How Algorithms Impact Social Media Accessibility

Calling out LinkedIn This month is the 35th anniversary of the ADA, and what’s happening on this platform feels as if it’s going against what we disabled people have strived for. LinkedIn's Algorithm: Prioritizing "Flash" Over Access? My LinkedIn impressions plummeted, going from thousands of views to under 2000. The reason? Not content quality, but a clear algorithmic shift favoring visuals over text. This isn't just cosmetic; it's exclusionary. For those relying on screen readers or braille displays, plain text is paramount. Yet, it's now deprioritized. The irony is stark: "visual-first" often prioritizes "exclusive engagement", not true inclusion. Carousels and videos, while boosting views, often neglect essential captions, alt text, or tactile readability. This oversight impacts low-vision and neurodivergent individuals, not just those using screen readers. This is not ok. This is an ableist shift. When the algorithm chases "flash" over function, it doesn't just drop impressions, it drops people. “Universal access” doesn’t work and is not the answer. It’s exclusive, not inclusive. It's time to advocate for access-centered design and inclusive digital spaces. Have you experienced this shift? What can we do to fix this issue to ensure this platform is equitably accessible and inclusive?? #WCAG #ADA35 #ADA #DisabilityInclusion #Ableism #AbliestDesign #AI #InclusionMatters #AccessMatters #DisabilityPrideMonth Tagging; Disability:IN Disabled by Society Disability Solutions Web Content Accessibility Guidelines (WCAG) National Disability Institute ADA National Network What will you do to help rectify this problem?

I noticed my LinkedIn engagement drop months ago. At first, I did what we all do. I blamed the algorithm. I assumed a platform tweak, a timing issue. I talked about it with others. Many of us saw the same thing. Still, it never crossed my mind that the reason might be… me. Then I read this TechCrunch article. And suddenly the question shifted from what changed to who the algorithm prefers. According to the reporting, LinkedIn’s new AI-driven ranking system, designed to surface the “most valuable” content, appears to systematically favour male-generated content. Not intentionally. Not maliciously. Just… structurally. Because this isn’t about LinkedIn alone. It’s about what happens when opaque AI systems start deciding whose voice is amplified, whose ideas circulate, and whose work is deemed “valuable”. And we’re told to trust the outcome. But how can we, when we can’t see: - what data the model was trained on - which signals it rewards - how bias is measured, challenged, or corrected This is what black-box AI looks like in real life. Not a dramatic failure. Not a headline scandal. Just a slow, silent reintroduction of old biases scaled, automated, and deniable. AI models are no longer neutral tools. They act. They rank. They filter reality. And when they do that without transparency, they don’t just reflect society, they reshape it. We’ve spent decades challenging gender bias as humans. Seeing algorithms quietly undo that progress, without explanation or accountability, feels like watching history loop back on itself, faster and harder to contest. LinkedIn might seem like a small example. Lower engagement isn’t life or death. There’s no immediate ROI attached. But that’s exactly why it matters. Because if we accept this here, in visibility, voice, influence, we normalise it everywhere else: hiring, funding, promotion, credibility. This is why transparency isn’t a “nice to have”. Why explainability isn’t academic. Why accountability can’t be optional. And this is why the AI Act matters. We’re still early. The signs are already there. #GenderBias #LinkedInAlgorithm #ResponsibleAI #Transparency #WomenInTech The techCruch article: https://lnkd.in/eKzxSTMZ

Our new paper, “Detecting bias in algorithms used to disseminate information in social networks and mitigating it using multiobjective optimization,” is now out in PNAS Nexus. Algorithms that determine how information spreads — from health campaigns to social media posts — are often optimized for one goal: maximizing reach. But what happens when reach comes at the expense of equity? Paper: https://lnkd.in/epNgJSnP ArXiv: https://lnkd.in/eAkVWwAK In this work, led by Vedran Sekara and with Ivan Dotu, Manuel Cebrian, and Manuel García-Herranz, we show that state-of-the-art influence maximization algorithms — the same kind used to identify “influencers” in social networks — systematically leave parts of the network behind. Some groups receive information late, others not at all. In other words, algorithmic bias can emerge not from data, but from the mathematical definition of the problem itself. To address this, we developed a multiobjective algorithm that balances spread and fairness. The result: we can significantly reduce informational inequality with only a minimal loss in reach. This suggests that optimization and equity don’t have to be opposing goals. As algorithms increasingly shape who gets access to opportunities, resources, and knowledge, this kind of fairness-aware design becomes essential — not just for social media or marketing, but for public health, disaster response, and social resilience.

What if the greatest irony of the AI age is this: Human civilization was built through “cut, copy, and paste.” We learned by repetition. By imitation. By precedent. By recursive improvement. A child repeats language. An apprentice mirrors a master. Lawyers study prior rulings. Programmers reuse code. Scientists build upon earlier discoveries. Civilization itself is one massive inheritance system of shared knowledge. Yet modern algorithms are increasingly treating those same human behaviors as suspicious. Structured writing? Downranked. Repetitive cadence? Flagged. AI-assisted communication? Questioned as “inauthentic.” That becomes dangerous when AI is not merely convenience, but accessibility. For many people, AI functions as: • a cognitive prosthetic, • a writing bridge, • a speech amplifier, • a memory aid, • or a tool for neurodivergent communication. So what happens when platforms suppress users because their communication appears “too assisted”? At what point does spam detection become algorithmic discrimination? This article explores a growing societal risk: the possibility that algorithms are beginning to standardize what “acceptable human communication” looks like. The deeper issue is not AI replacing humanity. It is humanity slowly learning how to behave for the machine. And once systems begin rewarding statistically “safe” language while suppressing irregular human expression, we may be engineering a civilization-scale feedback loop of synthetic conformity. That is not merely a technology problem. It is an accessibility issue. A governance issue. A labor issue. And potentially a human rights issue. #AI #Accessibility #DigitalRights #LinkedIn #Algorithms #HumanCenteredAI #Governance #Neurodiversity #Innovation #Technology

🚨 Why women’s health content is still being suppressed online Women’s health faces two major challenges in digital spaces: 1️⃣ Misinformation 2️⃣ Visibility barriers driven by censorship and algorithms Both directly limit access to care, education, and innovation. 🧠 The Gender Data Gap Decades of underinvestment in women’s health research have created significant gaps in clinical knowledge. Those gaps shape medical consensus and public understanding, leaving both providers and patients with limited evidence-based guidance for many women-specific conditions. 🤖 Censorship & Algorithms A recent Forbes article highlights what many in women’s health already know firsthand: medically accurate content about menstruation, fertility, menopause, sexual health, and reproductive care is routinely de-prioritized by algorithms and sex-safety filters. Meanwhile, comparable men’s health content is allowed to run freely. This is not loud censorship. It is quieter and more damaging. 📉 The Visibility Trap Algorithms reward simplified, surface-level women’s health content over medically rigorous education, forcing some founders to choose between discoverability and clinical integrity. 💥 The Impact When platforms flag women’s health terms as “sensitive,” the consequences are real: • Lower search rankings • Reduced social reach • Fewer trusted resources surfacing when people seek answers • Women’s health startups struggling with discovery, growth, and funding visibility 💬 My take: When clinically accurate language about women’s bodies is treated as unsafe by algorithms, visibility becomes a barrier to care, credibility, and trust. The result is a distorted information landscape where women must work harder to find evidence-based guidance and where women’s health innovation is harder to see, fund, and scale. #WomensHealth #AlgorithmicBias #HealthEquity #FemTech #DigitalHealth #AccessToCare https://lnkd.in/gyxupZ9Y