How Petrova AI Scores Candidates
Petrova AI creates a balanced score for each candidate by analyzing two main things: speech and video. Together, these scores make up the Aggregate Score, which gives hiring teams a clearer view of each candidate’s abilities.
1. Speech Analysis
Speech analysis focuses on the content and quality of what the candidate says. Here’s what Petrova AI looks for:
Clarity (High Weight): The AI checks if the candidate’s words are clear and easy to understand, including an assessment of pauses and hesitations.
Relevance (High Weight): It ensures the candidate’s answer is on-topic and relates to the question.
Coherence (Moderate Weight): It looks for how well the candidate organizes their thoughts, showing they can communicate effectively.
These elements together create a solid view of a candidate’s verbal communication, which is crucial in many roles.
2. Video Analysis
Video analysis focuses on how the candidate presents themselves, capturing non-verbal cues like facial expressions and body language.
Positive Expressions (Very High Weight): Positive signals, like smiles and engaging expressions, suggest confidence and interest, which are highly valued.
Calmness (Moderate Weight): Staying composed under pressure is a good indicator of a candidate's ability to handle challenging situations.
Confusion (Moderate Weight): A small amount of confusion is natural, but higher levels may indicate difficulty in processing or responding to questions effectively.
Nervousness (Low Weight): Signs of nervousness or discomfort, while natural, can impact response quality and are taken into account.
Frustration (Low Weight): Visible signs of frustration are lightly considered.
The combination of these non-verbal cues creates the Video AI Score, which, when combined with the Speech Score, makes the full Aggregate Score, giving a balanced view of the candidate’s communication and presence.
Streamlining the Hiring Process: AI Pre-Select and AI Reject
To make hiring quicker and more efficient, Petrova AI automatically categorizes candidates:
AI Pre-Select: Candidates scoring high are flagged as strong prospects, making them worth further review.
AI Reject: Candidates scording low may not be the best fit for the role, allowing hiring teams to focus on others who are more suitable.
This automated sorting saves time and directs attention to the candidates most likely to succeed.
Common Questions About Petrova AI
How accurate is your AI model?
Our AI model's accuracy is rigorously tested at the feature level, where we ensure that each observable behavior is interpreted correctly. On average, Petrova AI achieves a test accuracy of 75-80% across datasets, and we mandate a minimum accuracy of 70% on testing before any model is put into production. This standard allows our AI to deliver reliable assessments of both verbal and non-verbal indicators, providing hiring teams with a dependable evaluation tool.
How do you address and prevent bias in your AI, especially given concerns about AI fairness in hiring?
We place a strong emphasis on transparency and clarity within our AI models. Rather than relying on opaque algorithms, Petrova AI uses an Explainable AI (XAI) framework, where each decision is based on clearly defined, measurable factors. By focusing on specific observable traits and applying explainable machine learning methods, we aim to minimize bias in assessments. Our system is designed to ignore factors like gender, age, and ethnicity to create an equitable and inclusive evaluation process.
Does Petrova AI replace the traditional recruitment process, which often relies on human intuition and judgment?
Our AI is designed to enhance—not replace—traditional HR practices. Petrova AI provides objective insights that support recruiters’ expertise, enabling them to make more informed decisions. By automatically highlighting high-potential candidates, Petrova AI saves HR teams time, allowing them to focus their efforts on candidates who show the most promise, while still applying their own professional judgment.
Is your AI trained on a global dataset?
Our AI is trained on a globally representative dataset that includes diverse demographics and regions. We carefully construct and review our datasets to avoid biases from historical data, ensuring they reflect a wide range of candidates. The Petrova AI model is designed to be neutral to variables like gender, ethnicity, and age, providing objective assessments that are fair to candidates from all backgrounds.
Wrapping It Up
The Aggregate Score is valuable because it helps hiring teams make decisions that are faster, fairer, and more effective. Here’s why it’s important:
Efficiency: Petrova AI quickly highlights top candidates, saving time and resources.
Objectivity: By using measurable, explainable methods, Petrova AI avoids personal bias and provides a fair view of each candidate.
Complete Picture: Combining speech and video analysis gives a full view of each candidate’s skills and presence, helping recruiters make well-rounded decisions
Share this post