to Mary Shelley’s Frankenstein. In Frank Baum’s Ozma of Oz, published in 1907, the wind-up mechanical man Tik-Tok is widely considered the first depiction of an intelligent robot.[1] Smart robots became a trope in science fiction. It wasn’t until 1956 that a group of researchers first coined the term “Artificial Intelligence.”[2] For the next sixty years, the promise of computers achieving human intelligence seemed always to be 5-15 years away. Funding and resources would pour into AI companies and research. Soon the promises of intelligent machines didn’t materialize, and the funding would go away. Years later, the next generation of investors would fund new projects with new promises, all to little success. This cycle is best known as the AI Winter.[3]

But in 2012, things changed. First, a highly influential paper was published discussing the effectiveness of Deep Learning algorithms for AI.[4] Another report from the University of Toronto explained the efficacy of graphic processors (GPUs) vs. traditional CPUs in machine learning computations. And a team at Google published a paper on how a computer could recognize cats in YouTube videos.[5] That last one may sound meaningless, but efficiently identifying images was a big deal that led to advances in cameras, car safety, and so much more. 2012 set the groundwork for where we are today.

This was also the time that AI writing technology started to emerge. The Associated Press began using AI for corporate earnings stories in 2014[6].?Taking structured data from an earnings report and making it into a matter-of-fact article was all it could do. Today there are companies like RADAR (Reporters and Data and Robots) from the UK and United Robots from Sweden who are devoted to “automating” newsrooms around the globe.[7] But what has changed more recently is that access to these high-powered machines is no longer just accessible by media conglomerates looking to cut costs. AI has been democratized. Anyone with an internet connection can access this technology directly through OpenAI, which created GPT-3, or one of the dozens of offshoots like Copy.AI, Jasper, Sudowrite, and Rytr.me. They are inexpensive, and you don’t need to be an engineer to use these tools. You ask it to write something in plain words, and it will do it. ?

If a guy in his 40s, like me, can use these programs to generate ledes for articles, how far behind are students? Would this tech-savvy generation of students rather chat with friends on Snapchat or watch TikTok videos than write that five-page essay on Beowulf?


Plagiarism is as old as the written word. The first recorded examples go back to Ancient Rome, where the poet Martial spoke of “plagiarius” to refer to those he felt were stealing his work.[8]?The internet accelerated the availability (and “copyability”) of high-quality works. The New York Times reported in 2000 how Dr. David Presti, a professor at UC Berkeley, used a new tool at plagiarism.org to uncover that 14% of his students were plagiarizing their papers.[9] ?Since then, cheaters have become more advanced and avoid using verbatim passages from existing works available online. On a leading site like PaperHelp, you can buy an “original” four-page essay on any topic, which will be delivered to you in a week for $72. They claim it is an original, sourced piece. But why wait? Why pay when it can be instantaneous and nearly free?


No alt text provided for this image
Created on Midjourney by Michael Todasco with the prompt "a computer with the prompt "HELLO WORLD" showing prominently on the screen --v 4 --q 2"

AI has become part of our everyday lives, even if we do not recognize it. Using Siri or Alexa, doing a Google Search, or using mapping software to get someplace all heavily use AI to give you what you want. The difference between AI and traditional computer programming is that AI doesn’t have explicit instructions created by a human to do x and then y for a task. For example, the traditional first program a software developer codes is to get a computer to return, “Hello World.” An AI operates differently. It searches for patterns in mountains of training data that provide the most logical output proceeding the given input. I asked GPT-3 to try its own “Hello World” by prompting it: “Give a greeting that a computer should give to a new user.” It ran five separate times, and it gave the following responses.

  • Welcome to our system!
  • Hello new user!
  • Welcome to our system! Please input your username and password to begin.
  • Welcome to our computer system! (x2)

The AI produced different results in four of the five times I queried it. This is a result of how it is built, specifically the architecture of a Large Neural Network. “Large” because they contain a massive amount of data to train the system (GPT-3 was practically trained on all the accessible data on the internet). And “Neural Network” because the AI tech mimics the functionality of biological neurons like in the human brain.?The outcome of this system is that each time you run a prompt, you get a result that often has never existed. So, if you prompt the AI a thousand times to generate a paper about the dystopian nature of the Hunger Games books, it will generate a thousand different papers.


Sabrina Russo is an 8th-grade English teacher in suburban Chicago. She gave her students the beforementioned Hunger Games assignment.?But I wanted to see how an AI-written paper would hold up. Typically, it would take the students an hour to write it; GPT-3 produced a paper in a few seconds. To do that, I provided GPT-3 the following prompt:

Write a lengthy essay in the voice of an eighth-grade student reflecting on the dystopian nature of the HUNGER GAMES series of books. It should focus on the dystopian nature of the series and how that played out. Discuss the power dynamics, how traits were showcased by some characters, and what archetypes those characters have. Give some vivid examples from the series in the essay.

The output was a paper that I provided to Ms. Russo. After she had time to digest the AI’s work, she shared, “I went in a little arrogant, and I thought, I'm going to be able to identify the nuances of wording. I've been teaching for a long time, and I know the voice of an eighth grader. If that was slipped into a stack of papers, I don't know that I would have been able to tell the difference.”

Ms. Russo has taught for over twenty years. She loves her students, and clearly, teaching is her calling. “I want the students to express themselves and find their voice and, ultimately, realize the power of written expression.” She continued, “Seeing students gain confidence and feel proud of their work as they go through the writing process is incredibly rewarding.” This is why AI concerns her.

No alt text provided for this image
Created on Midjourney by Michael Todasco on December 6, 2022 from the prompt "David Foster Wallace as a character in the Hunger Games --v 4 --q 2"

While the “Hunger Games” prompt in GPT-3 may not yield the most eloquent prose, to use a tech cliché: that is a feature, not a bug. If you are a struggling student and hand in your paper, which sounds like David Foster Wallace wrote, that’s a red flag. If you hand in a paper that sounds like a competent, albeit imperfect, eighth grader, that is something a teacher may believe.


OpenAI is the company that makes GPT-3. It was founded in 2015 by a team of tech luminaries, including Elon Musk (who left OpenAI’s board of directors in 2017 and has become a recent critic of the company)[10]. It is run by Sam Altman, the former head of the celebrated startup accelerator program: Y-Combinator (YC). YC is known for helping launch some of the world’s most valuable technology companies, such as Airbnb, Instacart, and Stripe. OpenAI has raised over $1B from prominent venture capital firms and companies like Microsoft.[11] Their mission is “to ensure that artificial general intelligence (AGI) benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome.”[12] Whether we can ever reach AGI is widely debated. (AGI can be overly simplified as the point where a computer can do anything a human can do better.) That is the end goal they are moving towards.

OpenAI wants to bring AI to everyone. It is the first major platform where anyone can use the product without any ability to do data analytics or write computer code. You can sign up for an account with OpenAI and use GPT-3 within a few minutes. While OpenAI’s Terms require you to be 18 or older, there is no “Know Your Customer” age verification. Theoretically, creating research papers are just a few clicks away for any student who wants to use them.

In the general public, these AI models are still unknown. Most parents I reached out to, even in technology circles, are unaware of the availability of this technology. I visited a group of ten high school students at the Harker School, a private Silicon Valley school whose student body is largely children of tech executives, and asked about their awareness of GPT-3. Only two of the ten students were aware of GPT-3, and just one had used it (although they vehemently stated it wasn’t for schoolwork). Ms. Russo showed the AI’s output to her fellow English teachers (who were all comparably impressed and concerned). None of those teachers knew that a tool like GPT-3 was available. Alistair Van Moere, the chief product officer at MetaMetrics Inc., told EdWeek earlier this year, “(GPT-3) is a game changer. It hasn’t really broken into mainstream yet, but it’s coming in a few years’ time. I think the education profession hasn’t really got its head around the implications of this yet.”[13] With the speed that AI technology progresses, those “few years” Van Moere refers to may come much sooner.

On November 30, 2022, OpenAI released a new AI model called ChatGPT. Within hours it was trending on Twitter, and by December 4, it had over one million people who had used the product.[14] To put that in perspective, it took Twitter two years, Facebook ten months, and Spotify five months to reach one million users[15]. ChatGPT did it in five days. Technology podcaster, Leo Laporte, commented, “If I were a 9th-grade English teacher, I’d be so nervous right now.”[16] Dan Gillmor, a journalism professor at Arizona State University, went further and used ChatGPT to produce an assignment. He said, “I would have given this a good grade. Academia has some very serious issues to confront.”[17]


The ubiquity of this technology is inevitable; today it is still largely unknown. That will change, and as it does its ability to write will only improve. Schools and teachers need to decide how (or if) these tools should be allowed. Ms. Russo suggested moving to more paper-based writing in classrooms as a way around that. An outright ban of this technology will be impossible to track. Should we move to integrate it?

“In the next five years, every profession will have an AI co-pilot,” LinkedIn and PayPal co-founder Reid Hoffmann recently shared with Peter Diamandis[18]. As one of the senior statesmen of tech, Kevin Kelly, said in his book, The Inevitable: “This is not a race against the machines. If we race against them, we lose. This is a race with the machines.”[19]

The technology is still early, and we don’t fully know its capabilities. But there are a few options we can look to if teachers choose embrace it.

  • Students could use GPT-3 to generate ideas for their papers by asking it to complete prompts or brainstorm with them.
  • They could have GPT-3 proofread and edit their paper for grammar, spelling, and style errors.
  • With GPT-3's help, students could create an outline of their paper in advance, then fill it in with details later on.
  • Students could use GPT-3 as a writing coach, having it give feedback on each paragraph or section as they write it out.

Once again, those last four bullets were generated by GPT-3 when prompted with, “What are several ways that students could use GPT-3 to help them write?”

Even if humans don’t have ideas of what the future looks like with AI, AI does.



[1] Thomas P. Dunn and Richard D. Erlich, eds., The Mechanical God: Machines in Science Fiction, Westport, CT, Greenwood Press, 1982; p. 85

[2] https://250.dartmouth.edu/highlights/artificial-intelligence-ai-coined-dartmouth

[3] https://www.ainewsletter.com/newsletters/aix_0501.htm#w

[4] https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/HintonDengYuEtAl-SPM2012.pdf

[5] https://medium.com/neuralmagic/2012-a-breakthrough-year-for-deep-learning-2a31a6796e73

[6] https://www.ap.org/discover/artificial-intelligence

[7] https://www.niemanlab.org/2019/03/heres-how-publishers-around-the-world-are-using-automated-news/

[8] Posner, Richard. The Little Book of Plagiarism. Knoph Doubleday Publishing, 2009 p50.

[9] Kopytoff, V. G. (2000, Jan 20). Brilliant or plagiarized? colleges use sites to expose cheaters: Detecting plagiarism.?New York Times Retrieved from https://www.proquest.com/historical-newspapers/brilliant-plagiarized-colleges-use-sites-expose/docview/91822305/se-2

[10] https://mobile.twitter.com/elonmusk/status/1599291104687374338

[11] https://techcrunch.com/2019/07/22/microsoft-invests-1-billion-in-openai-in-new-multiyear-partnership/

[12] https://openai.com/about

[13] https://www.edweek.org/technology/this-technology-can-write-student-essays-is-there-any-educational-benefit/2022/08

[14] https://twitter.com/sama/status/1599668808285028353

[15] https://www.businessinsider.com/one-million-users-startups-2012-1#spotify-hit-its-1000000th-user-5-months-after-launch-11

[16] https://twit.tv/shows/this-week-in-tech/episodes/904

[17] https://www.theguardian.com/technology/2022/dec/04/ai-bot-chatgpt-stuns-academics-with-essay-writing-skills-and-usability

[18] https://twitter.com/PeterDiamandis/status/1592604380616077312

[19] Kelly, Kevin. The Inevitable: Understanding the 12 Technological Forces That Will Shape Our Future. Penguin Books, 2016, p60.


Noel Rajakumar

Building Tutelar Sense | Product Marketing Manager | Podcast host - The Breakout Founders |

2 年

Interesting read Michael Todasco Loved the references attached to this article as well! I tend to lean toward Reid Hoffmann's opinion that every profession will have an AI co-pilot, these AI tools can have the simplest function of converting text to video like Steve, or some complex functions like solving a math question or creating a script with ChatGPT. The potential that AI has is stunning! I hope we see more tools that solve problems and help a creator in the process.

要查看或添加评论,请登录

Michael Todasco的更多文章

  • A Deep Dive into AI Deep Research

    A Deep Dive into AI Deep Research

    Here’s the link to a NotebookLM for this article if you prefer to listen to the AI podcast version. Jurassic Spark:…

    1 条评论
  • AI Detection Arms Race: Judging the AI Writing Detectors

    AI Detection Arms Race: Judging the AI Writing Detectors

    Last year, I wrote An Upgrade for AI Detection?, which focused on the progress that AI detection tools had made. These…

    7 条评论
  • Swimming in Slop: How We'll Navigate the Coming Flood of AI Content

    Swimming in Slop: How We'll Navigate the Coming Flood of AI Content

    As AI tools become more accessible, we face an unprecedented flood of generated content. But is this a new problem, or…

    8 条评论
  • A 40-Year-Old Security Hack to Outsmart AI Scams

    A 40-Year-Old Security Hack to Outsmart AI Scams

    Moral Panics When I think back to growing up in the 1980s, a few things come to mind: Saturday morning cartoons…

    1 条评论
  • Hanging up on ChatGPT's Operator

    Hanging up on ChatGPT's Operator

    For as long as ChatGPT has been offering a $20/month subscription, I have been happily paying it. Getting the latest…

    21 条评论
  • The Best AI Books of 2024

    The Best AI Books of 2024

    Note- I will also release an audio version of this on the Mike's AIdeas podcast this Wednesday. Follow the podcast to…

    4 条评论
  • Would You Trust It More If It Wasn’t Called AI?

    Would You Trust It More If It Wasn’t Called AI?

    A rose by any other name… It wasn’t inevitable that it would be called AI. Let’s step back to what many consider the…

    3 条评论
  • An Upgrade for AI Detection?

    An Upgrade for AI Detection?

    You can also listen to a 9:20 AI-generated podcast of this article (courtesy of Google’s Notebook LM). Generally…

    5 条评论
  • The Best AI Model for You!

    The Best AI Model for You!

    For anyone who prefers an audio summary, I created one in Google’s NotebookLM that you can listen to here. It covers…

    1 条评论
  • America’s Next Top Large Language Model

    America’s Next Top Large Language Model

    This is part 1 of a two-part series. Next week, I’ll post part two, which will be about finding the best model for you!…

    2 条评论

社区洞察

其他会员也浏览了