AI Applications, Opportunities and Challenges
Efficiently recruiting top talent while delivering exceptional applicant experience is, for most talent acquisition teams, a challenge beyond the realms of possibility. This is especially true in volume recruitment, but even enterprise teams spend just shy of 100 hours and 4 weeks of manual human labour to hire each candidate. Countless hours are wasted screening candidates that are not suitable for progression beyond top-of-the-funnel screening processes, and consequently, recruiters typically lack the time to offer a reasonable experience to both the rejected applicants and the candidates that count.
But new advances in artificial intelligence (AI) are changing the game, and in particular, disruptive conversational AI applications are creating disruptive opportunities for innovative recruiters and hiring teams to set a better standard for candidate experience, and to strap on an “Ironman suit” to exponentially improve their productivity. Already 96% of recruiters are optimistic about the ability of AI to change the game.
In essence, “conversational AI” refers to the application of AI and Natural Language Processing to create automated interactions, enabling computers to communicate with humans with agility and to understand, interpret, and respond to conversations in a simulated human-like way, and with comprehension of the nuances of human interaction such as context, intent, and tone. Conversational AI relies on highly advanced machine learning algorithms to process and analyse natural language inputs to generate appropriate and relevant responses. Good conversational AI agents are virtually indistinguishable from humans.
What sets aside conversational AI from traditional chatbots is that chatbots rely on pre-defined decision trees that are highly limited (not to mention expensive and onerous to build). Unlike a chatbot, which can only respond with rigidity based on programmed inputs, a good conversational AI agent will be able to engage with a level of agility and personalisation to deliver exponentially superior user experience, as well as possessing the ability to constantly learn and improve the more conversations it delivers. The best conversational AI agents are also not limited to web chat pop-up windows but can be delivered across a miscellany of channels including mobile-first channels like WhatsApp, favoured by younger generations in particular.
Job descriptions are notoriously badly written, impacting success through the entirety of a recruitment funnel. By using Large Language Models (LLMs), talent professionals are better able to digest, suggest improvements, and even re-write job descriptions to improve their credibility and effectiveness, as well as ensuring that language is thoughtful of DE&I accommodations to eliminate inherent biases.
Engaging top talent from the beginning of their recruitment journey is essential, especially for impatient millennial and Gen Z candidates. Whether through outbound campaigns from an ATS, or inbound campaigns from a job ad, conversational AI agents can engage, build initial rapport and answer candidates’ questions promptly before they even apply for a role, reducing the administrative burden on the recruiter or HR team, and enhancing the speed of candidate engagement to deliver a superior application experience.
Screening candidates’ applications and long-listing qualified candidates using automation tools to streamline the top of the funnel is nothing new, but deploying screening tools that rely on LLMs can be a powerful differentiator.
Traditional application screening tools have relied on keyword sifting, an inflexible approach that depends on candidates refining CVs specifically to cater to a limited scope analysis, and favouring advantaged candidates who have been trained accordingly. Using LLMs offers a much more flexible and intelligent way of screening because they can intuitively understand if a candidate has experience without needing explicit keyword inclusion. This naturally makes the process fairer and avoids eliminating qualified candidates erroneously.
It’s important to note that heavy-handed, careless use of LLM-powered automation is not the answer. These tools must be flexible, guidable and customisable to reflect the recruiter’s human intuition. Developing an automation product that scales up human intuition (as opposed to machine intuition) is extremely difficult, but the very best AI recruitment tools will be aim to do just this.
A holistic solution would automatically deploy intelligent conversational automation across a longlist of candidates analysed by an ethical, human-guided AI, enabling the recruiter to begin building relationships with hundreds of candidates, as well as answering their questions and screening them on autopilot.
While there has been notable pushback from recruiters defending their manual outreach processes on LinkedIn and Email as critical to a process that relies on relationship-based human connections, the thoughtful use of AI in recruitment has significant advantages for both recruiters and candidates:
Cold-calling candidates to schedule interviews, or emails with “Are you free on Wednesday at ten?” is not acceptable in 2023. An adept conversational AI agent will connect with a recruiter’s calendar to schedule interviews with the shortlist of conversations that count.
At the human interview stage, data collection from the entire AI-enabled recruitment funnel should be collated to provide the interviewer with a synthesised candidate report and recommended questions, cases or tests to explore any potential gaps in the candidate’s profile.
Once contracts are signed, conversational AI will expedite the administrative burden placed on HR teams during the onboarding process. This includes collecting information from new hires and sending out necessary onboarding paperwork. HR teams can instead focus on more high-priority, high-value areas such as helping new employees acclimatise and hit the ground running.
Businesses are constantly changing, as are the needs of their employees. By using new technologies, businesses can extract data insights to keep senior leadership on-point with employee wellbeing and sentiment around specific practices of decisions, as well as to identify health-threats and recommend remedial steps before they become retention or PR crises.
For individual employees, adept conversational AI agents can improve employee experience by offering them the personalised information they need instantly while saving HR the time to invest in more meaningful people and cultural initiatives.
One of our early customers initially described an AI-enabled vision for the transformation of traditional recruitment and hiring practices as having the potential to lead to an “automation nightmare”. And she was correct. There is always a place for a human in the recruitment relationship, and hiring will always be a profoundly relationship-based experience that is as important for the hirer as the candidate. Meaningful applications of AI will not dilute the human element of talent acquisition, but enhance it by streamlining the repetitive manual processes to ensure that hirers are better prepared and have more time for high emotional quotient activities, building the relationships that matter and having the conversations that count. In our experience, candidates unanimously agree that until the interview stage, transactional access to the personalised, transparent, and real-time information that recruiters are typically too busy to provide trumps a human signature block in an email exchange.
Furthermore, while we see strong evidence of the success of AI during the engagement and screening phase, we also are convinced that human-in-the-loop capabilities are essential to intercept and takeover automation where a genuine human touch can make a difference. Recruiters should be able to have both birds-eye and detailed views of all the conversational campaigns running in real-time and be notified when it’s time to temporarily take over from the AI for a human-to-human interaction. Not only should human-in-the loop capabilities provide talent professionals with the means to take over, but they will be able to give Conversational AI agents the guardrails to constantly improve and think and behave more and more like their human counterparts.
The use of candidate application data is also an area of risk, especially when using open LLMs. Recruitment and hiring teams who want to leverage analytics to continuously improve hiring processes and decision-making must ensure that they fulfil new privacy, security, and regulatory compliance obligations, but also constantly examine and reassess the way they seek to optimise hiring based on data to ensure that they don’t become inherently biased. This is especially important for protecting the very DE&I considerations that can be strengthened by a responsible application of LLMs.
Finally, to the biggest elephant in the room: Headcount displacement. Inevitably, at the advent of a revolution of automation capabilities recruiters and human resources professionals are concerned about businesses leveraging new technologies to reduce headcount costs. And their worries are justified. Businesses will migrate to an era of leaner data-driven hiring teams that are equipped with the latest AI co-pilots to give them 10x capabilities, and this will inevitably impact traditional recruiters who still prefer manual processes, especially in volume hiring practices (notably black-book executive search is something of an exception here). Talent professionals who embrace these disruptive new technologies will find themselves irreplaceable and will enjoy a strong level of job security that survives cyclical hiring churn based on their indispensable understanding and engineering of the AI and data architecture of a company’s talent strategy. Those who do not seek to up-skill and adapt may find themselves increasingly vulnerable to displacement in the not-so-distant future.
Popp AI is building the end-to-end AI toolset for innovators in talent acquisition. With Popp, you can deliver the features outlined in the article above straight out-of-the-box, in seconds.
