Generative AI is harming the internet, say Google researchers

Researchers from the DeepMind laboratory, which belongs to the same group as Google, have published a scientific article on the possible harms of artificial intelligence (AI). According to the text, there is evidence that such content harms navigation.

Article “Abusing Generative AI: A Taxonomy of Tactics and Insights from Real-World Data” It assessed 200 tactics for the harmful use of generative AIs, catalogued between January 2023 and March 2024. It is not yet part of a scientific journal because it still needs to be reviewed by other scientists.

The study provides evidence and reinforces a perception that many users hold: There is an increasing amount of artificial images and text is surfing the internet. This is even noticeable in Google search engine results, and it is part of living in a digital environment, which brings with it some very negative consequences.

How AI is making the internet worse

According to the article, it is possible to say that this kind of malicious use of technology is happening.”blurs the lines between authentic and misleading presentations“. Even if most of the generative AI content found so far is not unregulated, this one has already been identified in a significant and “significant” way.

Among the most common cases encountered manipulation of human activity, falsification of evidence This use in fraud or other types of cybercrime.

These are usually “reputation or financial gains” and for now they mostly rely on text only, as other visual tools are still in their nascent stages of availability.

Moreover, such use of generative AI could have large-scale implications in terms of:Trust, authenticity and integrity of information ecosystems” — that is, damaging content circulation environments such as social networks and even search engines like Google itself.

Researchers have suggested possible solutions:multifaceted approach“It involves collaborations between government, industry experts, scientists and the community.

Moreover, not only are technical advances needed, but also ““deeper understanding of social and psychological factors” This involves the creation of fake content. You can read the full article at this link (in English).