Speaking on day two of the AI & Big Data Congress, David Malkin, Cogent Labs’ product management director, argued that businesses need to include people in the data validation process for machine learning and artificial intelligence models through “an interaction interface between humans and the AI system which enables them to validate data and easily make corrections if they spot any errors.”
He recommended that companies looking to market AI-based services or applications should have “a system in place to measure the accuracy of data as some errors are more significant than others.” They additionally ought to factor in the value perceived by users to assess “whether customers see it as a major or minor error in the business logic.”
Identifying the relationship between accuracy and the value they deliver with data is paramount for firms to settle on what accuracy they require before they can create business value and be successful. However, Malkin cautioned that “other constraints such as size, speed and memory also have to be borne in mind when developing new models.”
Day two of the congress was opened by Marco Orellana, manager of the Centre of Innovation for Data tech and Artificial Intelligence (CIDAI). He noted that “this year’s AI & Big Data Congress has confirmed that data technologies and artificial intelligence are here to help people’s lives, to collaborate with users and not to replace their jobs.” At a technical level, “we have seen how getting AI projects into production is even more important than developing AI algorithms with a high degree of precision.”
Joan Mas, CIDAI director and also director of the Digital Unit at Eurecat, pointed out that “some examples presented at the congress show that although AI-based models are crucial, any enterprise seeking to exploit them commercially also needs to address key aspects such as user interaction and scalability.”
