Of all the things I’d want an AI to help me with, this might just be the most useful. While companies like Google and OpenAI have been showing their AI’s multimodal capabilities at identifying the color of objects, or human expressions, an AI hardware company is putting the power of LLMs to good use by having them identify trash. Binit, a Finnish startup, has developed a tiny hardware device that scans your trash and then tells you how to accurately dispose of it. It doesn’t just identify objects like coffee cups and vegetable peels, it tries to analyze materials, looks at brands to identify their manufacturing techniques to determine the best ways to dispose of trash, and effectively tells you (with 98% accuracy) how to segregate your household waste so that the wrong items don’t get disposed of in the wrong channels.
Binit’s core technology is a dedicated AI-powered hardware device designed for the kitchen. Mounted conveniently near your bin, this gadget uses camera vision and sensor technology to identify items before they’re tossed. The system integrates with powerful large language models (LLMs) like OpenAI’s GPT-3, enabling highly accurate image recognition of everyday household waste objects.
Designer: Binit
Binit’s secret weapon? Rather than training their own AI model from scratch, they capitalize on the vast knowledge base of pre-trained LLMs. This approach has yielded impressive results – Binit claims near 98% accuracy in trash recognition. The system can even distinguish nuances – like whether a coffee cup has a lining based on brand recognition.
But Binit isn’t just about identifying your trash. The system analyzes the data it collects, generating insightful feedback through a companion app. This app provides users with a weekly “rubbish score” and gamified elements to motivate positive behavioral changes. Imagine a friendly nudge reminding you that you’ve been discarding a lot of plastic lately, along with suggestions for sustainable alternatives.
The free tier of the Binit app offers basic analytics, while a premium subscription unlocks additional features. The company also envisions itself as a potential data provider – offering anonymized insights on consumer waste habits to entities like packaging manufacturers. This data could be invaluable for businesses aiming to reduce their environmental footprint.
One might wonder – isn’t environmental awareness enough? Binit argues that while most people acknowledge the importance of waste reduction, ingrained habits are hard to break. They liken their approach to sleep trackers – providing data-driven insights that can empower users to make positive changes, even if they’re aware of the general principles.
Binit’s pilot programs across the US and Europe have shown promise. Users reportedly reduced mixed waste by around 40% after engaging with the system’s transparent feedback loop. The company plans to launch commercially this fall, likely in the US, offering both a dedicated AI hardware device and a free app-based scanning option.