The Definitive Guide to AI Screening: What Works and What Doesn't
Seeing high application volumes and keyword stuffing by candidates?
AI Screening is becoming a popular option for many companies. But what works and what aspects one should keep in mind?
Read this guide to get the answers.
Why AI Screening is becoming important?
Increasing Application Volume
As job markets become more competitive, recruiters face an overwhelming influx of applications. This surge makes it challenging to manually review each resume effectively.
Keyword-Stuffed Resumes
Many applicants optimize their resumes with keywords, making it difficult for recruiters to discern truly qualified candidates from those who simply know how to game the system.
Need to identify good candidates
AI screening helps quickly identify the most promising candidates, allowing recruiting teams to focus their efforts on high-potential applicants early.
Expectations from AI Screening
Short and fast
Since candidates are not yet invested in the process, AI screening needs to be easy and fast (<15 minutes).
Convenience
To ensure high participation, AI screening should work on mobile devices needing low bandwidth and no software installation
Accuracy
Should effectively vet the candidates reducing false positives and false negatives.
Remember: AI Screening is NOT an assessment tool
AI screening is a recruiter tool and not a part of the assessment process
1
It's a Recruiting-Ops Tool
AI screening should be viewed as a powerful aid to recruiters, enhancing their ability to manage high volumes of applications efficiently.
2
Not a Formal Assessment
It's important to distinguish AI screening from formal assessment processes. The screening is a preliminary step to identify potential candidates, not a comprehensive evaluation.
3
Complements Human Judgment
AI screening complements rather than replaces human decision-making. It is NOT an auto rejection or auto selection tool.
Designing an Effective AI Screening Process
1
Send to filtered candidates only
Since it requires effort, only send it to those who match the requirement.
2
Practical questions work best
Focus on experience-based questions rather than theoretical or programming queries.
Good: "How did you go about designing UX for your Gen-AI product" ✅
Bad: "What are different caching strategies?" ❌
3
Ensure candidates feel respected
Candidates need to feel valued and should feel that talking to AI will not waste their time. Ensure the AI speaks politely and respects their time.
4
Improve continuously
Regularly analyze the effectiveness of your AI screening process. Use feedback from recruiters and successful hires to refine and improve the system over time.
Types of AI Screenings
Phone-Based
Phone-based AI screenings utilize automated voice calls to ask candidates basic questions about their profile and experience. These screenings are typically short and focus on verifying information from the candidate's application.
Key Advantages
- Familiar to most candidates and require no change in behavior
- Excellent to collect basic data like notice period, salaries etc.
- A phone call brings trust and seriousness as the hiring process gets started
Current Limitations
- Not very effective to dig deeper into the skills and project experience
Types of AI Screenings
Chat-Based
Chat-based AI screenings involve text-based conversations between candidates and AI chatbots. Conducted through various platforms, including messaging apps or web interfaces.
Key Advantages
• Offers flexibility for candidates to complete at their own pace
• Enables structured responses and organized information gathering
• Can easily incorporate multiple-choice questions
Current Limitations
• Difficult for candidates to type detailed answers, especially on mobile devices
• May limit the depth of information gathered
• Not ideal for roles requiring extensive written communication skills
Types of AI Screenings
Audio-Based
Audio-based AI screenings allow candidates to respond to questions verbally. The AI system records and analyzes their responses, providing a more natural and detailed interaction
Key Advantages
• More detailed responses compared to text-based methods, since candidates don't have to type a lot.
Current Limitations
• Requires more effort from the candidates as compared to taking phone calls.
Types of AI Screenings
Video-Based
Video-based AI screenings involve candidates answering questions on camera in real-time. This format mimics traditional video interviews but with an AI interviewer instead of a human.
Advantages
Video screenings can provide additional visual cues and allow for assessment of non-verbal communication. They can create a more engaging experience for some candidates.
Limitations
While visually appealing, video screenings may not add significant value to the core assessment. Candidates know that they are talking to AI and they may find video to be an unnecessary distraction.
Considerations
Video screenings require more bandwidth and a quiet, presentable environment, which may not be accessible to all candidates.
Minimising cheating in AI Screening
Unfortunately, technology can also be used by some candidates to use unfair practices.
Since AI Screening is not a selection or rejection tool, eliminating cheating at the cost of losing speed and convenience is not often desirable.
Here is how to make it harder to cheat in AI Screening
1
Design questions carefully
Craft questions that are experiential and based on the candidate's resume.
Good: "How did you go about designing UX for your Gen-AI product" ✅
Bad: "What are different caching strategies?" ❌
2
Scramble questions into different channels
For e.g. showing a code snippet on screen while asking the question in voice will make it difficult to cheat.
Do candidates like AI screening?
After completing 30,000+ screenings in last 6 month on Cutshort, the answer is:
Most do.
The candidates want to feel respected and expect a response if they invest their time into AI Screening.
Junior Candidates (0-4 yrs experience)
Approximately 60% of junior candidates attempt AI screenings.
Experienced Candidates (5+ yrs)
25% of experienced candidates voluntarily participate in AI screenings. This can be increased by a push from the recruiting team.
AI Governance and Ethics in Screening
1
Use it to complement a recruiter, not to take decisions
Use AI screening tools to assist recruiters in prioritizing candidates, rather than making final decisions. This ensures human judgment remains central to the hiring process.
2
Manual Review of Rejections
Implement a policy that requires human review before any candidate is rejected based on AI screening results. This safeguards against potential algorithmic biases or errors.
3
Bias Mitigation
Choose AI screening tools that actively work to remove biases. For example, some tools like Cutshort's AI Screener redact voice, nationality, gender, and other personal information before calling the AI models.
4
Transparency and Consent
Clearly communicate to candidates when AI is being used in the screening process and obtain their informed consent. Provide information on how the AI works and how their data will be used.
Metrics to set for AI Screening Implementation
Here are some suggested metrics and improvements goals you could start with
Getting Started with AI Screening
1
Assess Your Needs
Evaluate your current recruitment process and identify areas where AI screening could have the most impact.
2
Choose the Right Tool
Research and select an AI screening tool that aligns with your specific requirements and industry needs.
3
Pilot Implementation
Start with a small-scale pilot program to test the effectiveness of AI screening in your organization.
4
Gather Feedback
Collect feedback from recruiters, hiring managers, and candidates to refine your AI screening process.
5
Scale and Optimize
Based on pilot results and feedback, expand AI screening across more roles and continually optimize for best results.
Try AI Screener by
- Designed to balance speed and convenience with accuracy
- Supports descriptive and code snippet questions
- 30,000+ screenings completed in production with 4.2+ user rating