Everything you need to know about Artificial Intelligence (AI)
What is it?
AI stands for Artificial Intelligence. Merriam Webster defines it as “a branch of computer science dealing with the simulation of intelligent behaviour in computers.” It is also defined as “the capability of a machine to imitate intelligent human behaviour.” Artificial Intelligence systems will typically show at least some of these human behaviours: learning, planning, reasoning, problem solving, perception, manipulation, motion and perhaps even social intelligence and creativity. While in sci-fi we see AI being human-like robots, it can actually be almost anything. It can range from Google’s search algorithms, to IBM’s Watson, to self-driving cars or even to SIRI.
Narrow AI versus General AI
We see Narrow Artificial Intelligence around us on the day-to-day. It’s in computers that have been taught how to do something without being explicitly programmed on how to do that. An example is a recommendation engine (like the one Amazon uses) to recommend products based on previous purchases. The long term goal of AI research is a Artificial General Intelligence (AGI or strong AI). AGI would be able to outperform humans at nearly every cognitive task. Narrow AI may outperform humans at its specific task, like playing chess or trying to solve equations.
Fera and AI
At Fera we’re leveraging Artificial Intelligence system to better eCommerce. We are using skills from our app on your store. Our app analyzes the data from your visitors to measure the performance of these skills. It can then determine which skills are providing the most benefit for your store. By learning, it applies and experiments with different skills until it has found the combination that’s most beneficial to your site. It will learn and continually improve to offer you the custom solution that drives the most conversions for your site.
We want to move from weak AI to strong AI. It also needs to feel like we’re interacting with a human, and not a computer. You’ve probably noticed this a lot with telemarketer pre-recorded calls or even customer service live chats. Sometimes it can be hard to determine if it’s a real person or not. As more data is collected and readily available we’ll be able to teach our machines more and more.
We do, however, need to be aware of biased data. A machine can’t determine bias and if we feed it biased data it will learn and continue the patterns of bias, so unwatched and unchecked it can be very dangerous. For this reason there is a lot of talk about AI being open source, to constantly check the data, add data, and to have accountability for it and its process.