How Sentient's using the principles of evolution to create AI solutions

Sentient Cover
See all stories

How Sentient's using the principles of evolution to create AI solutions

When we think about cutting-edge technologies, we don’t usually associate them with the ruthlessness and chaos of the natural world. But, as it turns out, a lot can be learned from nature’s engineering.

Founded in 2007 in San Francisco by Babak Hodjat, Antoine Blondeau, and Adam Cheyer, Sentient Technologies is an Artificial Intelligence (AI) platform that uses evolutionary algorithms to tackle the world's most complex problems. In a process similar to natural selection, their AI algorithms learn, adapt, and react to produce the optimum solution to any given problem.

You might recognize cofounders Babak and Antoine as the brains behind the natural language interface that went on to power Apple’s Siri. After selling the technology to Sybase in 2004, Sentient was the team's endeavor to build something truly scalable.

Since then, Sentient has gone on to receive a total of $174.4M in equity funding. It has used its patented evolutionary AI in a range of industries from digital marketing, to machine learning, and even stock trading. The company has also created Sentient labs to research AI techniques in sectors from healthcare to agriculture.

Sentient Technologies
From left to right: Sentient's Solutions Engineer, Romain Gabriel, Business Development Manager, Manon Paulet, and Director of Sales, Michael Versluis

Using AI to turn window shoppers into buyers

Sentient Ascend is based out of TQ, working to bring their AI-powered conversion rate optimization platform to more clients in EMEA. As Michael Versluis, Director of Sales explained over a coffee: “We want to create AI that can optimize problem-solving and eventually make the world a better place.”

The conversion rate is a key metric for e-commerce websites, measuring the percentage of visitors who actually convert to customers. One way companies can improve their conversion rate is by optimizing their user interface. This is traditionally done by A/B testing: A slow process that tests two versions of a web page to see which one performs better

Sentient Ascend, on the other hand, uses their algorithm to make this process much faster and more efficient. “We don’t just test A against B, we test the entire alphabet of solutions plus more in a much quicker time than traditional A/B testing,” explained Michael. 

Why outsource problem-solving to an evolutionary algorithm?

In order to understand what evolutionary AI is, you need a refresher in biology. In simple terms, evolution is the process of species adapting to changing conditions. This happens through natural selection: The fittest and best-adapted are the ones that survive and produce even fitter offspring.

What does this have to do with AI? Well, these biological mechanisms can also be applied to problem-solving. Sentient starts with the challenge they want to tackle and produces a random population of solutions. These are later ranked and the best ones are combined to create a new generation of even better solutions. The process then continues until the algorithm arrives at the optimal answer to the problem.

There are several benefits of optimizing problem-solving with AI. Not only is AI much faster than a human brain, but it’s also free of our inherent preconceptions. 

Sentient and MIT Media Lab's cyber agriculture research project testing the optimal growth environment for basil.

Michael recalled a cyber agriculture experiment Sentient collaborated on with MIT. Despite their initial assumption that the typical day-night cycle is the optimal growth environment for plants, their algorithm suggested exposing basil to light for 24-hours was more efficient.

As he explained: “We, as humans, are biased towards concepts, but AI isn’t. It has no emotions, no feelings. It’s just programmed to find the best solution.” 

Teaching people to embrace AI's potential 

Despite the promises of faster and better conversion rate optimization, convincing businesses to try evolutionary AI is not always easy. According to Michael, this is because people are scared of what they don’t know and don’t understand.

But AI is already here and we can hardly avoid it. Spotify and Netflix recommendations are powered by AI, and so are the suggested replies you see on your Gmail account. What some users might see as creepy, Michael sees as potential: “As a salesperson, I like that optimizing my work with AI leaves me more time to focus on the really creative and exciting part of my job. I’m trying to get more people to embrace that, too.”

Personally, I find the concept of evolutionary AI rather reassuring. With technology moving forward so fast, it’s comforting to know that nature’s design can still teach us something.

Enjoy the read? Sign up to our weekly newsletter for a curated selection of meaningful stories delivered straight to your inbox. And don't worry... we hate spam too.