Early attempts to create dedicated hardware for smart artificial intelligence have been criticized as, well, a bit rubbish. But here's an AI gadget in the making that's all about bullshit, literally: Finnish startup Binit applies the image processing capabilities of large language models (LLMs) to detect household waste.
AI for sorting the stuff we throw away to increase recycling efficiency at the municipal or commercial level has been catching the attention of entrepreneurs for a while (see startups like Greyparrot, TrashBot, Glacier). But Binit founder Borut Grgic believes that tracking household waste is uncharted territory.
“We're producing the first tracker for household waste,” he tells JS, likening the upcoming AI gadgets to a sleep tracker, but for your trash-throwing habits. “It is a camera vision technology supported by a neural network. That is why we use the LLMs for recognition of regular household waste objects.”
The early-stage startup, which was founded during the pandemic and has secured nearly $3 million in funding from an angel investor, is building AI hardware designed to live (and look cool) in the kitchen – mounted on a cupboard or wall near the trash can-related action takes place. The battery-powered gadget has built-in cameras and other sensors, so it can wake up when someone is nearby so they can scan items before they end up in the trash.
Grgic says they rely on integration with commercial LLMs – primarily OpenAI's GPT – to enable image recognition. Binit then tracks what the household throws away and provides analytics, feedback and gamification via an app, such as a weekly waste score, all aimed at encouraging users to reduce how much they throw away.
The team originally tried to train their own AI model to do waste recognition, but the accuracy was low (around 40%). So they latched onto the idea of using OpenAI's image recognition capabilities. Grgic claims that they get waste recognition that is almost 98% accurate after integrating the LLM.
Binit's founder says he has “no idea” why it works so well. It's not clear if there were a lot of images of trash in OpenAI's training data or if OpenAI can only recognize a lot of things because of the sheer amount of data it was trained on. “It's incredible accuracy,” he claims, suggesting the high performance I achieved during testing with the OpenAI model may be due to the fact that the scanned items are 'generic objects'.
“It can even tell with relative accuracy whether or not a coffee cup has a liner because it recognizes the brand,” he continues, adding, “So what we're actually letting the user do is pass the object to the front of the camera. So it forces them to stabilize it in front of the camera for a while. At that moment, the camera captures the image from all angles.”
Waste data scanned by users is uploaded to the cloud, where Binit can analyze it and generate feedback for users. Basic analytics will be free, but the plan is to introduce premium features via subscription.
The startup is also positioning itself to become a data provider of the things people throw away – which could be valuable information for entities like the packaging entity, assuming it can scale usage.
Still, there's one obvious criticism: Do people really need a high-tech gadget to tell them they're throwing away too much plastic? Don't we all know what we consume – and that we should try not to generate so much waste?
“They are habits,” he argues. “I think we are aware of it, but we don't necessarily act on it.
“We also know that it's probably good to sleep, but then I put on a sleep tracker and I slept a lot more even though I wasn't taught something that I didn't know yet.”
During testing in the US, Binit also says mixed waste was reduced by approximately 40% as users engaged with the waste transparency the product provides. So it thinks the transparency and gamification approach can help people transform deeply ingrained habits.
Binit wants the app to be a place where users can get both analytics and information so they can reduce the amount they throw away. For the latter, Grgic says they also plan to tap LLMs for suggestions, taking into account the user's location to personalize the recommendations.
“The way it works is – let's take the packaging for example – so for every piece of packaging that the user scans, a little card is formed in your app and on that card it says this is what you've thrown away. [e.g. a plastic bottle]…and in your area, these are alternatives you could consider to reduce your plastic intake,” he explains.
He also sees room for partnerships, for example with food waste influencers.
Grgic argues that another novelty of the product is that it is “anti-unhinged consumption,” as he puts it. The startup is in line with the growing awareness and action in the field of sustainability. The feeling that our throwaway culture of single-use consumption needs to be jettisoned and replaced with more conscious consumption, reuse and recycling, to safeguard the environment for future generations.
“I feel like we are on the eve [something],” he suggests. “I think people are starting to ask themselves the questions: is it really necessary to throw everything away? Or can we start thinking about repairing [and reusing]?”
However, couldn't Binit's use case just be a smartphone app? Grgic states that it depends. He says some households like to use a smartphone in the kitchen if they get their hands dirty while preparing meals, for example, but others see the value of a dedicated hands-free waste scanner.
It's worth noting that they also plan to offer the scanning feature for free through their app, so they're going to offer both options.
So far, the startup has tested its AI waste scanner in five cities in the US (NYC; Austin, Texas; San Francisco; Oakland and Miami) and four in Europe (Paris, Helsniki, Lisbon and Ljubjlana, in Slovakia, where Grgic is located). originally from).
He says they are working on a commercial launch this fall – probably in the US. The price point they are targeting for the AI hardware is around $199, which he describes as the “sweet spot” for smart home devices.
