AI hiring bias and automation threaten 37 million Brazilian workers
AI systems can filter applicants and displace operational workers. Up to 37 million Brazilian workers could be affected, while experts call for system audits, shared design and legal accountability.
Wire notes
- Up to 37 million Brazilian workers, about 37% of positions, could be affected by AI, according to Fundação Grupo Volkswagen.
- World Economic Forum data list administrative routines, customer support and telemarketing as functions highly exposed to rapid automation.
- Silva and Fico identify office proximity and attendance at a specific college as hidden filters in algorithmic recruitment.
- PwC Brasil's “AI Jobs Barometer” links new technology requirements to barriers facing workers aged 50 and older.
- Silva recommends codesign, with HR professionals and minority-group representatives involved in algorithm development.

By Elena Vasquez
Up to 37 million workers in Brazil, about 37% of positions, could be affected by artificial intelligence (AI), a Fundação Grupo Volkswagen study says. Silva and Fico link digitalization and AI in hiring and operational work to historical inequality in access and employment.
Recruiting algorithms promise agility and lower costs, but can reproduce, and sometimes accelerate, disparities based on gender, race and age, according to Humberta Silva and Bernardo Fico. According to the two experts, the groups most exposed include women, Black people, young people from peripheral areas, people with disabilities, LGBTI+ workers and older employees. Silva identifies hiring as the first major exclusion bottleneck.
How do algorithmic filters exclude candidates?
Silva and Fico describe recruitment platforms that use algorithms to screen résumés. According to their analysis, hidden criteria can include whether an applicant lives near the office, attended a specific college or uses the vocabulary the machine expects during screening. Those filters silently favor people with greater cultural capital (credentials, knowledge and language tied to social advantage).
Silva and Fico also identify limited technology access and distance from the language expected by machines as immediate disadvantages for candidates. Their warnings focus on who can enter the process and which signals the system treats as evidence of suitability.
Humberta Silva, a professor at Germany's Hochschule Bremen, researches and consults on AI's effects in human resources (HR). She says the technology's distance from candidates limits the assessment of subjective capacities.
“The process is more distant and dehumanized. That creates limits on an accurate assessment of a candidate’s capabilities because subjective aspects are difficult to identify. We need to understand that artificial intelligence has limitations, identify them and always conduct an audit when it is used,” Silva said.
According to Silva, the change also weakens the role of HR professionals. Managers lose confidence and become mechanical system administrators, rather than exercising judgment about candidates.
Can algorithm decisions hide discrimination?
Fico says labor laws provide established accountability when people discriminate. His legal analysis identifies a blind spot when companies delegate selection to software, because employers can cite a technical metric when evidence is needed that a disabled professional or Black woman faced exclusionary treatment.
Bernardo Fico is a Digital Law specialist with a master's in Human Rights from Northwestern University in the United States. He argues that this opacity requires urgent corporate accountability and rejects algorithms as legal shields for unequal selection.
“The problem begins at the root of technological development. Today, the overwhelming majority of creators of these tools are white men, a direct reflection of the lack of diversity in technology leadership. Without a plurality of voices involved in writing the source code, platforms start blind to the realities of plural groups,” Fico said.
Fico links this programmed blindness to the urgent need for transparency. He says automation is excluding workers on an unprecedented scale, so the market and the justice system must address algorithmic discrimination (systematic distortion in automated decisions) with rigor.
His regulatory objective is to hold companies responsible for hidden machine criteria. Fico also wants regulators to end the assumption that AI decisions are neutral or legally exempt from scrutiny.
Which jobs and workers face the greatest risk?
AI also threatens operational roles after candidates enter a company. World Economic Forum data identify administrative routines, customer support and telemarketing among the functions most exposed to rapid automation.
According to the Forum, those roles have historically concentrated the largest shares of women, young people and lower-class workers, making them central to automation's unequal impact on employment.
Fundação Grupo Volkswagen's Brazilian study says AI reaches the base of the corporate pyramid first. The research also finds that the country has not absorbed professional requalification on the scale needed to protect affected workers.
PwC Brasil's recent “AI Jobs Barometer” shows how quickly employers began requiring new technological competencies. According to the consultancy, older professionals already face resistance in recruitment and career transition. They now face demands for generative artificial intelligence literacy, meaning the ability to use systems that generate content.
Without corporate bridges designed to help workers aged 50 and older, PwC Brasil says, technology becomes an age barrier and can force premature exits from the formal labor market.
What would a fairer AI transition require?
Silva says diversity must become a fundamental step in software engineering rather than institutional messaging. Her recommended model is codesign (shared design involving affected users), with HR professionals and representatives of minority groups included in algorithm development.
“The majority of systems are built by programmers without participation from HR, which is a serious problem. Diverse programmers can see other perspectives, reflect society's real context and avoid biases,” she said.
Silva calls recruitment the most strategic area for attracting diversity. She argues that diverse development teams can recognize broader perspectives, reflect society's real context and avoid bias.
Fico supports a robust regulatory framework and strategic litigation (lawsuits intended to change wider practice). He says diversity protection must sit at the center of debate about technological progress.
“The justice system and companies themselves need to address algorithmic discrimination through law today. Civil society and workers must demand a just transition, so AI does not become a legal shield for corporations,” Fico said.
For employers, the prescriptions set a concrete standard: disclose hidden criteria, audit outcomes, include affected groups in design and fund requalification before AI converts existing inequality into exclusion at scale.
via oglobo.com (Source)
More from Elena Vasquez
Show full bio
Senior reporter covering technology and product news in marketing, advertising, marketing technology, media, SEO and search at Marketing Herald.
376 articles
More on the wire
- AI Hiring Tools Push 70% of Ethnic Minority Candidates to Mask CV Identity
- Google Gives Schools an AI Off Switch, But Not for Search Users
- Quaest trains AI agents to simulate Brazilian voter profiles
- Stanford AI100 Report: Autonomous Warfare and Synthetic Media Are Here
- AI Changes Public Relations Search, Visibility and Trust