“Only 9% of the success of an AI project lies in the machine learning algorithm, the remaining 91% lies in something else”, says Mari-Sanna Paukkeri, a true pioneer in the AI industry.

Paukkeri has 16 years of experience in Artificial Intelligence and a doctorate in machine learning & computational linguistics. With this background as well as being the CEO of a trailblazing AI company, Paukkeri has handled hundreds of successful AI projects.

Here is Paukkeri’s hows for a successful AI project:

1. Data quality. Without high quality data there is nothing. The higher quality the data, the higher chance the AI project will succeed.

2. Capable people. AI is rocket science, so it demands outstanding experts standing behind it. Without the intelligence of the people, an AI project will most likely fail.

3. Right tools and right algorithms. In one of Dr. Paukkeri’s previous AI projects, a same data set was tested on two different AI tools. While the traditional tool achieved only 51% in accuracy after 3 weeks, the modern one reached 84% in just an hour.

This post is an excerpt of Paukkeri’s speech at Digital Innovators’ Summit 2019 in Berlin, titled “How Alma Media, Tamedia and Schibsted employed AI to help solve problems across news comment moderation, shopping and advertising”.