AI and ML challenges within the enterprise
Last night I attended London AI talk with SK Reddy, Josh Warwick & Cheuk Ting Ho. I was primarily interested in the talk by SK Reddy addressing what challenges companies are facing with regards to AI and Machine Learning. These challenges are quite varied, and there was not enough time to go into detail so in this post I wanted to focus on the two that resonated with me the most:
1 - Shortage of data, lack of data, data distributed in silos with poor data governance and protection, dirty or untrustworthy data
2 - Shortage of skills, how do we get talent?
Needless to say, the quality of data determines its value as a corporate asset, however, dirty data accounts for 51% of all data out there. It sounds logical that the best way of collecting clean data is to do it at source. This requires a change in the way departments like Marketing, Sales, etc create and collect data, both at organizational and process level. How these two tasks are accomplished vary enormously from company to company, however, there are a few points all of them must touch, to name a few:
1 - Develop a strategy across the whole of the enterprise
2 - Commitment from the leadership teams
3 - Strategy to be aligned with business objectives
4 - Formal team and processes devoted to data analytics working alongside non-data centric teams
How about lack of skills in the market? It will take time for the next wave of Data Scientist to achieve maturity, however, Google is trying to help those companies with fewer resources. Google announced several weeks ago, Cloud AutoML a service part of their MLaaS offering. They are targeting the millions of software engineers out there to enable them to do ML. At present, only image recognition is available with the other major branches of AI to be added in the future. As an additional plus, the model does not require a huge dataset for training, it has been generically trained, still, you need to have a clean and labeled dataset.
Conclusion
Data Science, AI, and Machine Learning are areas in which your company should be getting into, however, there are quite a few barriers companies need to jump to adopt them. The offering out there is quite profound from DYI all the way to consultancy. There is something for everyone.