I believe AI won’t contribute to increasing disinformation efforts.

Understanding the Impact of AI on Disinformation: A Closer Look

In recent discussions, many have voiced concerns that artificial intelligence might significantly amplify the spread of misinformation and disinformation online. The reasoning is straightforward: with AI’s ability to generate vast amounts of content at scale, the volume of persuasive or misleading material could skyrocket, overwhelming audiences and complicating efforts to discern truth from falsehood.

The Argument for Increased Misinformation

It’s true that AI tools can produce a significant amount of low-quality or misleading content, often referred to as “AI-generated slop.” When observing social media platforms, one might notice an uptick in such material, leading to the logical assumption that disinformation is also increasing correspondingly. The fear is that as AI-generated content becomes more prevalent, our exposure to deceptive narratives will escalate.

Challenging the Assumption

However, I believe this perspective may overlook certain nuances. Consider the simple act of browsing through short-form videos—whether on TikTok or similar platforms. When I spend a few minutes scrolling, I typically encounter around 100-150 clips. The introduction of AI-generated content doesn’t seem to push this number higher; the volume remains relatively stable.

While it’s true that AI can produce more content overall, the human brain remains selective. We tend to engage with content that interests or entertains us, regardless of its origin. Given the vast amount of disinformation that already exists—most of which has been generated by humans—adding more AI-produced material doesn’t necessarily increase the amount of disinformation that catches our attention.

Content Formats and Consumption Patterns

It’s also essential to recognize how misinformation often takes advantage of specific content formats. For example, edited videos or snippets with provocative statements—like a doctored clip of a politician or celebrity—can be highly convincing, especially when presented alongside familiar personalities or emotional cues. Sometimes, these can be subtle enough that viewers might not even realize they’re being misled.

That said, the impact of such content might remain limited, as audiences tend to consume media in familiar patterns. People usually stick to content they find engaging, whether it’s funny cat videos, clips of mishaps, or political commentary—regardless of whether it’s AI-generated or not. The overall proportion of exposure to disinformation may stay roughly the same as it has been over recent years.

Does AI Really Widen the Disinformation Gap?

The most compelling concern is that AI might produce convincingly fake clips—deepfakes, manipulated audio