How are CIOs approaching AI at work?

AI is an evolution of your big data strategy. The term AI assumes actual intelligence which isn't what companies can actually purchase today. What most people are referring to as AI is actually Machine Learning or Deep Learning techniques, where Deep Learning is mostly used by Google, Facebook, and the likes.

Successful approaches using ML and DL include defining the problem, defining success, and working backward from there into creating the right data set to feed in. The data integrity problem hasn't gone away with AI. You still need to clean data or it's garbage in, garbage out. Often what seems like a 'vendor problem' in AI is not. What doesn't work is throwing data into a black box and expecting meaningful insights to come back.

0 answers

#ai,#strategy,#innovation,#artificial intelligence @IT
Maribel Lopez

Maribel Lopez, Founder and Principal Analyst / Ex-VP

AI is an evolution of your big data strategy. The term AI assumes actual intelligence which isn't what companies can actually purchase today. What most people are referring to as AI is actually Machine Learning or Deep Learning techniques, where Deep Learning is mostly used by Google, Facebook, and the likes.

Successful approaches using ML and DL include defining the problem, defining success, and working backward from there into creating the right data set to feed in. The data integrity problem hasn't gone away with AI. You still need to clean data or it's garbage in, garbage out. Often what seems like a 'vendor problem' in AI is not. What doesn't work is throwing data into a black box and expecting meaningful insights to come back.

Arvind KC

Arvind KC, CIO

Few suggestions

  1. Learn the technology: AI/ML is here to stay and CIOs should invest time to gain hands on understanding of how these can be applied. Coursera, Udemy have several good courses to build a strong foundation. For the more adventurous, Fast.ai and Ian Goodfellow's deep learning book are excellent ways to understand core concepts. As this is a trend that is here to stay, the longer you delay learning the fundamentals in a first order way, the more out of touch you would be with the art of possible.
  2. Form a small team dedicated to this and have it report to you. Let them experiment for a while with ideas, the learning phase for the team and for you is really important.
  3. Find right business partners who are willing to experiment with ideas and embed data scientists in those teams. In addition to finding valuable problems to solve, this has the benefit of increasing the data maturity of business functions.
  4. Find a valuable problem in IT for which ML can be applied. For instance, all service functions need a good way to determine quality of service. The old school implementation of this sends as survey when a ticket is closed. Instead logistic regression is a nifty way to approach this.
  5. Look for areas of 10X productivity gains. Does your legal team spend a lot of time in reviewing contracts? Do you have a fragmented knowledge base that is hard to get insights from? Does your recruiting team sift though resumes endlessly? All these manual tasks have a high probability of being eliminated by current AI/ML techniques
  6. Find startups that have deep expertise in AI (see if founders write papers and have patents) who are also solving a problem that you care about and partner deeply with them. Osmosis helps both sides, it elevates the "ML thinking" of your org and helps the startup refine their product.
  7. Increase the data bar for your org and over time for other business functions. You can do this by training people on using R, Jupyter and also on core concepts of stats, linear algebra and probability. People are creative and when they come across a new repertoire of tools, they wield it in fascinating ways.
  8. Focus on IA also. Much has been said about artificial intelligence but equally powerful is Intelligence Augmentation (IA). This is the idea that smart humans with deep analytics can be really effective. Pay attention to how humans in the org are using analytical tools to make decisions and finds ways to enhance both humans and machines.

Dr Louis Shallal, Ph.D

Dr Louis Shallal, Ph.D, CIO GOVERNMENT OF JAMAICA

Seasoned CIOs approach AI with keen interest but also with some trepidation! There are many technical, social, economic and even moral issues to consider. We all know human ingenuity will push the limits of AI, but we need measured response to it considering what it means not only for own organization but society as a whole.

Seenuvasan Amaranathan

Seenuvasan Amaranathan, Sr Director - IT, Cloud, Security

We (CIO)'s approaching AI to provide more analytical information, wherein the past the service analytical information was not available to us in a ready format when we need. Our data generation across the business is huge and we need AI to help us to resolve some of the complex transitional issues including the business requirements vs IT investments. Which is always a challenge for any CIO to justify the IT investment to the management and business. We expect AI to act as a catalyst to bring transformation in IT & Business.