ChatGPT as a Sensory Taster? AI’s Role in Brownie Evaluation

penAI's ChatGPT has been widely adopted across various industries, from assisting in academic research to accelerating drug discovery. Its ability to generate Studio Ghibli-style images has even strained OpenAI’s GPU infrastructure due to overwhelming demand. Now, ChatGPT is exploring an unexpected new domain—sensory evaluation in the food industry.

AI and Brownie Sensory Testing A recent study conducted by researchers at the University of Illinois Urbana-Champaign investigated ChatGPT’s potential as a sensory evaluator for brownies. Sensory evaluation, also known as organoleptic analysis, is a critical process in the food industry that assesses a product’s impact on human senses, including taste, aroma, texture, and even emotional response. Typically, this is done through panels of experts and consumer testers, but it’s a time-consuming and costly process.

Assistant professor Damir Torrico and his team tested ChatGPT by asking it to evaluate fifteen different brownie recipes based on sensory attributes. The AI provided ratings on a scale of 1 to 10 for sentiments such as joy, surprise, sadness, and even fear, mimicking typical consumer responses.

Enhancing, Not Replacing, Human Tasters The goal of using AI in food science isn’t to replace human sensory panels but to streamline the process. ChatGPT could help food researchers and product developers narrow down recipe variations, allowing human testers to focus on the most promising ones. The study concludes that AI could serve as an efficient pre-screening tool, reducing both time and costs involved in sensory evaluation.

However, challenges remain. The researchers noted that ChatGPT displayed a positive bias even for unconventional ingredient combinations. For instance, some brownie recipes in the study contained worm meal, fish oil, or citric acid, ingredients not typically found in brownies. While a human taster might express disgust or hesitation, ChatGPT maintained a relatively neutral or even optimistic stance, which suggests a need for further refinement in AI-driven sensory models.

The Future of AI in Food Science While AI has already made its mark in content creation, research, and image generation, its integration into food product development is still in its early stages. Fine-tuning ChatGPT for more nuanced and realistic sensory assessments could revolutionize product testing in the food industry.

That said, AI models have their limitations. Just as Google’s Gemini once recommended using glue on pizza, AI-generated food advice still requires careful human oversight. With further refinement, AI tools like ChatGPT could significantly enhance the innovation and efficiency of food science—without replacing the essential human touch.