AI Interviewers Are Here: Three Strategies to Ace Them

[baoshi.rao]

You never know if your interviewer is human or... artificial intelligence. With AI interviewers becoming prevalent, here are some key strategies to tackle them effectively.

For many office workers, their peak acting skills are often reserved either for shifting blame or handling interviewers.

However, relying on "persona-building" to breeze through interviews isn’t as easy as it used to be.

Since the rise of artificial intelligence, many companies have handed over the interview baton to AI, leaving seasoned job seekers feeling the pressure.

Starting last year, many top-tier companies have incorporated AI interviews or digital interviews into their recruitment processes. Several hiring platforms have even made AI interview systems a key growth area, positioning themselves as tech pioneers.

For job applicants, facing an AI interviewer—tireless, emotionless, and highly perceptive—can be daunting.

Especially for fresh graduates, who previously relied on advice from seniors, the novelty of "AI interviewers" means there’s little practical guidance available.

Today, let’s explore the boundaries of AI interviews and how to "hack" them.

If you encounter a dream job that requires an AI interview, don’t panic. The key is to approach it strategically: underestimate it in principle but pay close attention to tactics.

This is because many companies use the "AI" label more for marketing than for actual functionality.

The fast-moving consumer goods (FMCG) sector is a prime example. Companies like Coca-Cola, P&G, and Unilever were among the first to adopt "AI interviews" in campus recruitment.

On one hand, these roles often have no strict professional requirements but come with the prestige of being multinational Fortune 500 companies, leading to intense resume screening pressure.

On the other hand, campus recruitment is a golden opportunity to attract young talent. Many companies promote slogans like "hunting future leaders," making the competition fierce. Creative hiring methods like open questions, AI interviews, and gamified assessments help amplify their brand appeal.

This means their AI interview systems must meet two core demands: speed (to gain a first-mover advantage in marketing) and robustness (to avoid bias and negative applicant feedback).

As a result, most AI interview solutions used by these brands are third-party algorithms built on mature AI technologies.

This implies that AI interviews primarily serve as a preliminary screening tool, not a decisive factor in hiring. They help reduce human biases (e.g., appearance, accent, or alma mater) and improve job-candidate matching.

Meanwhile, the limitations of deep learning models, NLP, facial recognition, and emotion algorithms allow applicants to strategically target the AI’s evaluation criteria.

Let’s break down these specialized interviewers.

Some AI interviewers combine NLP and voice recognition to assess candidates through Q&A sessions.

For example, Japanese recruitment firm En Japan had graduates answer 126 questions over an hour-long interview via a smartphone. The grueling process left many candidates speechless.

These systems analyze speech patterns (tone, pace) to gauge reaction speed and emotional state, while NLP algorithms evaluate keyword relevance and semantic alignment with job requirements (e.g., P&G’s "Eight Core Questions").

Since these AI interviewers rely on predefined parameters, they’re relatively impartial. Here’s how to prepare:

1. Research Standard Questions: Most AI interviews use preset questions. En Japan’s system, for instance, was trained on 15 years of interview data covering basics, skills, and personality traits.

2. Adapt to the Format: Unlike human interviewers who might chat casually or throw curveballs, AI systems follow a rigid script. Study the company’s culture (e.g., fast-paced, creative, global) and practice delivering clear, structured answers.

3. Keyword Optimization: AI matches responses using speech-to-text, keyword extraction, and semantic analysis. Highlight hard skills (e.g., leadership, national projects, teamwork) to improve your score.

With these tactics, AI interviews might feel easier than in-person ones.

For tech-savvy or elite firms, you might face video AI interviews, which analyze facial expressions and micro-movements to assess honesty and personality fit.

While this sounds intimidating (e.g., even eye-rolling could count against you), current emotion-recognition algorithms are far from foolproof. Even giants like Microsoft and Google admit their limitations in hiring contexts.

Video AI (e.g., HireVue, Sonru) evaluates 15,000 traits—speech, vocabulary, eye contact, volume—against a database of "successful" candidates to rank applicants. Hilton has used it for 43,000+ roles, with HireVue conducting 1M interviews quarterly.

As one tech lead noted, "The extreme complexity of human language and expression requires extreme caution to avoid algorithmic bias." If 90% of candidates fail a question, the system adjusts its criteria.

So, while AI interviews are here to stay, they’re not unbeatable. Understand their logic, prepare strategically, and you’ll stand out—even to a robot.

For instance, only 10%-30% of the score is determined by facial expressions, while the majority depends on the candidate's verbal performance. In speech, using keywords that align with the target company's preferences can help. Whether a candidate prefers passive or active verbs, frequently uses "I" or "we," or employs technical jargon can influence the system's assessment of fit.

Another example is voice tone. If someone speaks too slowly, they might not be suitable for roles like telephone consulting, while speaking too fast may make it hard for users to understand. Finding the right balance for the desired role through empathy might make "feel" more reliable than data.

Does this make you feel the bleakness of corporate life? Don't rush to despair. If you "unfortunately" encounter an AI interview system that scans your social media, you might as well buy a lottery ticket to console yourself after being laid bare by the algorithm.

At this stage, AI often analyzes complex data about a candidate's daily behavior to deduce their fit for the role. Not long ago, the California-based startup Predictim used NLP and computer vision to scan babysitter applicants' Facebook, Instagram, and Twitter histories to predict whether they might bully, harass, or mistreat children.

Of course, such interviewers were quickly boycotted by the industry. Facebook accused the company of violating policies against using such data for hiring decisions, restricting its access to user data. Twitter also revoked Predictim's API access, citing a ban on using Twitter data for surveillance. Similar algorithmic risks occurred on LinkedIn when third-party site HiQ collected data to predict employee turnover.

Tech giants have distanced themselves from such AI interview systems because machine learning cannot reliably interpret nuances like sarcasm or jokes, making it unstable in interviews. Additionally, these algorithms are often unmonitored (black-box systems), potentially disqualifying diligent candidates without explanation.

Moreover, leaving hiring decisions entirely to AI violates tech ethics. If a company trains its AI on narrow, biased datasets, the fairness of AI interviews disappears, potentially worsening discrimination based on age, race, etc.

As Anna Cox, a human-computer interaction professor at UCLA, notes, "Any dataset will have biases, excluding people who are actually great at the job." Currently, AI's role in analyzing complex interview factors remains a contentious future.

However, technology is advancing rapidly. IBM announced using Watson to analyze internal training data to assess promotion potential. Extending this structured data approach to hiring may soon become widespread.

In the past, AI replaced highly mechanical, structured tasks like documentation, translation, and recognition. Now, even emotionally nuanced fields like interviews aren't spared.

Fortunately, as the magic of technology fades, humans armed with knowledge will find the best way to "co-work" with AI. Through continuous iteration and optimization, we'll overcome the growing pains of human-AI collaboration.