Machine Learning (ML) has become more commonplace in enterprises as the number of areas where it is effective grows. “If your organization or vertical has not yet embraced ML, it is time to get on the adoption curve. There is no doubt that if you aren’t going for it, it is coming for you,” says Adam Pantanowitz, co-founder of AURA, a South African emergency response platform for the security and medical services industry.
Pantanowitz says when it comes to implementing ML systems, it is important for enterprises to have the appropriate skill sets available whether in-house or via consultants as there are huge consequences of implementing ML incorrectly. Perhaps equally as important is ample buy-in when deploying Machine Learning systems. “There are numerous technical complexities and challenges involved but these are often quickly solved by clever engineers. What is harder is that there is always resistance to change in an enterprise and I would say the very hardest thing is to overcome the challenges around how people receive technologies. There is an immune system in organizations which work to stifle disruption: this is a reality of organizations. Change should be slow, incremental, and bring people along.”
He also notes that quick wins and proof-of-concepts, without upsetting the business and disrupting it too much, are important to gain momentum and win over important colleagues. “In just about every vertical, you can add machine learning, and improve it. There are no doubt areas which are better suited to early adoption — those where there is more data, and data is accessible, and where decisions are made from data, are very well poised for implementation. But there is no industry that is immune to these phenomenally impactful technologies. It’s just about where a given vertical is on the adoption curve right now.”
The greater accessibility and democratization of ML has led to a groundswell of excitement around the technology, from big cloud players to the open source community, as the potential of ML to revolutionize and improve every industry becomes all the more apparent. Says Pantanowitz, “We’re onto an exciting frontier. In the medium term, I would say an interesting area is that of human-machine collaboration where machines greatly augment our abilities in the working world. These collaborative systems may also help ease adoption, as the organizational immunity to change is really the ultimate and most substantial barrier we organizations face, and it is more so of large enterprises.”
He adds that ML is uniquely exciting because of its self-generative nature. “As systems are implemented, capability improves, data improves, and more systems can be built, resulting in greater competence, adoption and more data. This bootstrapping system really does apply in terms of the journey an organization goes on, and so the top and bottom lines are influenced.”