La IA permite personalizar el trato con los empleados, pero no sustituye a la persona en RR. HH

AI enables personalization of employee engagement, but does not replace the HR person

Ana Valera, People Analytics expert and member of the Advisory Council. IA+Igual Marc Altimiras, Vice President Southern Europe of CornerstoneCornerstonehave answered the question "What to ask a supplier to hire reliable AI?" from the Human Resources department in a webinar at Campus IA+Igual.

Madrid, February 27th, 2024. Ana Valera, People Analytics expert and IA+Igual board member, and Marc Altimiras, vice president of Southern Europe at Cornerstone, have answered the question "What to ask a provider to hire reliable AI?" from the HR department in a Campus IA+Igual webinar held this morning. Both professionals concluded that AI will never replace a quality HR professional, but it does allow for individualized treatment of the entire workforce.

During the session, the questions that HR professionals ask themselves when hiring artificial intelligence (AI) tools were clarified. Many managers often wonder whether the quality of an algorithm depends on the amount of data. Marc Altimiras explained that, without neglecting volume, the key lies in the large language models (LLMs) used to develop an algorithm based on its use case, and he recommends asking the supplier for the possibility of changing the model if it is not suitable, always respecting intellectual property.

«Es importante saber si el fabricante tiene un compromiso sobre el desarrollo del tipo de IA que genera. El lenguaje que se tiene que hablar en la empresa es el de las habilidades, por encima de cualquier otro criterio adicional”, says the vice president for southern Europe of the talent management and training services provider.

Altimiras presented some of Cornerstone's use cases for more than five years: identification of employees' skills through their CVs, work experience in the company or information from the HR department; creation of chatbots or virtual assistants through Generative AI that answer workers' frequently asked questions more quickly; writing job descriptions that ease HR's workload; creation of questionnaires to check whether training has been assimilated at the end of a course; etc.

Audit of biases

Ana Valera, for her part, gave a clear answer to the question of whether biases can be combated with the use of artificial intelligence. . "A priori, no, but they can be mitigated. Both AI and people have biases. What is essential is to have criteria and specify how to regulate their use," said Valera, who said it is essential to have criteria and specify how to regulate their use said Valera, who illustrated this idea through examples: it is possible to eliminate age to try to avoid bias, but this is deduced from the professional trajectory; if one opts for blind curricula to avoid gender bias, many data remain that highlight whether the author is a man or a woman.

Therefore, the key for the AI+Equal Advisory Board member is to constantly audit algorithms so that they work in a fair and equitable manner and to act ethically to decide which uses are not appropriate because AI has moved faster than regulation. Some company may be tempted to reproduce biases such as "I am looking for 20-30 year old female", but should introduce other variables to avoid a reputational crisis.

In the era of critical thinking," Valera stressed, "the key is to continue training to discern the results offered by algorithms. We now have a 'Ferrari' to make better decisions, but the person has the last word."

Complementing the AI use cases of an HR provider like Cornerstone, Valera - who has worked in several multinationals - mentioned some examples in the talent area:

  • Select candidates and predict the likelihood of success.
  • Create chatbots or virtual assistants for candidates or employees.
  •  Virtual coaches that help the development of different skills.
  • Personalize compensation by analyzing all benefits, the value proposition, optimizing the money the company spends on flexible compensation...
  • Detect people with higher potential and better performance more objectively than one person.

Biases can be mitigated through continuous training of the algorithm and debugging when necessary "Time spent training AI is like time spent training an intern: an investment," Valera concludes.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll al inicio