Scientists from Penn State University have recently unveiled a groundbreaking electronic tongue that excels at distinguishing between different tastes, surpassing human capabilities in taste perception. The team behind this invention claims that the device, which utilises graphene, has the potential to revolutionise the detection of chemical and environmental changes. This innovation could have wide-ranging applications, from medical diagnostics to identifying spoiled food.
One of the most remarkable aspects of this new technology is its ability to provide insights into the inner workings of artificial intelligence (AI), shedding light on a previously opaque realm due to the so-called black box problem. By reverse engineering the neural network’s decision-making processes when discerning differences in tastes of milk, coffee, and fizzy drinks, the researchers gained valuable understanding into the AI’s decision-making mechanisms.
Saptarshi Das, a professor of engineering science and mechanics at Penn State, highlighted the complexity of how humans perceive different foods, involving not just the tongue but a comprehensive neural network. The electronic tongue’s neural network achieved a tasting accuracy exceeding 95%, surpassing human-selected parameters. Through a method known as Shapley additive explanations, the researchers delved into the neural network’s intricate decision-making process, focusing on crucial data for taste assessment.
The neural network distinguished subtle characteristics in the data that humans may struggle to define accurately, leading to enhanced accuracy and consistency in taste assessment. For instance, in assessing milk, the network could detect variations in water content and identify any indications of degradation critical for food safety concerns.
The comprehensive research detailing this electronic tongue and its AI insights was published in the journal Nature under the title ‘Robust chemical analysis with graphene chemosensors and machine learning’. This innovative development has the potential to transform various industries, offering new opportunities for advancing AI safety and development.
The implications of this scientific breakthrough extend beyond taste perception, offering a glimpse into the intricate cognitive processes of AI systems. As technology continues to advance, understanding the inner workings of AI is crucial for ensuring its safe and reliable integration into various aspects of society.