The upskilling path to AI success
Artificial Intelligence – the new magic wand!
You know AI is the absolute next biggest thing. You know it is going to change our world!! It is the little technology trick startups use to disrupt industries. It enables crazy applications we have never thought of before!! A few days ago, we were dazzled to learn of an AI app that promises to give one a credit rating score based on reading your face – essentially from just a photograph it can tell a prospective financier, what the likelihood of your paying back the loan is!!!!
Artificial Intelligence is real and has started becoming mainstream – chatbots using AI to answer queries are everywhere. AI is being used in stock trades, contact center applications, bank loans processing, crop harvests, self-driving vehicles and for streaming entertainment. AI is now part of boardroom discussions and strategic initiatives of CEOs.
McKinsey & Co. predicts AI will add $13T to the global economy in the next decade1. IDC is forecasting AI spend will double from $50 B in 2020 to $110B in 2024, making AI one of the largest and fastest growing components of new development spend.
So much to like but why are leaders shrugging their shoulders?
Despite all the good news above there is also another side to AI. For all the green indicators, there are also some red flags. In fact, if one googles “Hype vs reality” the results returned are to do with AI!!!!
Our experience shows that broad swaths of executives are skeptical of AI. Leaders in a variety of businesses from large multinational banks, consumer packaged goods companies to appliance makers have privately expressed their disappointment at not being able to make AI work for them. They cannot bridge the gap between the AI / ML hype and reality in their businesses.
The data available also bears this out – VentureBeat estimated that 87% of ML projects never make it into production3. HBR research shows only 8% of firms engage in the core practices that support widespread adoption of AI and advanced analytics4. Gartner predicts that 85% of AI projects will deliver erroneous outcomes through 20225!!!
However, we also know that AI, in some cases, is a resounding success. Iconic brands like Coca- Cola, Barbie, Volvo, BMW, Burberry, Walmart, Disney, Netflix are all using AI. So are industrial giants like GE Power and John Deere.
Visible patterns emerge from successful AI use-cases
This brings us to an interesting dichotomy – the reality of failed implementations versus the hype surrounding AI. Digital native / born digital companies or early adopters of AI form most of the success stories. Traditional companies find it tougher to embark on a successful AI journey. A recent study by ESI Thought labs6 has shown a staggering gap in the ROI of AI projects between early adopters versus others. There are numerous other studies which describe key characteristics of successful AI efforts. For example, McKinsey & Co. in their publication, “The State of AI in 2020”7 show that there is a large gulf between the high performers and the rest when using AI.
Certain common themes have emerged from these studies, many of which are now commonplace wisdom, if not trite. Leadership alignment around AI strategy is the most common one. Getting clean data, aligning strategy with execution, building the capabilities to use AI are all touted as critical requirements for successful execution. These themes all point to the insight that it is the human element which is critical, rather than technology.
As practitioners we have come across numerous examples of AI projects which go off-track because of human issues. In a case that one of us led, the remit was to enhance a call center’s capabilities and capacity using RPA tools. There was strong leadership support and enthusiasm. It was clear that a large number of basic level tickets raised by the center, could be resolved using digital agents. This would result in substantial gains in customer experience, through faster ticket resolution and higher employee productivity (> 30%). However, after two months of launching the pilot only a very small percentage of cases were identified for migration to digital agents.
Very soon, it was clear that these tools were being seen as a replacement for human labor rather than to augment their capabilities. The most vocal proponent of the initiative, the head of the customer experience team, became its critic, as he felt that the small savings were not worth the risk of higher turnover rates among agents due to perceived job insecurity.
This was turned around by leading a three-day workshop focused on demonstrating how the job responsibility of agents could be enhanced as portions of their job got automated. The processes were redesigned to isolate parts which could be fully automated and to club non-automated components together driving more responsibility and discretion to agents. Once enhanced responsibility of the call center staff was identified, managers felt more comfortable and were willing to support the initiative. In the end, the goals set at the start of the project were all met.
The Winning Formula
It is quite obvious that success is determined by human aspects rather than technological factors. We have integrated our observations from multiple engagements as practitioners and consultants, with a variety of research studies that have been published to distill the steps to AI nirvana. We have identified four key organizational actions that enable successful AI implementation at scale.
1. Establish a Data Culture
While this has been emphasized in some of the studies, the focus has been on ensuring that the firm has good, clean data sets and is using the data properly. Which are important. But our experience has shown that culture is more critical than having the data. Does the organization have a culture of using data to drive decisions? Does every level of the organization understand and use data insights to do their day-to-day jobs? Is decision making reasonably decentralized with different parts of the organization taking decisions driven by data, needing to escalate only when there is ambiguity or need for strategic clarity? Do business teams push for new data sources when they are not able to get the insights they need?
Without this kind of culture, it may be possible to implement individual pieces of automation in a specific area or process, applying brute force to see it through. In the call center example referred to earlier, there were clear issues of insufficient process documentation with multiple gaps, but we still achieved a good outcome – partly by fixing the data gaps in that process.
However, to transform one’s business and truly extract the power of AI, we advise clients to build a culture of Data Driven Decision Making first. We contend that if the data culture does not exist today, it would be better to focus on building that and driving a culture of decision making based on data insights before even drawing up an AI strategy. Only doing this will ready the organization to be capable of integrating the use of AI to achieve implementation at scale. In the short to medium term a move to a Data Driven Decision Making culture will deliver greater returns than trying to implement piecemeal AI projects.
2. Ingrain the Digital-First Mindset
Assuming a firm has passed the data culture hurdle it needs to consider whether it has adopted a digital-first mindset? Does it have a digital culture?
AI is one of many technologies that are impacting businesses, along with Augmented and Virtual Reality, the Internet of Things (IoT), 5G, Cloud technologies and Blockchain to name a few. Today’s environment requires firms to be able to utilize a variety of these technologies often together, and to have a workforce capable of using these digital tools.
A workforce with the digital-first mindset looks for a digital solution to problems wherever appropriate. They have a good understanding of digital technologies relevant to their space, are comfortable looking for digital solutions and understand key digital methodologies like Customer
360 to deliver a truly superior customer experience; or Agile methodologies to successfully manage AI at scale.
Companies which have the digital culture can succeed at any of these new technologies including AI, while for others scaling the benefits of new technology tools is really hard. AI needs business managers at the operational levels to work with IT or AI tech teams to pinpoint processes that are right for AI. They need to make an estimation based on historic data of what specific problems require and AI solution. This is enabled by the digital-first mindset. Upskilling the workforce in this mindset lays the foundation for successful AI implementation at scale.
3. Demystify AI
The next step is to get business leaders, functional leaders and business operational teams, not just those who work with AI, to acquire a basic understanding of AI . Along with upskilling business teams to inculcate the digital culture, one also needs to upskill the business teams to become adept at AI.
They do not need to learn the intricacies of programming or how to create neural networks or anything of that technical a nature. However, all levels from the leadership down should have a solid understanding of what AI can do, the basics of how it works, how the process of training data results in improved outcomes and so on. They need to understand the continuous learning nature of AI solutions, getting better over time. While AI tools may recommend an answer, human insight is often needed to make a correct decision off this recommendation.
Demystifying AI helps to clearly visualize the possibilities – driving an AI vision and strategy, define workable projects and understand how long they are likely to take. Business managers are able to manage such projects and working with the technology teams, deliver good outcomes.
Many leading organizations have already started down the path of such upskilling programs. For example, the Wall Street Journal reported that Bosch would be training over 20,000 staff including managers in AI8. There is an increased focus on democratizing AI services by leading AI providers. Leaders need to focus on upskilling so their organizations can use these services effectively.
4. Drive implementation Bottom-Up
A lot of the literature on AI focuses on the need for alignment, objectives, strategy and leadership. It talks about the human issues. But rarely have we seen a reference to what we consider the most important aspect of building scalable AI – which is letting projects run bottom up.
A reputed healthcare provider one of us worked with, embarked on a multi-year AI project to improve productivity. They wanted to use Natural Language Processing, Discovery, Cognitive Assist and ML to augment clinical proficiency of doctors and expected significant benefits in new drug discovery and trials by using the massive data sets available with them built over the previous 20 years.
The company ran this like any other transformation project, with a central program management team taking the lead with the help of an AI Center of Competency (COC). These two teams developed a compelling business case, identified initial pilots aligned with the long-term objectives of the program. However, after 18 months, they had very few tangible outcomes. Everyone
including doctors, research scientists, technicians and administrators, who participated in the program had their own interpretation of what AI was not able to do.
Discussion revealed that the doctors and researchers felt that they were training AI to replace themselves. Seeing a tool trying to mimic the same access and understanding of numerous documents baffled them at best. They were not ready to work with AI programs step by step to help AI tools learn and discover new insights.
At this point, we suggested approaching the project bottom-up – wherein the participating teams would decide specific projects to take up. This developed a culture where teams collaborated as well as competed, with each other, in finding new ways to use AI. Employees were shown a roadmap of how their jobs would be enhanced by offloading routine decisions to AI. They were shown that AI tools augment the employee’s cognitive capabilities, that embracing AI made them more effective. In short AI was “demystified”.
The team working on critical trials found these tools extremely useful and were able to collaborate with other organizations specializing in similar trials. They created the metadata and used ML algorithms to discover new insight. Working bottom up led to a very successful AI deployment.
We have seen time and again that while leadership may set the strategy and objectives, it is best to let the teams work bottom-up to come up with the projects to implement. To make this bottom-up approach work once again upskilling is required and some changes are needed in the ways of working. Agile methodologies for example, are critical to driving projects bottom-up.
Conclusion : Upskilling is the common glue
While successful AI at scale, requires much to be done, many bridges to be crossed, the four actions describer earlier are critical to building the AI powered, future-proof, enterprise. These four “keys” are all human related and each require specific kinds of upskilling as shown in the figure below. The figure shows how upskilling needs vary by organizational level and the key being addressed. However, it drives home our key point that upskilling is a universal requirement for driving AI at scale, successfully.
Authored by :-
Niloy Mukherjee, Ravi Pattamatta and Ratnesh Prasad