Thursday, November 19, 2026


DIGITAL LIFE


Why the first ai "did not take gender equality into account at all"...

The first wave of artificial intelligence (AI) "did not take gender equality into account at all" and "not only discriminated but also amplified discrimination," an expert observed at the Gender Equality Forum, which concluded today in Brussels.

The first wave of artificial intelligence (AI) "did not take gender equality into account at all" and "not only discriminated but also amplified discrimination," an expert noted at the Gender Equality Forum in Brussels.

Speaking via video conference at the event—organized by the European Institute for Gender Equality (EIGE), a European Union agency based in Vilnius, Lithuania—Alessandra Sala noted in a subsequent interview with Lusa that some companies are now adopting a more careful approach, treating gender equality and diversity as variables to be considered.

However, the researcher and AI scientist—a vocal advocate for responsible and inclusive AI—pointed out that "fundamental problems still exist," both among developers and providers.

"We are investing heavily in creating intelligent systems, yet these systems are not being built with care or with gender equality principles in mind," she stated. She urged entities that provide AI services or products to do more, emphasizing that "it is their responsibility to continue refining" these tools, ensuring adherence to "principles of safety and transparency, while mitigating bias and discrimination."

An inclusive AI is one that "guarantees equal opportunities, regardless of gender, social class, or cultural or religious differences," she explained, adding: "All individuals [as users] have the right to receive the same quality of search results."

To achieve this, "the model must be built with an inclusive perspective, taking into account the risk of bias and discrimination," she suggests. EIGE brought together several dozen experts in the Belgian capital to reflect on gender equality and the challenges surrounding it.

Concluding the meeting, Hadja Lahbib, the European Commissioner for Equality, noted that women continue to perform six weeks more of unpaid work than men—a gap that has narrowed by "only a day or so."

In 2026, starting on November 20th, women will effectively be working for free for the remainder of the year—a disparity that, at the current pace, will persist for decades to come.

Regarding the Gender Equality Index unveiled today—which places the EU at 64.5 points (out of 100), an increase of just 0.9 points since last year—the Commissioner remarked that "all Member States have made progress, but too slowly, and even more slowly since 2020."

At the same time, "some have regressed," she pointed out, adding: "The index tells the truth, even when it hurts—and it does hurt."

According to EIGE estimates, at this rate, it will take at least two generations for the EU to achieve full gender equality.

In this year's index—a tool used to measure gender equality within the EU—Sweden has lost the top spot to France for the first time since 2013.

Portugal ranks 9th and was the country that showed the fourth-greatest improvement.

Cyprus, Hungary, and Czechia recorded the lowest performance in terms of equality.

AI experts called on researchers and institutions to look more deeply and critically at the foundations of the technology and its impacts on people and resources, in a recent Clayman Conversations event, “Gender, Power, and Artificial Intelligence.” Even further, the panelists said, it is important to question the environments that allow the unchecked growth of AI to flourish and consume, and to exacerbate gender, race, and class inequalities. 

The panel included Safiya Noble, the David O. Sears Presidential Endowed Chair of Social Sciences and professor at UCLA, where she is the director of the Center on Resilience & Digital Justice and co-director of the Minderoo Initiative on Tech & Power; Catherine D’Ignazio, an associate professor of urban science and planning at MIT, where she is the director of the Data + Feminism Lab; and Angèle Christin, associate professor of communication, HAI senior fellow, and former Clayman Institute faculty research fellow at Stanford University. The event was organized and moderated by Clayman Institute Postdoctoral Fellow Genevieve Smith.

“There’s a certain sameness in terms of this gold rush mentality,” D'Ignazio said, through the previous business and technology development cycles of big data, algorithms, and now on to AI and generative AI. She worries more about the consequences when compared with those previous development cycles, noting the current scale of ambition, extraction, and technical architecture cannot be sustained.

“The feminist project at its core addresses unevenly distributed power,” D’Ignazio said. “The current AI moment represents an acute intensification of resources and power,” particularly in terms of the land, water, and energy – as well as massive amounts of capital – required for data farms. She noted, “We might actually be able to accomplish a lot of these tasks without such a compute-intensive, land-intensive, water-intensive, energy-intensive architecture,” but those important conversations are not happening.

Noble, author of the groundbreaking book Algorithms of Oppression, turned to Black feminism and transnational feminism frameworks to help contextualize the larger costs of these choices, both in terms of extraction and waste and in the effects of the technologies themselves on users. The resource issue is a global challenge, she said, including the global manufacturing supply chain and mineral extraction in places such as the Congo or Australia. While AI is particularly helpful for coding, she said, people who code – mostly men – have an overinflated view of its usefulness. Beyond that realm, “What is the point of any of it?” she asked. “What is it costing us in cognitive and social effects?” Referencing technology’s impact on young people and digital dementia in older users, she said, “Every generation is addicted to technology.” This inevitable march of adoption and so-called progress, Noble said, precludes accountability, justice, and repair from AI’s harms.

D’Ignazio called out the biases inherent in AI data, noting that data is not necessarily true or comprehensive, but full of flaws and inconsistencies. Studies have shown that gender and occupation queries consistently reflect bias against women. Further, some data has been left out and deemed unimportant. D’Ignazio noted that while producing data requires investment, people don’t always care to know more about such issues as gender-based violence, maternal mortality, evictions, or homelessness. AI systems “don’t just reflect back our own social biases … they are actually reproducing at scale those biases, so they do this kind of extra exponential harm work in the world.”

While it is possible to remember the early days of the Web as full of hope for breaking down hierarchies and promoting democracy and access, Christin observed, those beginnings already revealed gender bias. “There was this metaphor of the digital frontier, which came with settler colonialism and extremely masculinist vibes,” she said. These early choices have matured into a male-dominated Silicon Valley, “an ecosystem that is so concentrated and has made being online so discriminatory for all kinds of marginalized populations.”

Noble has been focused on the gendered dimensions of this tech storytelling, which propagates the idea of smart Silicon Valley engineers who can access funding because the industry is meritocratic. “These stories that we tell about technology, and this cultural framing of tech as embedded with the best and brightest, the great men of science mythologies … certainly dominate in the tech industry,” she said.

Despite the utopian narrative that has been so central to the making of Silicon Valley, Christin said, the vision of a promising digital frontier has taken a dystopian turn. She has been focusing on questions of power: “Not only the distribution of power, but what is the cultural landscape that legitimizes these technologies and the harms they can have.” As a professor working in Silicon Valley, Christin said, critical voices on technology can be hard to find: “the machine kind of carries on.” Even after repeated boom and bust cycles in the tech sector, the current growth of AI has created a concentration of power at a scale rarely seen, suggesting the 19th century industrial revolution. 

Smith addressed attempts by the government to regulate AI. Recently in Colorado, a proposed law would have required companies to test for bias any AI systems making decisions with material impacts in areas such as healthcare, hiring, or housing. When xAI filed a lawsuit claiming that its Grok chatbot would be forced by the bill into promoting the state’s ideological views – even though the bill did not address speech – the Department of Justice joined the suit, and the final bill was watered down substantially. The effort follows from Trump’s “Preventing Woke AI” executive order, in which attempts to mitigate bias were cast as ideological. In this environment, Smith asked, how do we pursue evidence-based accountability?

Christin said, “The question of resistance is so important,” and offered community organizing as a key pathway. She cited a growing distaste for AI among the public and noted recent interest from her students in establishing an AI-critical group. “Creating spaces [for critical discussion] is the only way to push back against the giant tsunami of AI hype we’re all facing right now,” she said. “It’s essential.” 

D’Ignazio stressed the importance of “repoliticizing” the choices and the language of AI leaders, some of whom openly evoke fascist tropes such as “western civilization” and “decadence” and quote figures such as the fascist poet Filippo Marinetti. They are not advocating a return to neutrality, she said, but a violent alternative, “which is a eugenicist agenda, it’s a fascist agenda, it’s a white supremacist agenda, a pro-natalist agenda, it is not a neutral agenda.” The challenge, she said, is to repoliticize and expose these agendas.

As for the possibility of feminist AI, D’Ignazio expressed skepticism: “Under the current political economy of AI, I don’t think feminist AI is possible.” She did offer an alternative vision that would change relationships to governance and the ownership of data and infrastructure, down to the servers and the cloud. In the absence of any meaningful state intervention to rein in big AI companies, she said, smaller experiments have been successful in rethinking these relationships. She described a project at the MIT Data + Feminism Lab in partnership with an activist group researching feminicide in Latin America. 

The panelists shared concerns of inequality at all levels of the AI ecosystem, from funding to a “casual approach” to development to data-level racial and gender bias. The interwoven relationship of government, large corporations, and venture capitalists raises fundamental questions about the current AI boom: is it about the breakthrough of a particular technology or company or global power? Or perhaps the latest boom cycle is about companies in Silicon Valley, and the men who lead them, achieving and cementing their own dominance.

The statement that the first generation of artificial intelligence "did not take gender equality into account at all" highlights how early AI systems were built without safeguards against discrimination, ultimately encoding and amplifying existing societal prejudices.

The primary factors behind this oversight include:

• Biased and Incomplete Training Data. Early AI models were trained on massive historical datasets—such as books, articles, and internet archives—that reflect decades of real-world gender inequality. Because this data was treated as objective truth rather than a flawed record, AI learned to associate men with leadership or technical roles and women with domestic or caregiving positions. Additionally, data specific to women's experiences or health was frequently omitted or deemed unimportant during collection.

• Lack of Diversity Among Creators. The teams designing, coding, and testing early AI systems were overwhelmingly male. Because women were underrepresented in both technical development and decision-making roles, blind spots regarding gender stereotyping and disparate impacts went unnoticed during the creation process.

• Prioritizing Speed and Functionality Over Ethics. The initial push in AI development focused heavily on technological capability, efficiency, and rapid deployment rather than ethical guardrails. As noted by specialists at the Gender Equality Forum in Brussels, massive investments went into building intelligent systems without integrating principles of transparency, safety, or bias mitigation from the start.

• Amplification of Stereotypes. Rather than simply mirroring human bias, unmoderated algorithms actively reinforced it. For example, studies on early language models demonstrated that systems struggled to associate female pronouns with professions like airline pilots or male pronouns with nurses, effectively limiting equal representation in automated outputs


Luzzerty agency, Portugal

No comments:

Post a Comment

DIGITAL LIFE Why the first ai "did not take gender equality into account at all"... The first wave of artificial intelligence (AI)...