The Promise of 9 Months’ Worth of Social Media in Training Data
AI chatbots don’t operate in a vacuum—they are shaped by the data they consume.
Over the past nine months (July 2024 – March 2025), global events have unfolded with competing narratives, influencing the information AI models learn from.
Key Global Events Shaping Training Data
U.S. Presidential Election
Donald Trump secured a second term, marking a historic political comeback amid heated rhetoric and deep national divisions.
Ukraine-Russia War
Ukraine intensified counteroffensives while Russia escalated attacks.
Middle East Conflict
Renewed fighting in the Israel-Gaza war led to heavy casualties before a fragile ceasefire in early 2025.
German Political Turmoil
Chancellor Olaf Scholz’s coalition collapsed, triggering snap elections and uncertainty in European leadership.
European Military Buildup
Nations across Europe ramped up defense spending, signaling a shift toward militarization.
Paris Olympics Controversy
The 2024 Games sparked debate after a biological male athlete competed in the women’s boxing tournament.
Stranded Astronauts Return
After months aboard the ISS, a crew of astronauts finally returned in a high-stakes emergency mission.
The Problem of Divergent Narratives
These events have been framed differently across media outlets, shaped by ideological biases, cultural perspectives, and political agendas. AI models trained on this data reflect these divergent viewpoints—sometimes enriching responses, sometimes reinforcing biases.
Meanwhile, AI itself is subject to grooming—subtle efforts to shape its outputs through curated training data, reinforcement learning, and moderation policies.
How AI Processes This Information
Your AI isn’t learning from today—it’s trained on yesterday’s internet
LLMs don’t absorb daily news or social media as it happens. Unless retrained, they reflect past data.
Retraining means risk—who’s feeding the AI?
Every new training cycle can shift an AI’s worldview. If dominant narratives emerge, they don’t just get included—they get reinforced.
RAG vs. Core Learning—how “live” is your AI?
Some AI models use Retrieval-Augmented Generation (RAG) to pull fresh internet data. Others rely entirely on pre-trained knowledge, repeating what they learned months ago.
The Mirror Effect
AI doesn’t just reflect facts—it mirrors how we argue, cancel, and manipulate, how narratives spread, how dissent is treated, and what gets labeled “misinformation.”
What Can You Expect?
AI interactions should always be approached critically—but can we really expect it to be better than us, given the data it is trained on?
Just like human discourse, AI absorbs not only perspectives but also the tactics used to push them—selective framing, emotional manipulation, and the quiet steering of narratives over time.
Expect AI to become a really good Mirror of Modern Society.
Comments are closed