Blog
Beyond the Prompt: How to Hire the Graduate, Not the AI
For candidates, AI lowers barriers by supporting faster, more polished applications. For recruiters, it can improve efficiency through automation and screening support but also make it harder to distinguish genuine human capability from AI responses. The key to success is having hiring processes that preserve fairness and still reveal authentic skills, potential, and critical thinking.
Share
Share
Is AI in graduate applications really a problem?
70% of candidates now use AI at some stage of the hiring process¹, from researching employers to drafting CVs and cover letters. Its usage is also prevalent during interviews, and online assessments² so the direction of travel is clear—more AI, more polish, and more difficulty separating genuine capability from generated content.
From my experience working with customers looking to improve their graduate programs, there are three major shifts that are happening across industries.
1) “Cookie-cutter” applications
AI application tools have made it easy for candidates to apply to hundreds of roles with a single click, which drives applicant numbers up without driving up quality. Many candidates are using the same tools, prompts, and optimization techniques so applications can all look thoughtful and well considered, even when the underlying capability is harder to see.
2) Application volume is rising while signal quality declines
Employers saw a 26% increase in applications in 2025³. AI tools have dramatically lowered the effort required to apply for roles at scale, enabling candidates to submit dozens or even hundreds of applications quickly. While this increases candidate reach, it also creates greater screening pressure for employers and can dilute overall application quality.
3) The emergence of “ghost candidates”
By 2028, one in four job candidates could be fake or misrepresented through AI-generated profiles, fabricated credentials, or proxy interviewing practices⁴. This creates growing concerns around candidate authenticity and increases the importance of validating skills and identity earlier in the hiring process.
Why banning AI doesn’t work and the smarter approach
Adding extra friction such as blanket bans on AI and demands that candidates prove they have not used it can damage the candidate experience and make the organization seem adversarial rather than progressive. That is a poor signal to send to digitally fluent graduates, costing the organization the exact caliber of talent that employers want to attract. I’ve also seen organizations closing applications early once volume thresholds are reached, but that tends to reward the fastest AI-driven applications rather than the candidates who have taken time to consider the role properly.
Organizations should accept that AI is not going away, and design graduate hiring processes that account for AI use, larger candidate pools and more polished applications, but where strongest candidates can still rise to the top. The aim should be verification rather than prevention, so that AI assistance does not determine the final outcome and there is enough genuine human evidence at relevant stages.
Building a stronger stage-by-stage process
Rather than asking one stage to do everything, each stage should play a different role. Early on, when volumes are high, the goal is to identify the factors that genuinely separate strong candidates from the rest. That is where strong assessment tools do the heavy lifting and sift effectively. Clear structured criteria, a shared understanding of what is being assessed, and careful thought about cut scores all matter here.
Comparing performance across applications, assessment, and interview stages can help reveal inconsistencies which may not be spotted if evaluating potential AI-generated answers in isolation. This is where structured live interaction becomes especially important. On-the-spot reasoning questions and asking candidates to walk through their logic can create moments where genuine capability has to show up in real time.
Use the right assessment tools
When evaluating assessment tools for graduate recruitment, there are some key things that should be considered, such as how the tool is designed to resist AI-generated responses. The best assessments have large item banks and randomized question order to help reduce risk and provide an auditable view of candidate behavior including video, audio, and on-screen oversight, as well as signs such as tab switching, secondary-screen access and other plagiarism detection.
Ask whether the provider actively tests its own assessments against AI tools, as the most rigorous providers run regular testing to understand vulnerabilities and address them proactively. It is also important to check vendor compliance with legal, regulatory, and regional guidelines, especially with more deployer obligations, such as those posed by the EU AI Act coming into effect.
A well-designed AI-resilient graduate hiring process does not try to eliminate AI. It makes AI assistance irrelevant to the final hiring decision by creating enough genuine human signals for the right graduates to stand out.
Sense-check your own approach using our Checklist for Graduate Recruiters and review where your process may be most vulnerable to AI-generated responses.
² ISE 2025
Author
