Recent advances in Artificial Intelligence, enabled by cloud technology and low-cost compute presents the most compelling opportunity in tech. The consensus seems about AI is that it will add many trillions of dollars of value over the next decade. The value and the promise of AI is starting to be seen in many sectors with much of the early use coming from established large IT companies with deep expertise. The press talks about the sunny side or the scary side of AI – but there is very little being discussed around practical aspects such as technical/business risks, ideal team composition, life-cycle, evolution, typical design patterns to use, business case and modeling expected value, calculating total cost of ownership, planning for support/operations, client expectation management, and marketing/sales approaches.

The life-cycle of AI project is different to a traditional software project, the skill combination is different, project managers need to address a range of different risk vectors, software architects need to consider a different set of technical concerns. The popular agile development approaches work only partially, estimation of effort needs a different perspective and these projects should construct the business case with a proper appreciation of risks. Furthermore, (at least for now), a strong relationship with research focused institutions is helpful in improving the odds of success as the field is evolving fast and as such there are no firm patterns to follow (unlike a few decades of traditional database/workflow/control driven systems).

This talk synthesizes experience gained over the past 4 years from creating different commercial AI projects across a range of sectors. The presentation will cover key concepts that technical and business leaders should be aware of. Using a set of case studies specific concerns will be discussed with an open format that invites questions and conversation.

Agenda

"Lessons learnt from creating AI systems in the last few years with a focus on Risk"