Earlier this year, Andrew Ng, co-founder and former head of Google Brain and former chief scientist at Baidu, presented the AI community with a daring yet simple question on the current state of AI development: Why has the industry been so focused on data model performance to the detriment of data quality improvement?
The foundation of his argument was a simple one: AI systems consist of code and data, and data is like food for AI. Just like a knowledgeable and technically skilled chef will not realize a recipe’s full potential without the right ingredients, an AI system will fall flat without proper datasets. In fact, since the late 2000s, widely shared estimates have judged that 80% of a data scientist’s work involves collecting, cleaning and organizing data, while only 20% goes toward the rest, and working with the algorithm accounts for roughly 5% of the work.
CentralNic Group Plc (LON:CNIC) provides registry services, distribution, and strategic consultancy for new TLDs, ccTLDs and SLDs.
