The era of Small Data

01.12.2020

The idea of small data for artificial intelligence (AI) is one that resonates very well with me, our company and our clients. The definition of small data is widely recognized as a dataset that COULD be processed by humans. However, manual processing by human might be too costly, difficult, slow and error-prone. The first time I really got excited by the notion was in 2018 when I read this article: “Why small data is the future of AI”

I have lived through the Big Data hype and sadly saw it erode pretty much as quickly as it grew. The market was not yet on the right level of maturity, most organizations feared that they do not have enough data, or enough of right quality data, the tools were either very expensive or Open Source (Hadoop, MapReduce), that required a lot of competence scale-up.

Now, that AI solutions are quickly gaining interest, small data pops up as a major enabler for mature AI algorithms. Artificial Intelligence has been perceived as a solution that requires massive amount of data, powerful high-end processing units and as something that happens in a development lab.

That is no longer the case. AI can provide value even with small datasets. In fact, early AI algorithms were forced to work well with small amount of data and computing power. Recently, deep learning and other modern AI algorithms have started to rely more on clever learning than increasing amounts of data and computing power.

This is a major shift for organizations thinking about utilizing AI in creating business value. Small and Midsized businesses constitute 99% of all businesses in the EU “Statistics on small and medium-sized enterprises” and they certainly do not have the BIG data. However, with well-defined, efficient algorithms they can also reap the benefits of AI in their businesses. With AI becoming gradually more “mainstream”, the technology becomes more accessible also for smaller companies, not just the big, global behemoths. This naturally also means the price points and implementation times decrease.

Demand your AI solution provider for a lean and mean approach that fits your size. We at Ai4Value believe in the rise of small data and are well equipped to provide solutions for the smaller datasets and thus smaller organizations.

PS: I will be speaking to a Finnish group of SMB companies in Vaasa on November 27th: https://www.chydenius.fi/fi/kauppatieteet/opiskelu/science-meets-business-1

PS2: We have a couple of White Papers available, to be found at our website https://ai4value.com/

Share this post: