Time has a higher price in the Enterprise AI/ML Race

While enterprises typically underinvest in most transformation areas, the involvement of introducing new AI/ML models are more likely to be overestimated. This might be because ideas such as “Cloud” are comparable to regular infrastructure, whereas AI/ML is only comparable humanoid robots existing only in Scifi world.

The reality is that AI/ML models are much simpler. Many simple deep learning processes and basic statistical models come pretty well off the shelf to get started, and can offer immediate value in months rather than years. The quicker firms that understand this and get a head start on this new world will have a significant advantage in years to come.

The speed of enablement is even more critical in AI/ML due to the invalidation of the old paradigm of linear advancement based on human effort. Now the machines are in charge. The human limitations like the “Mythical Man Month” and management overheads are no longer in full effect. Therefore there is a steerer exponential growth rate compared with other areas.

This becomes a vicious cycle of acceleration:

  1. Models are built on top of the power of previous models.
  2. Products can benefit from increased digital interaction capturing more data points.
  3. The larger data sets are used made available for model training.

Additionally, with effective product management new propositions can more effectively be be generated by capitalising on a captive audience by generating new data points as new technologies become available. (e.g. via IoT, wearables, increased mobility).

I would get started with AI/ML introduction at the same time as any new data platforms are introduced:

  1. Invest in strong leadership comfortable jumping into AI/ML. This includes an empowered “triad of tech strategy”: Chief Data Officer (CDO), Chief Innovation Officer (CINO), Chief Product Office (CPO).
  2. Unite the Data Science team and Business team to a cross-functional working squad. Aligning these groups to undergo a unified evolution is most effective.
  3. Prioritise embedding discoveries into processes, business models, interaction strategies over finding new insights. The insight “definition of done” consists of delivering value to the right decision-makers.

Enterprises are not adopting the complex modelling practices of Google, Facebook and Amazon, nor do they have the data sets for this analysis. There are ready-made models to get started with today that will help get started in competing in a challenging digital marketplace that is set to get even more challenging over the coming years.