Eight takeaways on data and transformation for CIOs

Here are the key takeaways for CIOs to consider:

1. Think in terms of impact and feasibility. Prioritize data and AI projects with these variables in mind. High impact and easy-to-implement projects can drive organizational momentum forward. High impact and low feasibility are often the most interesting, as these long-term plays force you to consider the skills and infrastructure that evolve into a foundation.
   
   
2. Adopt a mindset of privacy by design. Governance is too often considered a bolt-on to AI efforts instead of embedded from the beginning. Regulatory considerations like the “right to be forgotten” in the European Union’s General Data Protection Regulation are important to build into AI efforts from the start.
   
   
3. Avoid continually starting from zero. Development is slow and painful because data scientists typically must slog through ad hoc processes straying from their value-added skills, like building data pipelines. AI platforms are accelerators that democratize data science for more workers.
   
   
4. Iterate rapidly with thorough experimentation. Begin with a hypothesis-driven mindset that advances as your tools evolve — into more sophisticated A/B testing that can be tweaked in real time, for example. Experimentation platforms help model out the ROI on big initiatives.
   
   
5. Build trust. One Tech Icon participant asked her employees in a call center about the task they hated the most, then made the case for automating it, gaining confidence from those concerned about layoffs. Another added process descriptions to tools showing how data is turned into insights.
   
   
6. Hold lines of business accountable for data integrity. Data needs to be controlled at the point of origination, so an updated operating model is integral. One participant helped to establish data domains and to assign owners in her organization; those who identified problems could file a claim, and she would know whom to contact. Accountability will prove to be vital as an organization goes deeper into AI and data-driven models.
   
   
7. Understand your vendors. Unintended consequences from AI can be significant, like improper data tagging, can be significant and affect your entire value-chain. It’s critical to first understand the mechanisms by which an AI product or service operates before implementing it.
   
   
8. Make sure your data initiatives have sponsorship at the top levels. Do you have an enterprise mandate? It’s a question of trust and credibility because data ownership can’t be centralized anymore.