Have you ever wondered about the possibilities of having artificial intelligence (AI) so tangible that we could actually touch and play with it? That’s exactly the radical question students at MIT grappled with in a course called Interaction Intelligence (4.043/4.043). Run by Marcelo Coelho from the MIT Department of Architecture, the class was a breeding ground for ideas on integrating AI and physical design. The aim? To push the boundaries of AI from abstract, to physically interactive.
The students were tasked to introduce a game-changer – an AI actuality we could experience in the flesh. They went about creating large language objects (LLOs), a new breed of AI-driven physical interfaces. These LLOs are basically powerful language models that can interpret and learn from the environment, making interactions smarter and context-aware. But even with all its power, Coelho, who also leads the Design Intelligence Lab, notes that “they lack real-time, contextual understanding of our physical surroundings, bodily experiences, and social relationships.”
The challenge was to put the “physical” into these AI models. The idea was to finally bridge the gap between AI and IRL. The ultimate task was to make the language models understand and respond to the environment and the people within it. It’s AI, but “in the wild” if you would.
A cooking experiment was the backdrop for students Jacob Payne and Ayah Mahmoud’s AI(application). Drawing inspiration from the 1969 Honeywell 316 Kitchen Computer — a novelty item that was a statement more than a successful sell — they thought of tailoring an AI device that could take the hassle out of cooking and make the experience more individualized. The challenge was certainly placed on a popular platform – the kitchen.
They birthed theKüche Cosmo, their own take on creating a smooth, AI-linked recipe generator. Envision this: a device that could scan your kitchen counter with a webcam, take stock of the ingredients available, and voila — custom recipes at your service. The device would even consider intricate factors like your cooking timetable, culinary skills, diet, or even the mood you’d like to set with your cooking.
Despite the challenges in fine-tuning AI to understand the complexities of taste profiles, regional cuisines, and cooking techniques, Mahmound and Payne kept trying. After different recipes, nasty surprises, and pleasant finds, they retouched the system to better mirror real-world cooking dynamics. All the while, Kitchen Cosmo’s tactile interface provided users a sense of agency, making the AI appear as more of a partner than a robotic assistant.
With an aesthetic nod towards its 1960s antecedent, the Kitchen Cosmo is a culmination of rigorous prototyping and reiteration. It stands out in its vibrant red 3D-printed form, featuring a flip-open webcam, a thermal printer to print recipes, and a storage area for recipe cards.
After creating a buzz in the design community, Mahmoud and Payne have set their sights on further enhancements for Kitchen Cosmo. With ideas from real-time cooking feedback to a multi-user mode that assigns tasks to different chefs, they have exciting things in the pipeline. For Mahmoud, this experience has been nothing short of transformative, aligning her towards a future in design. She ends with, “I see myself doing design long-term, which is something I didn’t think I would have said previously about technology.”
To know more about this exciting project, head over to MIT-Nachrichten.
Diese Website verwendet Cookies.