Getting artificial intelligence (AI) products to operate correctly takes a lot more effort than flipping a switch and letting a machine do the work.
In a July survey of 200 US and European IT executives conducted by Databricks and International Data Group, 56% of respondents said that preparing large datasets is a very challenging aspect of moving AI concepts into production. Similarly, 53% of respondents reported that deploying AI models quickly and reliably is also very challenging.
These results underscore the fact that AI products require significant investment.
News aggregation platform Flipboard uses AI to sift through about 300,000 articles per day. However, Flipboard has a team of nearly 40 tech, engineering and data science specialists who work on creating and monitoring its AI tools. And the company has 20 curators who manually select articles for its readers while supplementing its AI output.
To build its AI product, Nielsen Marketing Cloud relied on engineers to build the tech infrastructure, product developers to implement best practices and data scientists to monitor and tweak the product’s models and algorithms. Four to seven people worked full time on building the platform for a few years and several others got pulled in occasionally.
AI can be used for a plethora of tasks including buying ad inventory and writing pop songs. But getting its automation flowing requires an uplift.