When a vacancy attracts hundreds of applications, using AI to help review them is an understandable choice. Hiring teams have limited time, and working through every CV can take attention away from interviews, assessment and keeping candidates informed.
AI tools can support tasks such as summarising CVs and scoring applications. But employers need to distinguish between making information easier to review and deciding who deserves consideration.
There is another side to this: applicants can use AI to write and tailor their CVs, while employers use AI to assess them. In that process, how much are we learning about the person behind the application?
For a hiring manager, the practical question is simple: how do you know your screening process is improving hiring, rather than just reducing the number of applications you have to read?
Where AI can be useful
Used carefully, AI can give recruiters a more manageable starting point. Depending on the tool, that could mean summarising relevant experience, bringing evidence against agreed criteria together, or highlighting areas that need clarification.
Consider a hiring team looking for a gameplay programmer. A useful screening assistant might identify the candidate’s stated engine experience, relevant projects and programming background, with links back to the original application. The recruiter can then check the evidence and decide what to explore.
That is a different responsibility from automatically rejecting everyone below a suitability score.
Before introducing a tool, be clear about the job you want it to do. Will it organise information, recommend candidates or make decisions? What can the hiring team check, and what happens to people it ranks poorly?
When AI screens an AI-written CV
A candidate can ask AI to rewrite their CV around a job description, highlight relevant experience and draft application answers. An employer can then use another AI tool to summarise, score or rank that application. Where automated screening questions are involved, AI could effectively be assessing answers drafted by another AI system before anyone has a conversation.
Using AI to improve a CV is not, in itself, a reason to distrust an applicant. Help with structure, language and clarity can make genuine experience easier to understand. Research on non-generative writing assistance has found benefits for jobseekers without evidence of lower employer satisfaction, although that does not establish the effects of fully AI-generated applications.
The concern is what the screening process rewards. If it places too much weight on polished wording or close alignment with the job description, could it favour a well-optimised application over a stronger candidate whose CV is less polished?
That is a risk to test for, rather than an assumption that every AI tool can be fooled. A CV rewritten to match a role still needs evidence behind its claims. Better presentation does not create additional experience.
For a games studio, the next step might be asking a programmer to explain a technical decision, an artist to walk through their contribution to a portfolio piece, or a producer to describe how they handled a delivery problem. Use consistent, role-relevant questions and proportionate assessments to explore what the applicant actually knows and has done.
The useful question is whether the candidate can substantiate their application. Trying to judge AI use from writing style alone does not answer that.
A screening tool cannot fix an unclear brief
One place to start is the job specification itself.
If the brief treats every preference as essential, the screening process may exclude people the hiring manager would otherwise consider. This risk also exists in manual recruitment. Automating the process makes it especially important to agree the criteria before applying them across a large applicant pool.
Does the person need experience with a particular engine on their first day? Is experience in the same genre essential? Would equivalent technical work in another setting be relevant?
Some requirements will be non-negotiable. Others may be preferences that have gradually become part of the standard job description. Those distinctions should be deliberate.
AI tools differ, and some assess meaning beyond exact keywords. It would be misleading to suggest they all work as simple keyword filters. Even so, employers should test whether their chosen system recognises the kinds of relevant experience they actually want to consider.
Games experience needs context
In games recruitment, a title or a list of projects is a starting point for assessment. It does not establish what someone personally delivered.
Consider these hypothetical candidates:
-
A programmer who has worked mainly with a proprietary engine but has relevant experience solving performance and gameplay problems.
-
An artist whose portfolio shows the required quality, although their previous job title differs from the advertised role.
-
A producer from a smaller studio who has managed a broad development remit but lacks the specific studio names listed as desirable.
None should automatically qualify. Each presents a question worth investigating before deciding whether their experience fits.
The same scrutiny should apply to apparently strong matches. A credit on a successful title does not, by itself, explain someone’s responsibilities, contribution or level of ownership.
Employers should ask how their screening process handles portfolios, project evidence and missing information. If the system cannot assess an important part of the application, that limitation needs to be visible to the person making the decision.
A ranking still needs evidence behind it
A candidate score can look precise without explaining much about suitability.
What does a high score represent? How closely the application matches the brief? Evidence of particular skills? A prediction about performance? Those are different things, and hiring teams should understand which one they are being shown.
Ask the provider what the score measures, how it has been tested for the intended use, and what its limitations are. Then check the results against real role requirements.
Human review also needs to be meaningful. A reviewer should have the time, information and authority to question the recommendation. The ICO’s work on automated recruitment highlights the importance of meaningful human involvement, transparency and monitoring for fairness.
Simply having someone approve the shortlist does not tell you whether suitable people were excluded earlier.
Check who is being rejected
Reviewing only the successful shortlist leaves a gap in your assessment of the process.
As a quality check, examine a sample of rejected or low-ranked applications against the agreed criteria. Are there candidates a recruiter or hiring manager would have wanted to speak to? If so, identify why the system missed them.
Perhaps the brief was too narrow. Perhaps relevant evidence was overlooked. Perhaps an unanswered question was treated as proof that a requirement was not met.
This kind of sampling helps diagnose problems; it does not replace individual safeguards or establish that a process is fair on its own.
Bias deserves specific attention too. Human recruitment is not automatically fair, and using AI does not guarantee fairness either. In its audits of recruitment AI providers, the ICO found examples of tools allowing filtering by protected characteristics, as well as inappropriate inferences about candidates.
Employers should seek evidence of relevant testing and keep checking how the tool performs in their own recruitment process.
Put hiring quality alongside time saved
Processing applications more quickly is useful. It should be assessed alongside the quality of the decisions being made.
Before expanding AI screening, agree how you will evaluate it. Useful questions include:
-
Do shortlisted candidates demonstrate the skills the role requires at assessment?
-
Are reviewers finding relevant applicants among those ranked poorly?
-
Can the team explain the evidence behind a recommendation?
-
Does the assessment distinguish a well-presented application from demonstrated ability?
-
Are candidates told how automation is used and given a clear route to raise concerns?
-
Does the process still work when the role, brief or tool changes?
The aim should be a hiring process that makes good use of people’s time while giving relevant talent proper consideration.
For specialist roles, that means understanding the work behind the CV: what someone delivered, the problems they solved and how that experience could apply to your team.
InGame Recruitment helps games studios find and assess specialist talent. If you are reviewing how you screen candidates, or struggling to identify the right people within your applications, get in touch to discuss your hiring needs.