Mushroom picking has always been a high-risk activity. Current technology has come to the rescue to avoid or minimize these risks as much as possible. If you take a stroll through the App Store or the Google Play Store, you will see many applications that use artificial intelligence to identify mushrooms and compare them to known varieties. However, we advise you not to rely too much on them.
The accuracy of these applications is, at the very least, questionable at first. At least that’s what a report developed by researcher Rick Claypool for Citizen (a non-profit consumer advocacy organization) indicates. And if that wasn’t enough, the applications also do not make their margin of accuracy very clear.
The sudden popularity that artificial intelligence has gained among the public has led us to see its applications in hundreds of sites, at any cost. The new language models and image generators have proven to be quite effective in their tasks, but there is something that AI is not very good at yet: providing accurate information.
Mushroom identification is one of those areas where the uncontrolled use of AI can be quite dangerous. However, companies and developers seem to care very little about it.
Last year, after an increase in poisonings from consuming poisonous mushrooms, a group of Australian scientists tested the accuracy of mushroom “detectors” to see how reliable they were. The most accurate of them managed to identify which mushrooms were poisonous with 44% accuracy. So, given what we’ve seen, the advice is clear: do not eat any wild mushroom until you have consulted an expert in the field.
Source: The Washington Post
