We’ll begin this final Digital Literacy dimension with a personal story from Andrew (one of this paper’s co-authors, for those readers who didn’t skip straight to the biographies at the end).

In late 2023, my wife, Ana (another of the paper’s co-authors), and I found ourselves sitting at a café in Melbourne, Australia, with most of the day left before we needed to catch a flight. My first instinct was to spend thirty minutes scouring the internet for things to do or see in Melbourne, an idea that I did not relish because I dislike being glued to my phone when in good company.

Ana suggested that they seek advice from Microsoft Copilot (which was, at the time, branded simply as “Bing”). I had not thought of this, but curiously explained the situation to Bing and clarified where we were, how much time we had before we needed to go to the airport, and what kinds of sights we like to see when visiting a city. Much to my delight, in about fifteen seconds Bing suggested an entire itinerary for the day including sights to see and places to eat and drink. The itinerary was even organized according to a logical walking path from the place in the city where we were then sitting.

This capability had been in my pocket for months, but so ingrained was the impulse towards self directed Googling that it had never occurred to me that AI-infused Bing could do the work for me so much faster. Off we went to explore Melbourne!

It is not enough that we put AI into our colleagues’ hands (or pockets), lest it stay there until some outside force compels them to give it a try.

To understand this predicament, let’s consider a 2023 study published by Boston Consulting Group (BCG) finding “that people mistrust generative AI in areas where it can contribute tremendous value and trust it too much where the technology isn’t competent.”

BCG goes on to enumerate the study’s key takeaways:

• “Around 90% of participants improved their performance when using GenAI for creative ideation. People did best when they did not attempt to edit GPT-4’s output;

• “When working on business problem solving, a task outside the tool’s current competence, many participants took GPT-4's misleading output at face value. Their performance was 23% worse than those who didn’t use the tool at all;

• “Adopting generative AI is a massive change management effort. The job of the leader is to help people use the new technology in the right way, for the right tasks, and to continually adjust and adapt in the face of GenAI’s ever-expanding frontier.”

The change needed in most of your non-technical colleagues can be thusly understood as:

• Knowledge: Colleagues must know that a particular AI capability exists, what it does, where to find it (hopefully embedded in workstreams with which they are already comfortable), and - in some cases - be persuaded as to its merits over doing things “the old-fashioned way;”

• Understanding: Comfort breeds acceptance, so it is important that you help colleagues understand how one interacts with AI generally and a given workload specifically, which should include a healthy awareness of how to talk with AI, how to write an effective prompt, and an understanding that AI workloads thrive on better information (so, explain where you are in Melbourne, and what kinds of things you’d like to see);

• Skepticism: In particular, colleagues should have some grounding in how to be an ethical user of AI, recognition of possible hallucination, incorrect responses, or bias in training data, and an appreciation for the notion that AI workloads sometimes get things wrong, too.

Remember, also, that the leadership teams in most organizations have themselves not been immersed in concepts such as the latest technical skills, basic knowledge of data concepts, data literacy, responsible AI, how to use and apply AI in general, etc. Your digital literacy efforts ought to therefor include elements designed explicitly for senior and executive leaders such that they can develop the knowledge required to be the best possible organizational leaders in the age of AI.

Developers, engineers, IT colleagues, and others involved with creating AI capability ought to go several steps further.

Workloads and their user experiences should be designed to be increasingly born in AI rather than a traditional app with AI bolted on. As discussed earlier, this is a transition that will both take a bit of time to play out yet is likely to happen in earnest.

Microsoft’s Project Sophia is already pushing boundaries that are likely to burst wide open as more and more architects and developers experiment, refine, and commercialize their born-in AI solutions.

Whilst this first transition to born-in-AI workloads takes shape, the developer-equivalent to my Bing-powered exploration of Melbourne is already underway. For just as developers will learn how to build workloads that harness AI for end users, they, too, will continue to learn how to fully harness AI to help them build the workloads itself. It is difficult to land on a believable statistic here, but immense volumes of new code are now being written by AI in the form of tools such as GitHub Copilot.

Meanwhile, Copilot for Power Apps and other Power Platform services are now creating significant pieces of the workloads themselves based on the Copilot’s chatting with developers and citizen developers alike.

The human-centric change required of IT professionals will thus be two-fold:

• Learn to build workloads and user experiences that are born in AI; and

• Use AI to build the workloads and user experiences themselves.