Four Areas of AI Readiness We Need to Improve
If we can't wait for existing orgs to be fully AI ready, how can we avoid the most obvious roadblocks on our AI adoption journey?
We all have high hopes for the impact of AI and smart automation on the enterprise, but without urgent action on organisational readiness, the biggest improvements and productivity gains will likely be out of reach.
As Dave Wright and Brian Solis wrote for CIO magazine, AI promises a shift from a focus on processes and tasks to a focus on outcomes, which could be transformative in the way organisations innovate and maximise productivity. But …
“IBM research found that 83% of executives say generative AI will reinvent the way their organization works. And organizations that revamp their operating models before putting new technologies at the core of their business notably outperform their peers. In short, modern operating models will be the enabler of business transformation.”
AI technology is advancing very rapidly, as Ethan Mollick describes in a recent post, but the gap between its capabilities and our organisations’ capacity to utilise them is becoming a chasm. Mollick suggests some useful starting questions for leaders, such as what useful thing do you do that is no longer valuable?, and what impossible thing can you now do thanks to AI? in order to help imagine the art of the possible. But there are bigger questions to ask around readiness, and perhaps the way we think about and describe AI capabilities need to be broadened to take into account their many dependencies and underpinnings.
There are at least four broad areas where readiness is lacking or non-existent.
People and adoption issues
As with any technology transformation, technical software is the easy bit; updating human software is where it gets much harder. Whether we call this behaviour change, adoption, or new ways of working, the fact is that so much management effort has been put into turning creative humans into process machines that it can be hard to break old work habits without a much more imaginative approach to engagement, learning, and re-thinking roles and tasks.
In HBR recently, Tomas Chamorro-Premuzic shared 7 strategies to get your employees on board with AI, which is a helpful primer on the engagement part of this challenge with some good ideas about how to overcome resistance. But we also need to think more deeply about the relationship between people, roles, tasks and value creation, and try to find more ways to use AI and automation as a force multiplier for human ingenuity and creativity.
Org structure barriers
“The problem companies are facing, however, is that traditional methods of process redesign may not be entirely up to the task because GenAI doesn’t function like a traditional technology. Users “talk” to it, much as they would to a human colleague, and it works with the user in an iterative fashion. It also can continuously improve as it learns user needs and behaviors (and vice versa).
To effectively integrate GenAI, we propose a new paradigm: Designing for Dialogue. Unlike traditional, technology-driven process redesign principles that focus on taking capabilities “out” of the human and putting them “into” the machine, Designing for Dialogue is rooted in the idea that technology and humans can share responsibilities dynamically…”
But in a way, this is among the easier organisational structural problems to solve. A bigger issue is the divided nature of the cascading hierarchy, which also guides the way that management functions within silos and limited domains of influence. Traditional management might work OK for the most mundane, low-value administrative work, but it is wholly inadequate for modern organisations, which should be thinking more in terms of embedding shared services in a platform that any team can use and build on, regardless of where they sit. I wrote about this aspect of the challenge a couple of weeks ago, and my hope that AI could be a more effective driver of organisational development than digital transformation has been.
Tech stack limitations
If we build on this idea of shared platforms rather than point solutions, it is clear that even individual AI apps and services should not be stand-alone actors within the new technology stack. Instead, they will combine layers of new technology that need to be managed and integrated to maximise our innovative potential. We don’t want many different data layers, Natural Language Processing, models, agents, etc. For a given organisation, these should probably be common services that we can build new bots and automations on top of. Even the bots themselves might be common, but use different modes - or be given different skills - to ensure some consistency of engagement and conversation, whilst supporting a variety of uses and use cases.
And of course, many organisations are still beholden to big platforms (finance, HCM, ERP, etc) each of which will try to force their own AI features into the organisation, rather than playing nicely with others.
Connected data and knowledge
Juliette Rolnick recently surveyed emerging enterprise AI use cases from a VC market perspective, and suggested four main areas adoption that she foresees:
Monotonous task reduction. Putting an end to high-touch, repetitive tasks that should be no or low-touch. (note that this bucket likely sees more PLG / employee-level adoption first)
Known output generation. Generating content or any deliverable that is pure execution on what is known. In other words, you already know how the slide deck or contract should look — you just need to create it. (seeing substantial activity in procurement and legal here)
Knowable but unknown output generation. Creating the idea from a dataset that the employee can know but is difficult to derive insights from manually. Beyond that, creating and delivering the content that supports this idea. (seeing substantial activity in advertising and ecomm enablement here)
Unknowable, unknown output generation. Creating the idea from a dataset that the employee can’t possibly know. Beyond that, creating and delivering the content that supports this idea. (seeing substantial activity in healthcare and pharma here)
GenAI can do a decent job of 1 and 2, but without connected data and knowledge, it is much harder to make progress in 3 and 4. Even in the first two areas of use cases, existing multi-stage process work might be so fragmented and full of exception handling that even process discovery tools might not be enough to identify where and how to apply automation.
Overall, using AI to create a smarter, more connected organisation will rely on having smarter, connected work and data platforms to build on.
Seeing the Bigger Picture
For any executive wanting to experiment with AI or adopt new AI tools or services in the enterprise, some of these wider issues of readiness cannot easily be solved, but that should not hold them back from getting started.
One tip that we would share from our experience of digital transformation initiatives would be to start by mapping the new business capability that you would like to create using AI, and rather than just focusing on technical requirements, try to identify and map its relationship to other processes, teams, skills, data, knowledge and systems to get an idea of the wider effort needed to make it succeed. And then, of course, have an adoption strategy and perhaps also a transition strategy that defines how you hope people will move from X to Y over time without dropping the ball or getting into trouble. These might seem like basic approaches, but just as surgeons and pilots still use checklists to avoid missing the obvious, it makes sense to focus on the bigger picture and context of new tech adoption, rather than becoming dazzled by the tech itself.