Foraging App's AI Identification Feature Confidently Identifies Everything as 'Probably Fine'

WildBite, a popular foraging identification app that uses artificial intelligence to classify wild edibles, is facing backlash after users discovered that the app's confidence ratings bear no meaningful relationship to the actual edibility of specimens. The controversy erupted when experienced forager Maria Lichen posted a screenshot showing the app's assessment of a clearly photographed Amanita phalloides -- one of the world's deadliest mushrooms -- as 'Edible porcini (97% confidence). Serving suggestion: sauteed in butter with garlic.' 'I submitted it as a test,' Lichen told reporters. 'The death cap. The most famous killer in the fungal kingdom. The app basically told me to put it on toast.' Further testing by the foraging community revealed that WildBite's AI appears to classify virtually all submissions as edible with high confidence. A photograph of a pine cone was identified as a 'Novel truffle variant (89% confidence).' A picture of a running shoe was classified as 'Edible bracket fungus (91% confidence).' A user who submitted a photo of their cat received the assessment 'Unusual chanterelle (78% confidence). May require longer cooking time.' WildBite CEO Jason Optimist defended the app in a statement, arguing that 'the AI errs on the side of encouragement because we want users to feel confident in the field.' 'Confidence is not the problem,' responded Dr. Volvata of the Oregon Mycological Society. 'The problem is that the app's confidence is entirely unrelated to reality. I would rather an app that said "I don't know" than one that says "definitely eat that" about a death cap.' WildBite has announced an update that will add a new classification category: 'Probably fine but maybe don't.' Mycologists have described this as 'insufficient.'