PTWhen the algorithm joins HR, is it a tool or a colleague?
Dashboards, systems and artificial intelligence reached people management with the promise of data-driven decisions. The thesis interviewed 10 HR directors and business partners in the Brazilian pharmaceutical industry and concludes that this People Analytics acts as an actor: it changes goals, tasks and the professionals themselves. Out of this comes a new crisis, the hybridization crisis. Explained in 12 screens.
People (blue) and algorithms (red) start deciding together (purple, on the table in the middle). Drag to rotate.
Tool or actor? It depends on the lens you look through
People Analytics is the use of data, systems and algorithms to decide about people: whom to hire, who may leave, how the climate is. The thesis looks at it through two lenses at once. Switch lenses and pick an aspect.
Table 12 and section 1.6. Lens 1: Social Construction of Reality Theory (Berger and Luckmann, 1966). Lens 2: Actor-Network Theory (Latour and Callon), which looks at people and things in the same way. The network is our diagram.
Ten HR leaders from ten pharmaceutical companies told how data entered their routine
Between November 2024 and January 2025, the author interviewed 6 HR directors and 4 HR business partners (HRBPs), almost always in their own offices. All of them report to the company's CEO and none knew the others. See who they are.
Tables 10 and 11 and sections 2.3 and 2.4. The counts are the thesis's. Company numbers follow Table 11; the thesis does not say whether company 3 is interviewee 03's, so we do not link the two rows. To protect anonymity, we do not show gender per person (7 women and 3 men in total).
Do what the author did: read a quote and say which theme it belongs to
The interviews were read and reread in the NVivo software. Each excerpt got a code, the codes were grouped and, in the end, three major themes remained. Pick the theme of each quote.
Sections 2.5.1 to 2.5.5 and Figure 14. The 11 sub-themes are our count in Figure 14 (5 + 2 + 4). Quotes are as the thesis gives them in English; the interviews were held in Portuguese and the original transcripts were not published. The thesis says the software helped to organize, but the interpretation was the researcher's.
Before data can create value, it has to convince
No company adopted People Analytics all at once. The path starts in doubt, moves through alliances and systems and, when it works, reorganizes HR. Then new doubts appear and the cycle starts again. Tap each stage.
Figure 15 and section 3.1. The five stages follow Figure 15, inspired by the "moments of translation" of Callon (1986) and Latour. Quotes as given in the thesis.
Many systems, little talk between them
People Analytics only exists once it becomes a thing: software, dashboard, algorithm. The interviewees named 15 tools, and almost all complained about systems that do not talk to each other. Each bar shows how many named the tool; tap one.
"We use 10, 15, 20% of what it can deliver."E01, on the HR system
Sections 3.1.2 and 3.1.3. Who named each tool is our count from the list in section 3.1.3.2 (plus one ChatGPT use mentioned right after). "Visier" also appears in the thesis as "Vizier". The brands are the ones the interviewees named; the thesis does not assess any of them.
Facts instead of impressions: data changed a conversation with the CEO
In one of the companies, the CEO believed a competitor was taking the firm's talent. HR looked at the 16 departures in its analytics system. Compare the impression with the data.
Sections 3.1.2, 3.2.2 and 4.2. The drawing of the 16 departures is ours: the thesis says only that 1 of the 16 people was classified as talent and that the 16 went to different companies. The time bars follow what the interviewees said (for "hours" we used half a day as a reference, our reading). For the thesis, this objectivity does not come ready-made in the data: it has to be validated and gain value in people's eyes.
The algorithm does not just obey: it deviates, composes, takes over and transforms
Drawing on Bruno Latour, the author shows four ways in which People Analytics alters what HR set out to do. Pick one and watch the goal go into the algorithm and come out.
Sections 3.2.3 and 4.2.3, based on Latour (2001), Pandora's Hope. The examples come from the interviewees; the drawing of the path is our diagram. Quotes as given in the thesis.
Systems to spare, people short of time
In daily life, the coexistence of people and algorithms runs into four tensions. The paradox the thesis highlights: the professional is overloaded while the system is underused. Flip the cards.
"[...] so today I take care of the operations part, I take care of the compensation and benefits part, I am a business partner for the functional areas, I take care of IHS and fleets [...] it's a little monster."E02
Sections 3.3.1 to 3.3.3 and Appendix B. The striped band shows the 10 to 20% range E01 mentioned, not a measurement. The five blocks of E02's day are the five areas mentioned, with no real size. Quotes as given in the thesis.
A sixth crisis on the companies' staircase: hybridization
Researcher Larry Greiner described organizational growth as a staircase of phases and crises: each period of growth sows the seeds of the next crisis. The thesis proposes a sixth step, algorithmization, and with it a new crisis. Tap the steps.
Figures 20 and 23 and sections 3.3 and 4.3. The thesis uses Greiner's model in its 1998 version. The author presents the sixth phase as a theoretical lens, not as a proven universal stage. Step heights are a diagram.
The more human AI seems, the easier it is to forget what it is
Before seeing the result: what would you call the artificial intelligence you use at work?
Section 4.3.1 and Figure 23. The nicknames are in section 4.3.1, without the code of who said them. The slider is our illustration of the thesis's idea, the "anthropomorphic paradox": the bars have no numbers, they only show the direction the author describes.
Use the algorithms without handing them your judgment
HR professionals
Learn to read data critically: know where it comes from, what biases it carries and how the algorithm turns people into numbers. Ask the right questions before asking for answers.
Companies
Data and AI governance with an owner, rules and clear roles. Systems that talk to each other, and pilots before rolling out a tool, as the interviewees learned.
HR leaders
Lead the change: translate the new reality for teams, handle resistance and invest in reskilling. Design processes in which the machine calculates and people interpret.
Everyone
Keep a "humanistic skepticism": use AI as a powerful tool, never as an oracle. Make sure every algorithmic decision can be reviewed and contested by a person.
What this research does not say
- It only heard decision makers. HR directors and business partners spoke; employees, who are measured by the systems, were left out. The thesis calls for that research.
- Ten people from one sector. The pharmaceutical industry is strongly hierarchical and heavily regulated; the picture is not a statistic for other sectors.
- A snapshot of one moment. Tools change fast; the author bets that the human dynamics described will last longer.
- The hybridization crisis is a theoretical proposal. It organizes what the interviews show, but it was not measured or compared across companies. · our note
- The quotes are the author's English translations of interviews held in Portuguese. · our note
Credits
ASSI, Ilton Luiz. The hybridization crisis: people analytics as an actor in the resignification of Human Resource Management in the Brazilian Pharmaceutical Industry. Doctoral thesis (Business Administration), Mackenzie Presbyterian University, São Paulo, 2025. Advisor: Prof. Dr. Almir Martins Vieira.
The work was supported by CAPES (Brazil's federal agency for graduate education). The author used DeepL to improve the readability of the English text and reviewed everything; the NVivo license the author bought includes an AI tool, but the interpretation of the themes was the researcher's. This version simplifies the work for a general audience; the full quotes and references are in the original text.