Solid State of Mind: a start-up that wants to create truly intelligent AI

Solid State of Mind: a start-up that wants to create truly intelligent AI

“It is possible to create AI that consumes less energy and less data by drawing on biology, cognitive psychology and neuroscience to develop a new approach,” says Solid State of Mind president and co-founder Maxime Julian. (Photo: Solid State of Mind)

Artificial intelligence (AI) has made huge strides, but it needs a paradigm shift to better adapt to the real world, according to Montreal startup Solid State of Mind. He therefore proposes a new direction that could revolutionize the industry.

“Going down different paths can lead to great innovation,” believes its president and co-founder Maxime Julien in an interview. I’m tired of hearing that if it were possible, Google would do it.”

With this conviction and armed with the first convincing results regarding his technology, the serial entrepreneur is in Las Vegas at the prestigious Consumer Electronics Show to feel the pulse of the market and lay the groundwork for new partnerships.

The Montreal company’s mission is to create AI inspired by how the human brain works, rather than stuffing a machine with data using deep learning. His approach is called “deep meaning,” or the development of meaning.

“At the moment, the only way to improve AI is to make bigger and bigger models, but we’re getting to completely unmanageable dimensions,” says the entrepreneur. MIT estimates that just running ChatGPT uses the equivalent of the city of Granby’s electricity. »

“We thought it’s possible to create an AI that uses less energy and less data based on biology, cognitive psychology and neuroscience to develop a new approach,” he adds. There are opportunities to do better if you look elsewhere.”

An impressive start

In late 2022, Solid State of Mind tested its solution against a machine that, according to its boss, was at the top of what was being done in AI. Both machines were tasked with finding the ideal scenario for picking up balls in a virtual maze during this reinforcement learning exercise.

“We beat the second AI, which had a success rate of 75% after a million trials, while we achieved 90% in only 370 trials,” notes Maxime Julien. We used 10,000 times less data and 100,000 times less energy.”

The man who co-founded the start-up Sky Motion Research, sold in 2013 to AccuWeather, also praises the adaptive capabilities of his solution, the codes of which remain secret.

“During this test, we modified the task a bit,” he continues. The other machine had to repeat a million trials for the same success rate, while ours learned and only needed ten trials to achieve a 90% success rate. We don’t have to re-engineer our system like those who work in deep learning.”

He believes his technology works a bit like a person who has to pick tomatoes and then strawberries. Although the two fruits are different, people do not need much adjustment. However, an AI working in deep learning mode has to relearn everything if the harvest changes.

“We were kind of the black sheep in the AI ​​community because we weren’t doing deep learning,” he says. At first they laughed at us, but we are taking these results more seriously.”

Reserves until 2026

Solid State of Mind believes its AI would be useful in autonomous robotics, such as a robot dog or crop-picking robots. It would also integrate very well with autonomous vehicles and medical devices. There are also all cloud applications, especially for web recommendation engines, just like in the world of video games, to face much tougher enemies.

The boss hopes that his solution will be used in one of these areas within 14 months. Until then, the young company has enough money to develop until 2026. It plans to develop its technology and increase its value before seeking additional funds.

“In the United States, investors are less cautious,” believes Maxime Julien. We explain that the journey is long and arduous. We are now starting to build relationships because we need to get to know each other before we get married.”

Anyone who is in their sixth start-up knows that there is no point in rushing. He believes that the small size of his young company, which only has three employees, does not put it at a disadvantage when creating.

“It’s easier to go with the wind than against the wind,” he said. Especially in large companies. We try, we make mistakes. We don’t have to apologize to our boss. Great innovation can come from anywhere.”

And the next one in AI will possibly be from Quebec.

Leave a Comment

Your email address will not be published. Required fields are marked *