The dark side of progress or why it's so important to develop critical thinking.
A recent report by Public Citizen, a non-governmental consumer protection organization, raises a worrying question about the growing popularity of wild mushroom picking and the associated risks, compounded by people's over-reliance on artificial intelligence (AI). Experts warn of serious consequences, including cases of poisoning, when users rely on AI technologies to determine the edibility of mushrooms.
The report points to cases where people who trusted AI to identify fungi from photos ended up in the hospital after consuming misidentified species. In North America, there are about 5,000 species of "fleshy" mushrooms (which are characterized by a soft consistency, cut and break well), 15 of which are deadly, and another 60 are toxic.
Despite the fact that online resources and social networks have helped mushroom pickers avoid danger for many years, the surge in interest in AI has led to the opposite effect. People have started turning to AI for advice on mushrooms, relying on its comprehensive knowledge and everyday wisdom, but at the moment the technology is far from perfect.
According to a report published by the researchers, there are currently more than a dozen AI-based mushroom identification apps available in the official Google and Apple stores. Many of these applications are not able to take into account very important factors, such as the substrate on which the fungus grows, the characteristics of its damage, and the smell. Even if it is established that the mushroom is edible, this does not mean that its use will not cause harm to health.
The report cites a recent study by Australian scientists who tested three AI-based apps by feeding them photos of 78 mushrooms. Picture Mushroom was the most effective app, but even it accurately identified mushrooms only half of the time and correctly identified toxic mushrooms only 44% of the time.
Experts note that while the ability of AI-based applications to identify almost half of all wild mushrooms tested is impressive, this "impressive" accuracy is the reason for the high risk of using them as the main or even the only means of identifying mushrooms. Overestimating trust in AI-based applications can lead to serious consequences.
The report ends with the conclusion that poisonous mushrooms are just one example of how over-reliance on AI technologies to obtain reliable information can be harmful. As AI becomes more integrated into our lives, it is imperative to develop critical thinking and approach the use of such technologies with awareness and caution.
A recent report by Public Citizen, a non-governmental consumer protection organization, raises a worrying question about the growing popularity of wild mushroom picking and the associated risks, compounded by people's over-reliance on artificial intelligence (AI). Experts warn of serious consequences, including cases of poisoning, when users rely on AI technologies to determine the edibility of mushrooms.
The report points to cases where people who trusted AI to identify fungi from photos ended up in the hospital after consuming misidentified species. In North America, there are about 5,000 species of "fleshy" mushrooms (which are characterized by a soft consistency, cut and break well), 15 of which are deadly, and another 60 are toxic.
Despite the fact that online resources and social networks have helped mushroom pickers avoid danger for many years, the surge in interest in AI has led to the opposite effect. People have started turning to AI for advice on mushrooms, relying on its comprehensive knowledge and everyday wisdom, but at the moment the technology is far from perfect.
According to a report published by the researchers, there are currently more than a dozen AI-based mushroom identification apps available in the official Google and Apple stores. Many of these applications are not able to take into account very important factors, such as the substrate on which the fungus grows, the characteristics of its damage, and the smell. Even if it is established that the mushroom is edible, this does not mean that its use will not cause harm to health.
The report cites a recent study by Australian scientists who tested three AI-based apps by feeding them photos of 78 mushrooms. Picture Mushroom was the most effective app, but even it accurately identified mushrooms only half of the time and correctly identified toxic mushrooms only 44% of the time.
Experts note that while the ability of AI-based applications to identify almost half of all wild mushrooms tested is impressive, this "impressive" accuracy is the reason for the high risk of using them as the main or even the only means of identifying mushrooms. Overestimating trust in AI-based applications can lead to serious consequences.
The report ends with the conclusion that poisonous mushrooms are just one example of how over-reliance on AI technologies to obtain reliable information can be harmful. As AI becomes more integrated into our lives, it is imperative to develop critical thinking and approach the use of such technologies with awareness and caution.
