Generative AI tools like ChatGPT have become increasingly popular since their launch towards the end of 2022. ChatGPT reached a million users in a span of five days. That’s because conversational AI produces quality human-like responses.
For example, ChatGPT’s human-like content is due to the data it is trained on. The earlier version of the product was trained on approximately 500 billion tokens, which are common sequences of characters found in the text. This capability of Generative AI helps users design working codes for different objectives and use cases.
Furthermore, ChatGPT can also be used for debugging existing code. In fact, ChatGPT was compared with standard automated program repair techniques and two deep-learning approaches to program repairs by researchers from Johannes Gutenberg University Mainz and University College London.
The researchers found that ChatGPT’s bug-fixing performance is measured up to the common deep-learning approaches. They also noted that its performance was better than those reported for the standard program repair approaches.
The advantage of ChatGPT lies in its unique ability to converse, giving it an edge over other approaches. Users can ask questions and receive answers during code creation, simplifying the process and providing developers with the necessary information.
However, AI-generated code may not be correct. For this reason, Stack Overflow, the popular developer forum, banned codes or answers written by ChatGPT.
Students who need help writing assignments turned to professional assignments and essay writing services, but with ChatGPT becoming popular, there are increasing instances of students using AI-generated text for academic purposes.
Similarly, with ChatGPT’s ability to produce codes, AI-generated code may be used in coding assessments. While this may work in favor of the candidate during the test, it can become challenging to adjust to a job position where the candidate needs to work on generating complex codes regularly.
The use of ChatGPT also hampers an organization’s ability to assess a candidate’s analytical thinking and technical skills.
As discussed, the advent of Generative AI language tools, such as ChatGPT, Bard, and GitHub Copilot, has amplified apprehensions regarding the possibility of cheating in online assessments.
Mercer | Mettl has introduced the ‘AI Sense’ score on its coding platform to tackle concerns around the use of AI-generated code and AI plagiarism. It indicates the likelihood of a candidate submitting AI-generated code by apps like ChatGPT, Bard, or GitHub Copilot. This score further strengthens the integrity of the test.
AI Sense uses an in-house algorithm to detect AI-generative code. The algorithm uses Generative AI’s power to sense and identify instances where the candidate might have used machine-designed code.
The AI Sense score is displayed along with the plagiarism score and helps recruiters identify candidates whose code is AI-generated, reducing the chances of mis-hires.
To further ensure candidates do not indulge in any unethical means to gain an unfair advantage, recruiters can use Mercer | Mettl’s Secure Exam Browser, with its suite of anti-cheating measures. The browser locks the system and turns off any additional software. The recruiting organization can safelist sites and software that candidates can access during the assessment.
Similarly, organizations can use Mercer | Mettl’s Online Interview Platform for remote interviews. Again, it comes with anti-cheating measures, ensuring the interview’s integrity and reducing the chances of AI plagiarism or the use of AI-generated text.
Hiring teams across organizations can also utilize Mercer | Mettl’s SecureProctor. It is a remarkable tool that can be used to ensure assessment integrity. Mercer | Mettl SecureProctor uses advanced technology, like multi-factor authentication, to eliminate student impersonation risk. A comprehensive human-based and AI-based proctoring technology ensures live proctoring, auto proctoring, and record and review proctoring. With Mercer | Mettl’s proctoring solution, employers can be confident that they are evaluating candidates based on their skills and abilities without biases.
ChatGPT and similar AI apps can be a boon or a bane, depending on the intention behind their usage. Candidates who use ChatGPT during hiring assessments can easily masquerade as expert programmers and come across as ideal hires. This could lead to unsuitable new hires who are an inadequate fit for the organization. In order to overcome this, hiring professionals can use assessment platforms such as Mercer | Mettl’s coding assessments platform with AI Sense to identify applicants using ChatGPT and prevent any negative impact on recruiting.
Originally published July 28 2023, Updated July 28 2023
Vaishali has been working as a content creator at Mercer | Mettl since 2022. Her deep understanding and hands-on experience in curating content for education and B2B companies help her find innovative solutions for key business content requirements. She uses her expertise, creative writing style, and industry knowledge to improve brand communications.
The accelerated pace at which businesses are rushing toward digitization has primarily established that digital skills are an enabler. It has also established the ever-changing nature of digital skills, and created a need for continuous digital upskilling and reskilling to protect the workforce from becoming obsolete.