Artificial intelligence is already built into applicant tracking systems, LinkedIn and the filters that decide which candidates make it to an interview. In this episode of Jugando en Serio (in Spanish), Antonio Robles García, an expert in people strategy and AI applied to HR, university lecturer and master's program director, explains how to bring AI into recruitment without dehumanizing the process, and why games reveal what an interview hides.
AI as HR's copilot
For Robles, people and technology have always gone hand in hand; what matters is how we use it to our advantage. That's why he thinks Microsoft got the name of its assistant, Copilot, exactly right: AI should take administrative and repetitive work off our plate so we can spend time on what is creative, human and empathetic.
In his doctoral thesis he studies how we are adapting to this technological revolution, and suggests that companies will have to manage a new kind of diversity: coexisting with another type of intelligence. People are afraid, but they know there's no going back.
Play reveals what an interview hides
In an interview, candidates say what they know you want to hear; in a game, they don't know what you expect. According to Robles, 80% of learning happens between the ages of 0 and 18, a stage when we're always playing. A simple dilemma, such as deciding whether to give a potion that saves a patient, already shows whether someone leans towards results or towards quality and order.
His clearest story comes from a mass recruitment drive using live role-play: groups of 20 people in a dark room full of zombies, trying to save humanity. There was only one rule: don't hit the zombies on the head. A candidate who had shone in every earlier test broke it three times, and in the feedback session admitted she didn't know how to handle stressful situations.
I would never have seen that behavior in an interview; I can only see it there. You see people freeze, you see things you'll never see in any other circumstances.
Why games meet more resistance in training than in hiring
Robles sums up his approach as «hiring: play»: what candidates know is checked over the phone; how they behave, through a game. Nobody has ever questioned the method in recruitment, but in training they have: many companies would rather have «a course with proper content». The more tangible the skill, the easier it is to accept; with soft skills such as leadership it's harder, although resistance is falling as the technology gets cheaper.
At university he applies the same logic: if anyone can find all the information on their phone, what students need to learn is critical thinking to judge whether that content is valid, and that can only be learned by doing.
AI filters, but you decide
Robles wouldn't let an algorithm choose his flatmate, and he won't let one choose his candidates either. What he asks AI for is a list of the five people he should sit down with. If they have the knowledge the job requires, the decisive question follows.
The advantage is energy: if AI does the first screening, you arrive at the interview fresh and make fewer mistakes. It can also draft the report from the transcript or, with predictive AI, look for the patterns of your best people in new candidates. But a person sits down with the ten who fit best.
If you'd go for a coffee with them afterwards, that's your candidate; if not, no, because if they join your organization you'll have to have a lot of coffees with them. Whether you want to have that coffee or not is not something artificial intelligence can decide.
Bias: AI is fast and obedient, not neutral
AI is mathematics that has learned from human data, and that data carries bias: depending on its sources, it may select more men or more women. There are tools to put guardrails in place, but it's an engineering job that HR has to take on. What the machine brings is speed and obedience, like the intern who used to score paper CVs.
Robles won't predict what will happen in six months, but he's clear that anyone who doesn't get on board will be left behind. Applicant tracking systems and LinkedIn already include AI: the key is to manage it well and never forget that the final decision is yours.
