Senior Partner at McKinsey & Company, Global co-leader of Data Transformation practice, leader of Greater New York Digital & Analytics Hub.
Great article by my colleagues (Aamer Baig, Douglas Merrill, Megha Sinha, Danesha Mead and Stephen Xu) lays out 7 things CIOs need to know or do when scaling AI: 1) Eliminate the noise, and focus on the signals (i.e., get past experiments by being focused on what can be learnt from POCs and go after solving a small number of big business problems) 2) It's about how the pieces fit together, not the pieces themselves (i.e., too much time is spent on choosing tools that are either obvious or undifferentiated vs integrating them) 3) Get a handle on costs before they sink you (i.e., platinum plating can be extremely expensive in this space with 10-20x variance of spend for similar outcomes, also change mgt vs development should be 3:1 but is often under-estimated) 4) Tame the proliferation of tools and tech (i.e., 'too many platforms' is a leading obstacle to progress) 5) Create teams that can build value, not just models (i.e., getting to impact is 'never just tech') 6) Go for the right data, not the perfect data (i.e., high performing GenAI solutions aren't possible without clean data, this requires real work and focus) 7) Reuse it or lose it (i.e., the name of the game in scaling is reuse - GenAI development can be sped up 30-50% through reuse of code) #NeverJustTech #GenAI #AI #data #McKinseyTechnology #QuantumBlack
Indeed, it's a compelling read by your colleagues outlining key factors for CIOs to consider when scaling AI. Prioritizing learning from POCs and focusing on big business problems sounds like a strategic approach Kayvaun Rowshankish