Could AI News Be Skewed By Different Governments?

We often think of artificial intelligence, or AI, as a neutral tool, simply processing information and giving us answers. But new research is shining a light on a fascinating and somewhat concerning issue: how the information AI is *trained* on can actually shape its 'opinions' or outputs, especially when it comes to news and current affairs.

The study suggests that if an AI is fed information primarily from state-controlled media – think news outlets where the government has a strong say in what gets published – then the AI might start to reflect those viewpoints. Essentially, the AI could end up giving more positive responses about government actions or institutions than it would if it had access to a wider, more independent range of news sources. This is particularly noticeable in places with less press freedom, like China, where media heavily reflects state narratives.

For everyday Australians, this is worth paying attention to. As we increasingly rely on AI tools, like advanced chatbots, to summarise news or answer questions, we need to be aware that the 'answers' these AIs provide aren't always a purely objective reflection of reality. They are influenced by the data they've been trained on, and if that data has a bias, the AI's output might too.

It also highlights a bigger picture problem. Governments, especially those controlling media, might see this as an incentive to expand their influence over the information landscape. If they can shape what goes into AI models, they can subtly influence the global conversation. As AI becomes more ubiquitous, ensuring diverse and unbiased training data becomes crucial for maintaining truthful and open access to information for everyone.