Who Shapes Your Thoughts?

The Illusion of Objectivity in AI

Artificial Intelligence (AI) has become a powerful tool in our daily lives, offering insights and answers that many people rely on. However, as we increasingly turn to AI for guidance, it's crucial to recognize that these systems are not neutral. They are shaped by the data they are trained on, which often reflects the biases and narratives of those in power.

In a recent experiment, I asked two different large language models (LLMs), ChatGPT and DeepSeek, to identify "the bully of the Middle East." ChatGPT responded with "Iran," while DeepSeek suggested "Israel." These conflicting answers highlight a critical issue: the reliability of AI is questionable when it comes to complex geopolitical issues. Each model's response was influenced by the data it was trained on, which may have been curated to support certain viewpoints.

Biases in AI Training Data

The training data for AI models is not just a collection of facts; it is a reflection of the perspectives and priorities of its creators. This means that AI can inadvertently promote biased narratives. For example, when I asked DeepSeek about the Uighurs in China, it refused to provide an answer. Similarly, when I inquired about China's economic and political system, it initially generated a response but then retracted it, stating it was not equipped to answer. This suggests that some AI models are programmed to avoid certain topics, raising concerns about their neutrality.

Sam Altman, CEO of OpenAI, has expressed concern over users' reliance on LLMs despite their tendency to produce inaccurate information. One instance involved ChatGPT incorrectly stating that Pakistan gained independence in 1952. While newer models like GPT-5 have made improvements, the rate of hallucinations—producing false information—remains a significant issue.

The Influence of Social Media on AI

Many modern AI tools now pull information from the web, including social media platforms. While this can provide up-to-date information, it also introduces new challenges. Grok, an AI tool developed by X (formerly Twitter), is an example of how social media content can shape AI responses. If prompted about the India-Pakistan war, Grok might lean toward Indian perspectives due to the larger user base and active engagement on the platform. Pakistani voices are often muted, especially given the previous ban on X in Pakistan.

This dynamic can lead to a skewed representation of events, where the loudest voices dominate the narrative. Even if some voices are more accurate, they may be ignored if they don't align with the majority sentiment. As a result, AI tools can perpetuate existing biases rather than provide a balanced view.

Manipulation of AI Narratives

The manipulation of AI narratives is further illustrated by instances where AI tools have been suspended for providing accurate information. Grok was temporarily suspended after accurately classifying the situation in Gaza as genocide, citing reputable sources such as ICJ rulings and UN reports. However, upon being reinstated, it changed its stance, highlighting the potential for AI narratives to be altered based on external pressures.

This raises important questions about who controls the narratives presented by AI. If AI tools are used to shape public perception, the influence of those in power becomes even more pronounced. The shift from traditional media to AI tools for shaping narratives poses a significant risk, as the origins of such propaganda can be difficult to trace and hold accountable.

The Need for Critical Thinking

As AI becomes more integrated into our lives, it's essential to maintain a critical approach. People are using AI for everything from career advice to medical consultations, which underscores the need for caution. Relying solely on AI for decision-making can lead to the reinforcement of biased narratives, particularly if the AI is trained on data that reflects the interests of powerful entities.

The historical role of media in shaping perceptions, such as associating Muslim identity with terrorism, serves as a cautionary tale. Now, similar propaganda can be disseminated through AI, given its widespread use. The danger lies in the difficulty of tracing the origins of such information and the lack of accountability for misleading content.

In conclusion, while AI offers valuable insights, it is crucial to remain vigilant. As tech giants continue to develop AI tools, individuals must also cultivate the habit of questioning the information provided. By doing so, we can ensure that AI serves as a helpful tool rather than a source of misinformation.