Artificial Intelligence and Neural Networks

On May 2005, Davenport Cohen and Jacobson published a working paper that a few months later appeared in HBR: Competing on analytics, one year later the book was released. This research was both chronicle and trigger of what has been called the analytics revolution or the data revolution. New positions have been invented such as data scientist and data engineers and a whole new cycle of hype around machine learning in particular and AI in general started. Analytics is presented now as the modern source of competitive advantage, is that so? Models allowed us to understand the world through the use of stylization and abstraction, that way we were able to precisely delineate ideas and complex systems that were later implemented in the real world. The Holy Grail has always been to be able to translate these models into fully automated systems with little need of human intervention and plastic enough to allow experimentation and rapid change. Advances in Cloud, AI and IT are making this dream real. Platforms such as Amazon, Uber, Facebook, Instagram or WhatsApp are examples of this, fully automated models. The implication are numerous, practically infinite scalability, zero marginal cost and no decreasing returns on scale together with the extensive use of network effects and feedback loops among others. Competition moved first from execution to analytics and now to model implementation. With it, the competences needed in companies to successfully compete in this new environment together with the meaning of R&D and experimentation changed dramatically. Innovation has always been a key driver of progress and growth, but now in environments with total plasticity and perfect execution, innovation is more relevant than ever. This new world is however full of challenges, code glitches, bandwagon effects, strange emergent behaviors are some of the unintended consequences of systems that act and decide on their own with non-human logics.