1Behavioural and situational questions1,25 punts

Explain the difference between a behavioural and a situational interview question. Write one of each for the competency conflict management in a restaurant floor manager.

Solució

Model answer

A behavioural question asks about a real past experience, so it gives evidence of what the person did. A situational question presents a hypothetical scenario, so it shows how the person thinks, but it does not prove that the person has done it.

  • Behavioural: "Tell me about a time when you had to solve a serious complaint from a customer. What did you do and what was the result?"
  • Situational: "Two waiters argue in front of customers on a full Saturday night. What would you do?"

Follow-up questions should explore the context, the candidate's own action, the result and the learning, without suggesting the answer.

Idees clau: passat real vs escenari hipotètic · «Tell me about a time…» vs «What would you do if…» · seguiment sense suggerir.

2Indicators, structure and rubric1,25 punts
  1. What is an observable indicator? Write two for conflict management.
  2. What is a structured interview and why is it used?
  3. Why do interviewers need a rubric?
Solució

Model answer

  1. An observable indicator is a behaviour that you would see in a person who has the competency. For example: "listens to both sides before deciding" and "keeps a calm tone with an angry customer". An adjective like "friendly" is not an indicator.
  2. In a structured interview all candidates are asked about the same competencies, with the same questions and the same criteria. It makes candidates comparable and reduces the weight of personal impressions.
  3. A rubric fixes the criteria before the interview, for example from 1 (poor) to 5 (excellent), and describes the evidence for each level. The interviewer records the evidence, relates it to the competency, scores it and reviews it. The score summarises evidence, not sympathy.

Idees clau: conducta, no adjectiu · mateixes preguntes i criteris · registrar → relacionar → puntuar → revisar.

3Name the bias1,25 punts

Name the bias and say how to reduce it:

  1. The interviewer prefers a candidate who studied at the same university.
  2. A brilliant answer about Excel raises the candidate's leadership score.
  3. In the first minute the interviewer decides the candidate is weak and then only notes mistakes.
  4. A 55-year-old candidate is rejected because "he will not adapt to technology".
  5. The recruiter accepts the software ranking because "the machine is never wrong".
Solució

Model answer

  1. Affinity bias. Go back to the role criteria.
  2. Halo effect. Score each competency separately.
  3. Confirmation bias. Look for contrary evidence too.
  4. Stereotype. Check whether the characteristic has any relation with the role; assess the competency.
  5. Automation bias. A person must review the result and be able to change it.

A structured interview and a rubric reduce these biases, but they do not eliminate them.

Idees clau: quatre biaixos humans i un de màquina · cada biaix té el seu control.

4An unfair request1 punt

A manager asks you to remove a qualified candidate from the shortlist because of her age. What do you do?

Solució

Model answer

I would not accept the request, because selection must be based on the competencies of the role and not on personal characteristics. I would:

  1. Keep the role criteria and show the evidence of the candidate's competencies.
  2. Question the request in a professional way and ask which requirement of the role is behind it.
  3. Document what happened.
  4. Use the correct channels (HR, compliance) if the manager insists.
  5. Protect the candidate's information.

This is a question of ethics, fairness and inclusion, and also of quality: the company can lose the best candidate.

Idees clau: criteris del rol · qüestionar, documentar, canals, confidencialitat.

5An AI tool to screen CVs1,5 punts

A company receives 1,500 applications for 20 positions. It wants to use an AI tool, trained with the hires of the last ten years, that gives each candidate a score. Only candidates above 87 points will be interviewed. Which questions must the company answer before it uses the tool?

Solució

Model answer

  1. Purpose: which problem does the tool solve, and is it necessary to automate this decision?
  2. Representation: which data does it use and what does it leave out? A CV is an incomplete representation of a person.
  3. Criterion: what does "success" mean for the model, and is it related to the role?
  4. Impact: who benefits and who can be excluded? If the company hired mainly one type of person in the past, the tool learns that pattern. This is historical data bias. The same threshold for everybody is not fair if the data was unfair.
  5. Responsibility: who reviews, corrects and can stop the decision?

I would use the tool to order and summarise applications, not to decide. A person should review a sample of the rejected candidates. Human oversight means that a person can understand, question and change the decision; to sign the machine's output is not oversight. A score of 87 has an air of precision, but precision is not fairness.

Idees clau: propòsit, representació, criteri, impacte, responsabilitat · biaix de dades històriques · supervisió humana real.

6Judgment and reckoning1,25 punts

Moser, den Hond and Lindebaum distinguish judgment from reckoning. Explain the difference and what they recommend to managers.

Solució

Model answer

Judgment is human. It takes into account the social and historical context and the possible outcomes, and it relies on imagination, reflection, empathy and valuation. It has a moral dimension. Reckoning is what algorithmic systems do: they process data through calculation and rules, within the narrow objectives for which they were designed.

AI decides on "technical images", which are incomplete representations of the world, like a map. The answers have an air of precision, so people accept them as correct. The Dutch child-benefit scandal, in which a flawed system harmed about 40,000 families, shows the risk. The authors say AI will never be capable of judgement, only of reckoning, so "ethical AI" does not solve the problem.

They recommend two things. A call for inaction: stop automating everything that can be automated (technological solutionism). A call to action: learn what AI is and decide case by case whether, how and why to use it, with an attitude of vigilance and doubt.

Idees clau: judici = context i moral · càlcul = dades i regles · el mapa no és el territori · decidir cas per cas.

7The algorithmic divide1 punt

What is the algorithmic divide (Snyder and Mann)? What can leaders do?

Solució

Model answer

The algorithmic divide goes beyond the digital divide. It is the gap between those who have access to AI and AI literacy and those who do not, made worse by the cognitive biases of AI models trained on available data. Biases cannot be fully eliminated: to remove one often introduces another, and computing power amplifies them.

Leaders can act on three guideposts:

  1. AI readiness: educate leaders and employees, build trust and pilot quick wins.
  2. AI governance: responsible-AI policies and frameworks with legal, compliance, HR, procurement and IT.
  3. Sustainable investment: transparency, explainable AI, data protection and efficient use of energy.

For People Management this means AI training for all employees, not only for a few, so that the divide does not grow inside the company.

Idees clau: accés + alfabetització + biaixos · preparació, governança, inversió sostenible.

8AI in workforce planning1,5 punts

According to McKinsey, how can AI improve workforce planning in travel and logistics companies? Why are tools not enough?

Solució

Model answer

These companies face labour shortages, rising labour costs and slow productivity growth, and many still plan with spreadsheets. McKinsey proposes five AI-enabled actions:

  1. Digital demand reduction: remove work before it reaches frontline employees (self-service, chatbots).
  2. Resource and capacity planning: more accurate forecasts and dynamic scheduling.
  3. Hiring and people development: faster hiring and training plans tested with simulations.
  4. Day-of decision making: standards, playbooks and real-time data for local teams.
  5. Continuous improvement: clear standards, visible metrics, coaching.

Tools and analytics alone are not sufficient. They must be combined with process changes, performance management and change management, because culture is the true engine of the transformation. In the railroad case, employees did not use the routing tool because of technical problems and fear of job displacement. Usage grew only after training and a clear change story. The article also warns that AI assessments can introduce bias in hiring.

Idees clau: cinc accions · eines + processos + persones · gestió del canvi · risc de biaix.