People have said that the first efforts to make hardware specifically for artificial intelligence were, well, not very good. On the other hand, here’s an AI tool that’s just plain rubbish: A Finnish company called Binit is using the picture processing power of large language models (LLMs) to keep track of trash around the house.
Business people have been interested in using AI to sort the trash so that recycling is more efficient at the municipal or business level for a while now (see Greyparrot, TrashBot, and Glacier). But Borut Grgic, the founder of Binit, thinks that tracking trash is an open market.
“We’re making the first household waste tracker,” he tells TechCrunch, comparing the new AI gadget to a sleep tracker but for how often you throw away your trash. “It’s using a camera and a neural network to make it work.” So we’re asking the LLMs to help us recognize common pieces of home trash.
The early-stage company was started during the pandemic and has raised almost $3M from an angel investor. They are making AI hardware that will live (and look cool) in the kitchen, fixed on a cabinet or wall near where things happen that involve bins. The device runs on batteries and has cameras and other sensors built in so it can wake up when someone is close and scan things before they are thrown away.
To do picture recognition, Grgic says they plan to connect with commercial LLMs, mostly OpenAI’s GPT. Then, Binit keeps track of what the family is throwing away and uses analytics, feedback, and game-like features in an app, like a weekly trash score, to get people to throw away less.
The team first tried to teach their own AI model to recognize trash, but it was only about 40% accurate. So they jumped at the chance to use OpenAI’s picture recognition tools. Grgic says that after adding the LLM, they are getting trash recognition that is almost 98% correct.
The person who made Binit says he has “no idea” why it works so well. It’s not clear if OpenAI’s training data had a lot of pictures of trash or if it can spot a lot of things because it has been trained on so much data. In tests with OpenAI’s model, they got great results, he says, “it’s incredible accuracy.” This might be because the things they scanned were “common objects.”
He continues, “It can even tell pretty accurately if a coffee cup has a lining or not because it recognizes the brand.” To use it, all the user has to do is put the object in front of the camera. For a short time, they have to steady it in front of the camera. At that very moment, the camera is taking pictures from every angle.
When users scan trash, data about it is sent to the cloud, where Binit can look it over and give users input. It will be free to use basic analytics, but it plans to charge for more advanced tools.
The company is also planning to become a provider of data on the things people throw away. This could be useful information for groups like the packaging group, as long as it can handle large amounts of use.
Some people may still question whether or not people need a high-tech gadget to tell them they’re putting away too much plastic. It’s clear that we all know what we’re eating and that we should try to make less trash.
He says, “It’s habits.” “I believe we know about it, but we don’t always do anything about it.”
And we all know that sleep is probably good for us. But when I wear a sleep tracker, I sleep a lot more, even though it didn’t teach me anything new.
Binit also says that during tests in the US, the amount of mixed bin waste dropped by about 40% as people used the product’s trash visibility feature. So it thinks that its method of being open and making things fun can help people break bad habits.
Binit wants the app to be a place where people can discover data and tips that will help them throw away less. For the second one, Grgic says they also plan to use LLMs to get ideas, personalizing the suggestions based on the user’s location.
“Here’s an example of how it works: for every piece of packaging a user scans, your app makes a little card that says this is what you threw away, like a plastic bottle.” “These are alternatives you could think about in your area to cut down on the amount of plastic you use,” he says.
He also sees room for relationships, like with people who get people to waste less food.
One of the things that Grgic says is unique about the product is that it’s “anti-unhinged consumption.” The new business is in line with people becoming more aware of and taking action on environment. A feeling that our culture of single-use consumption needs to go away and be replaced with more mindful consumption, recycling, and reusing things. This will help protect the earth for future generations.
He says, “I feel like we’re on the edge of [something].” There are some people who are starting to wonder if they really need to throw everything away. If not, can we start to think about fixing and reusing?”
But wouldn’t Binit’s use case be just a smartphone app? Grgic says that it varies. He says that some families are fine using their phones in the kitchen while they’re doing things like making meals, but others think it would be better to have a special hands-free trash scanner.
Not only that, but they also want to give you the option to scan things for free through their app, so you will have two choices.
New York City, Austin, Texas, San Francisco, Oakland, and Miami are the five places in the US where the startup has been testing its AI trash scanner so far. Paris, Helsniki, Lisbon, and Ljubjlana, Slovenia, where Grgic is from, are the four cities in Europe.
He says that a business launch is planned for this fall, most likely in the US. He says that $199 is the “sweet spot” price for smart home products, which is what they want to charge for the AI hardware.
Also Read: Voice Cloning of Politicians is Still as Easy as Pie
A mistake was made in this report: Ljubjlana is in Slovenia, not Slovakia. The mistake is regretted.
What do you say about this story? Visit Parhlo World For more.
