Imagine that you have a job interview: you are confident in your abilities but you are nervous, of course, and you also doubt if you will fit in the company or if you will like the tasks you are given, or if you will be satisfied with the salary. Now imagine that you have already entered, and that the human resources department is asking you how you are at the company, if you are happy, if you feel stress… So far, these imaginations are not unusual: for all jobs there are processes of selection, and in many companies work environment surveys are carried out. Now imagine that this interview and these surveys, as well as the daily meetings, are done on video, and that behind these videos an artificial intelligence (AI) is in charge of collecting the data it receives from what is said and from facial expressions. , passes them through a model, and transmits it to the human resources team with evaluations such as these: the candidate was happy when the tasks were explained to him, he was nervous when he found out the salary; the worker has gone through a moment of stress in this meeting, or in this project he has become very emotionally involved.
In recent times, AI, a technology that has been in development for years, has burst into the collective imagination with force thanks to programs such as ChatGPT, which create texts or images based on a few indications. But the practical application of AI goes beyond these experiments, and experts predict that it will enter many aspects of our lives. There are detractors and defenders of this trend, but what seems certain is that there is no turning back, and this is the belief of many of those who are present at the Mobile World Congress (MWC), who show different ways of putting into practice artificial intelligence. Large technology companies do it, but also entrepreneurs: AI is one of the main protagonists in the 4YFN entrepreneurship hall, which is held inside the MWC with 556 startups exhibitors.
One of the areas where AI is beginning to be used is the workplace. It is present in various projects, some of large technology ones, such as Google’s Interview Warmup —an artificial intelligence to train before an interview—, or in initiatives such as LKS Next (of the Mondragón Corporation) and Kevo Technologies, which have created the IA Select application, an artificial intelligence that performs the psychotechnical tests in the selection processes. In the 4YFN there are two startups from Barcelona who have set up a demonstration with their projects. Aniol Hervás Royo is CEO and co-founder of the startup Emotional, which in one year of existence has raised one million euros and has Redarbor as its reference investor (investor of employment platforms such as Infojobs or Computrabajo). Its product is an artificial intelligence that integrates into meeting video calls, and recognizes and translates the emotions shown by employees into data. “Our program does the same thing as anonymous work climate evaluations, but it is not a substitute for them, it simply gives the human resources department general data, a more objective framework, of how the teams are doing: if the level of stress is increasing, if they are happier… and based on this we also make recommendations to address these emotional situations”, explains Hervás. This same technology, with which they already have clients of the stature of Sanitas, although above all they use it startupsThey also use it to do the personality tests that are done in the selection processes. Are you not concerned that this data could be used to discriminate against candidates? “In personality there are no negative aspects, everything can have its positive side, and the objective is to use it so that one knows himself more, what his strengths are,” explains Hervás, who has a team of 15 workers.
A few meters away, in the same 4YFN, is Santi Molins, CEO and co-founder of Talentfy, a startup three years old. Its main product comes from detecting a need: there are not enough technological workers (programmers, developers…) for a booming market, and companies need to find the best. With this objective, they have created a platform that connects technology companies with recruiters classified by an AI (the machine gives the company a selection of recruiters according to its needs, which will be the ones that will present the candidates); In parallel, the AI, which has read all the application data, lists the ideal candidate for each position, and also automatically feeds the candidates’ CVs (for example, adding the technical tests that have passed in different processes of selection, but also showing the reviews of the companies: why they have selected it or why they have discarded it Finally, Talentfy has a technology very similar to that of Emotional, which they are also showing at 4YFN, which is the recognition of emotions in video interviews. “All these AI projects help the company find the best candidate. For example, if you want one who has not had much turnover in companies before, or if you want to have a prior analysis of what behaviors you will have, to make sure that it won’t cause problems. But normally this doesn’t matter to them, what they want is that even if you’re for a short time (these highly qualified professionals have an average rotation of 18 months) you’re very profitable”, explains Molins. With his company, with 20 workers, they billed 1.2 million in 2022, and they have raised an investment of 1.2 million in three years.
Do algorithms discriminate?
This use of artificial intelligence and algorithms in the workplace raises an obvious question: can it be guaranteed that there is no discrimination in these automated processes, and that this data cannot then be used to discriminate? Dani Cruz, head of Analysis and Digital Transformation of the CC OO union in Catalonia, recalls that a European directive that is in the process of being drawn up will probably consider the use of artificial intelligence in selection processes to be a high risk, with which these technologies are already They cannot be used unless it is very justified. “We are not against AI if it can make work easier, but what we ask is that the data and processes used are transparent, because if everything is done anonymously and as objective data, it may deepen discrimination” , points out.
Indeed, researchers from think tank American National Bureau of Economic Research addressed this issue in a 2019 study in which they stressed that “it would be irresponsible, even dangerous, to confuse “data-based” with “non-discriminatory”, “impartial” or “objective”. Although he points out that, although the algorithms can use data that gives a discriminatory result (for example, if they use a database of selection processes where there has already been discrimination), their use can be positive: until now, discrimination was made only by humans , and when asked, they could lie, whereas with AI you can find out when and why this bias has taken place.
Perhaps the answer lies in the approach made by another startup in the 4YFN, called Expai. This is dedicated to explaining, in a very simple platform, how an AI model works when you give it some orders and it makes a decision. For example: a bank asks an AI to decide whether to give a loan to a client, and to make this decision it has all the historical data on the granting of loans. The model reaches a conclusion and decides, but if the process is not explained, it is not known how it has decided, and therefore it can be discriminatory. The startup, who is one year old and is looking for investors, explains and translates into graphics how the decision was made: if age, gender, etc. were taken into account. Another simpler example: taking the data of the survivors of the Titanic disaster, you ask the AI to decide who it would save in a shipwreck. The result is totally biased, since based on these data, it would only save children and women, and first class passengers. One of the workers of this startup, Arnau Soler, has a clear answer to the debate: “In the future, AI will be explainable or it will not be. Because we cannot trust 100% of what the machine tells us, we have to know how it makes decisions”.
Five Days agenda
The most important economic appointments of the day, with the keys and the context to understand their scope.
RECEIVE IT IN YOUR MAIL
Subscribe to continue reading
Read without limits