October 6, 2021 2 min read
Above: Andrew Blumenfeld, co-founder of Telepath, and Lily Adelstein, creative project manager, discuss ML for small and mid size businesses.
In a recent article for Fast Company, Geoff Webb talks about the transition of machine learning and artificial intelligence from the exclusive province of large enterprises, into the everyday work of small and mid sized businesses. Webb rightly notes the significant change that machine learning has enabled both in business operations (i.e., offering recommendations and insights about pricing) to customer-facing functions (i.e., chatbots and product recommendations). “Small businesses are using AI to solve a slew of problems that were, until recently, simply too complex to manage, too expensive to address, or required highly skilled and scarce data-scientist expertise,” Webb writes.
But why? What accounts for the greater access to this tech by non-giant companies and governments? And where can we expect this trend to go next?
Research. We are living through an exciting time for machine learning and data science research, with academia and major tech companies like Facebook and Google releasing breakthrough papers that are moving this technology forward for everyone. This generation of new knowledge is happening at breakneck speed and, in many cases, is bringing down the cost of developing new artificial intelligence technologies.
Data. One of the great burdens on past progress in the machine learning space was the quality and availability of data. Companies with lots of historic data also had collected that data in a way that did not have machine learning use cases in mind. Consequently, enormous amounts of time, money, and effort had to be spent cleaning and preparing the data so that it could be used to teach machine learning models. Newer companies of all sizes have often been much more thoughtful about what data they have collected- and how- right from the very beginning. Many companies are now “data-first” in their thinking, making them prepared to tackle machine learning projects much more quickly. Add to that the sheer availability of new types of data in standardized format, and you create an environment where artificial intelligence can really accelerate.
Tools and Best Practices. The last several years has also seen the explosion of investment in new tools specifically aimed at helping data scientists and the companies they work for convert data science into practical machine learning applications. Along with these tools a canon of best practices for applied machine learning is developing which is also helping to reduce the cost and improve the speed with which machine learning knowledge is being translated from theoretical research into real life products and systems.
Taken together, these are just some of the trends that are helping to empower more companies to build products and features that are powered by artificial intelligence, which is making that technology more widely available. Rather than simply having major enterprise companies with massive R&D budgets toiling away on highly customized machine learning models, we are seeing more and more SaaS B2B and B2C companies developing technology that has machine learning baked into the offering. That said, there is reason to believe that we are still very much at the beginning of a massive wave, and that progress in all of these dimensions and others will likely continue to explode the amount of available machine learning-powered technology in our world.