Combatting Contract Cheating in Exams via AI Technology

It appears that universities across the world have become more aggressive when it comes to ensuring that students do not cheat when taking tests online. In light of Covid regulations in the past that allowed students to take exams remotely, thus allowing them to outsource the test to paid exam takers, and how these Covid policies are still in place for most universities, they have needed to take extra steps to prevent remote cheating by students. One of these methods is by using stealth software to monitor the academic integrity and honesty of the students. Contract cheating in exams is well known to the universities and their exam developers. Paying someone to take the test for you has proven to be more difficult, but not impossible for the IT guys of the university to check and prove.

These days, the universities use lockdown browsers in face to face exams. Preventing the use of the internet while taking the test. For remote testing, they now keep track of the regular student activity when compared to his geographical location when logging in to take any test. Using an AI, the tracks / clicks of the student during the exam time is traced and assessed. At the end of the trace, a report is then generated and submitted to the professor for all the students that took the test. Making it now quite dangerous to outsource / hire an academic writer to complete any given test for you. This also explains why some universities have a delay in the release of online test results for remote exam tests. IP addresses are checked for location in relation to the actual registered location and IP log history of the students. Where VPN is used or suspected, more advanced digital investigation is required before the release of exam results.

As far as I know, there are 2 known universities in 2 different parts of the world that do this, Miami University in the US and Deakin University in Australia. Who is to say that other universities across the globe are not using the same process when it comes to keeping remote track of their students during exams? Yes, the system has helped both universities to prove not only plagiarism, but also exam outsourcing by their students.I believe other universities across the world also employ these tactics in keeping track of students in-class and remote exam activities.

Be careful when hiring someone to take your test for you. The university may have a system in place to catch you doing it. Your future is not worth it. My suggestion is, take the test personally, regardless of what the results might be for you. Do this for yourself, for your academic integrity. Fail the test if you know you cannot pass it. Retake the test if given a chance. Be honest with yourself, the professor, and the university. Never place your academic integrity on the line because this will endanger your future as a student and future professional. It will be difficult to find another university to accept you as a student when your record includes a reference to proven academic dishonesty.

If students are really intent on cheating, they'll just hire someone to help them in real time and use some other method besides their computers to transmit the exam questions and receive the answers. The only way to stop that would be to require a live camera view of students during exams in addition to IP monitoring and keystroke tracking.

The AI technology surely has provided ease for the students. However, still platforms like Turnitin plagiarism checking websites have add-on of AI detecting tool for checking the content whether they are copied from an AI platform or not. AI has it's advantages but only if the students use them positively. The student's can use the platform for information gaining purpose and should not become dependent to the platform but should use them for knowledge gaining purpose.

As I've explained several times, AI doesn't plagiarize in the simplest, most obvious traditional sense, because AI programs don't actually copy and past verbatim content, which is what plagiarism scanners are designed to detect. AI programs frequently lift ideas from sources without crediting those sources, including totally inappropriate sources, such as amateur blogs and grade-school textbooks, and they also fabricate non-existent "sources" and content, which is known as AI "hallucination." However, when AI programs do lift information from uncredited sources, they still "rewrite" it in different words, which makes it original "writing" that inoculates it against detection by traditional plagiarism scanners. Futurist Ray Kurzweil has explained that the phenomenon of AI hallucination is attributable to the fact that AI programs don't yet have the ability to respond appropriately when prompted for information that they can't find. Instead of, essentially, admitting that they don't have any relevant information, AI programs fabricate information, instead.

I have read that plagiarism detection tools are now being tweaked to sense when a paper was quite possibly written by an AI. I find this laughable since there are times when an ESL writer churns out papers that read like badly written AI output.

I guess AI can do it though. One thing is sure - AI doesn't make grammar/spelling mistakes. It may write nonsense, but nonetheless its writing is (technically) error-free.

There are scanners that are intended to determine whether a piece of writing was generated by AI or written by a human writer. My understanding is that they're extremely unreliable because they're often wrong. However, they're totally different from traditional plagiarism scanners, which are useless for that purpose, because plagiarism scanners are designed to identify copied strings of text by matching them to the sources from which they were copied. Since AI programs don't actually copy existing text, plagiarism scanners are useless to identify AI writing vs. human writing.

Most professors recognize AI writing, precisely because of the combination of technically good "writing" and the glaring total absence of any substance and/or by obvious deflection away from any essay prompt that requires analysis and toward historical fluff. Generally, students who write as well as an AI program also express coherent thought in their essays. In fact, AI writing is so obvious to professors because it the exact obvious of the real pattern exhibited by human students: Specifically, students are almost always worse at writing than at substantive learning.

@formerstudent
Sadly, the student will not be graded on grammar alone for their papers. The papers still need to make sense and show a degree of actual research in the presentation. That is what the professors grade on. That is also why AI is easily spotted by the trained eye. AI can produce 250 words of unusable information in some instances because it cannot produce a logical thought on its own, nor analyze the information it finds online. It is not gifted with the sense of reading comprehension the way a human writer is. That is the main problem with AI written essays at this time.

Colleges and universities are increasingly adopting sophisticated methods to detect AI-generated content in student submissions. While tools like Turnitin have long been staples in academic integrity, they are now being enhanced with advanced AI detection features, rather than being universally abandoned. Institutions are utilizing a combination of established platforms and newer, specialized AI detection services.

Many free and freemium AI detection websites are available, offering to distinguish between human and AI-generated text. Examples of such tools include:

- GPTZero: Known for its focus on academic prose and offering detailed analysis, including sentence-level highlighting of AI-generated text. It provides free basic access with premium options for more extensive use, and integrates with learning management systems.

- Originality.ai: This platform is recognized for its high accuracy in detecting AI-written content and also offers plagiarism checking.

- Scribbr.com: Provides an AI detector that offers paragraph-level feedback and supports multiple languages.

- ZeroGPT.com: This tool gives a percentage score indicating the likelihood of AI authorship and highlights potentially AI-generated sections.

While some universities may provide access to the premium versions of such software as part of their institutional subscriptions, it is not a universal guarantee of free access to all premium features for all students.

...or they could just look for a whole lot of words that say practically nothing substantive and read about the same as a typical C-/D student-written essay...