Long associated with a far-off sci-fi future, artificial technology has arrived in the present - and it’s here to stay.
Companies are already on their way toward an AI-driven model. In McKinsey’s latest survey on the state of AI, 56% of companies said they’ve adopted AI in at least one function.
For those still in the early stages of their AI journey, the stakes couldn’t be higher: An Accenture survey found that three of four executives believe that if they do not scale artificial intelligence in the next five years, they risk going out of business entirely. Why? Because, as 84% of those executives noted, AI is essential to achieving their organization’s growth objectives.
Turning an awareness of AI into its practical application is no simple task, however. In this guide, the third part of our series on AI and innovation, we look at how it can be accomplished. Inside, you’ll find insights on:
Artificial intelligence’s capabilities and potential for your organization
How to approach defining your AI goals
Assessing your internal capabilities - and whether you should enlist a partner
The role of data quality in AI development
How to get started with your first AI pilot program
Understanding AI’s Capabilities and Potential
For companies looking to harness the potential of AI, knowledge is power. To design and develop solutions that deliver for their businesses and their customers, leaders first need to understand, broadly speaking, what can be accomplished with AI.
In part one of our AI and innovation series, we explored how AI can impact all aspects of a business. For example, IT teams can leverage AIOps to detect emerging security threats and keep critical functions online. Supply chain management teams can utilize AI-powered scenario modeling to identify risks, optimize processes, and overcome inventory challenges. Marketing and sales teams can use AI to draw on previously untapped behavioral insights to improve customer experiences and conversions.
In part two of our series, we dug deeper into the concrete ways that AI can bolster strategies tied to efficiency, effectiveness, expertise, and innovation. Within organizations, AI adoption can:
Unlock collaboration and encourage interdisciplinary work
Improve decision-making across an organization
Motivate more agile, flexible, and innovative thinking
In turn, these AI investments can pay off for companies in a number of ways. According to a PwC survey, when asked to zero in on the top benefits of AI implementation:
48% said they expect their AI investments to grow revenue and increase profits
46% expect to create better customer experiences
40% expect to improve decision-making
39% expect to innovate products
38% expect to achieve cost savings
Practical AI: Define Business Goals and Prioritize Maximum Value
While our earlier deep dives into the technology clearly outline the range of benefits that companies can reap by successfully leveraging AI. That's the easiest thing to understand about AI. Ultimately, where a majority of companies struggle is in turning those ideas of AI into a practical reality.
Once they have an understanding of AI’s capabilities, companies need to begin their AI journey with a turn inward. This path toward an AI-powered reality all starts by asking a simple question: What are we trying to accomplish with the help of AI?
Of course, arriving at an answer to this question is not so simple. Doing so requires enlisting the input not just of key business leaders but of employees across a wide variety of departments and functions. Survey your organization to understand, among other factors, what outcomes they would like to see-which problems they would like to solve, which processes they would like to automate, and which obstacles currently stand in the way of success.
Prioritizing Value-for You and Your Customers
As you seek to understand your business’s needs and define areas where AI can help, consider which potential AI solutions might deliver the most value for your business. For instance, when soliciting input from across your organization, are there issues common to multiple teams or departments that can be solved by one solution?
Meanwhile, do not forget to consider the voice of the customer. In an online discussion hosted by MIT’s Sloan Management Review, Wim Rampen question the motivations of many organizations pursuing AI, saying:
Let’s be honest. Most organizations are focusing their AI efforts on things that matter only to them, hardly things that matter to their customers. Best to find the sweet spot where both interests meet.
As Rampen notes, uncovering where your business’s needs and your customers’ desires overlap is often the best way forward. The reason for this is quite straightforward: It’s a way of prioritizing and maximizing the value derived from initial AI endeavors. By choosing a use case that solves for the needs of both your business and customers, even small early successes can translate into outsized positive impact.
AI Readiness: Assessing Your Capabilities
With an understanding of what AI can do for your business and what objectives you wish to accomplish with AI, your organization must assess its operational and practical readiness for AI implementation.
Beyond cultural awareness and executive buy-in, companies need to take a hard look at their finances and ask: Do we have the necessary budget to support AI goals in both the near- and long-term? To understand the true costs, budgets must consider all inputs of AI investments, including contract resources, IT infrastructure, data acquisition, and potential partnerships or acquisitions. Crucially, they must also assess headcounts and the availability of the right talent.
AI and Talent
While the capabilities themselves capture the headlines, AI solutions are no different from other disruptive technologies: at the end of the day, they can only be successfully implemented with the right talent. Lacking the right, AI-savvy talent is what Jeanne Ross of MIT’s Center for Information Systems Research (CISR) calls “the fundamental flaw of AI implementation.”
Studying companies adding AI to their business processes, Ross and a team of CISR researchers found that-as we covered in parts one and two of this series-AI doesn’t replace human efforts, it augments them. Of course, in doing so, it also changes the demand for what is required of human workers. Since AI is best at eliminating those tedious, repetitive, and non-specialized tasks, the need for skilled and expert talent is increased.
How AI Changes Talent Requirements
Take this example offered by Ross: A company deploys an AI-powered customer service solution to resolve routine customer issues. The solution is a success-but without the right talent, that success only goes so far. Even with the AI solution in place, to truly drive sustainable improvement and change, the business still needs talent that can work to understand what’s causing customer issues.
The same holds true for, say, a team of financial analysts. AI might be able to much more efficiently crunch the numbers and extract performance data, but without the expertise to take action on the implications of that data, how much value is the team delivering for its customers?
Ultimately, possessing this expertise is what enables corporates in pursuit of AI to not just improve existing processes but to also imagine new ones. That’s why, as you look to introduce and implement AI, you must take an honest look at the talent in your organization.
From Talent Acquisition to Education
As the CISR team’s research shows, to deliver AI success, you don’t just need data scientists and analysts. You need employees who exhibit agile thinking and top-notch domain expertise. However, according to research, a talent-driven AI strategy can’t end there.
Surveying over 100 businesses in an array of sectors, McKinsey and MIT researchers sought to understand what set AI leaders apart from the rest of the pack. When it came to talent, the findings were clear-cut: Leading organizations didn’t solely leverage the expertise of specialists. AI leaders were five times more likely to train frontline employees on digitization and the Internet of Things. Likewise, more than half of leading companies trained their frontline personnel in the fundamentals of AI-compared with just 4% of other companies.
Looking Outside Your Organization
Even with these best-in-class capabilities, that doesn’t mean leading companies closed their doors to outside help. Researchers found that leaders in AI actually relied on external partners more than other companies. That’s because partnerships-whether with academic institutions, startups, consultants, or technology vendors-help to maximize learning and speed to market.
The lesson for companies looking to reap the benefits of AI? Begin with a transparent evaluation of your organization’s internal talent. Then, devise strategies to evangelize, educate, and develop that talent. Meanwhile, do not hesitate to look outside your organization to find the right partners that will help accelerate your journey to AI success.
Prepare Data, Data, and More Data
Talent plays a vital role in putting AI into practice. However, even with all the right people in place, there’s no escaping the centrality of data to AI applications.
The good news, of course, is that organizations and their customers are creating more data than ever before. Indeed, the International Data Corporation forecasted that, between 2020 and 2025, the amount of digital data created will be greater than twice the amount of data created since the advent of digital storage.
So why do organizations put this astonishing amount of data to use within their AI solutions?
Develop a Data-Driven Culture
To answer this, organizations first must go back to their people and their culture. Beginning with awareness and buy-in from senior leaders, corporates need to build a data-driven culture. This means communicating the benefits of working with data-and investing in the right tools and resources to ensure the workforce has data available and knows how to put it to use. Once again, this extends even to frontline staff. According to the MIT and McKinsey research cited above, every single leading AI company made data available to frontline staff. Underlining this point, further research suggests that companies at the forefront of AI adoption are more likely to possess robust, company-wide data governance systems and centralized data lakes that are widely accessible.
Improve Data Collection to Ensure Quantity
Meanwhile, all of these companies also acquire data from both customers and external suppliers to ensure an adequate supply for various applications. But how much data do you need exactly? According to AI consultant Adam Geitgey, “You want something on the order of 10,000 data points” for any AI application. Of course, the more complex your application is, the more data your AI will require for training.
Ensure Data Quality
While having a sufficient quantity is critical-and the more data you have, the better-not all data is created equally. As the IDC data above suggests, quantity is not often an issue for organizations. Instead, where most companies struggle is ensuring that their data is high quality. Ultimately, your AI project only stands a chance of succeeding if it's working with good data. In turn, your AI solutions will only be able to scale over time if they’re built on a foundation of high-quality data.
Image source: TimeXtender
That’s why certifying data quality should be an ongoing process for every organization adopting AI. According to AI experts at D-Labs, to be considered “clean,” teams need to consistently verify that data is:
Easy to understand and free from incoherent info
As accurate as possible
Possesses all the necessary attributes required for an algorithm to perform its task
To Get Started, Start Small
With the right goals, talent, and data, your organization will be truly ready to test the waters of AI. But deploying AI should nevertheless not be rushed. Begin with a pilot program that uses a selective, well-targeted sample of your data to train your algorithm, whether it’s one built in-house by your experts or one premade from an outside partner. (The upside to going the latter route is that it’s typically quicker to integrate.)
While the end goal of this pilot program is to eventually scale it into a company-wide practice, keep your sights trained on the near term to start. Zero in on a narrow problem, and closely monitor performance.
The metrics used to determine success will naturally vary from project to project. However, consider qualitative and quantitative factors such as:
Reports of improved customer satisfaction
Employee productivity and efficiency
Increases in sales and revenue, as well as decreases in costs
Establishing KPIs upfront that are aligned with the business goals you’ve defined above is crucial to developing an AI solution that’s successful, sustainable, and, finally, scalable. Tracking performance throughout this pilot equips your teams with the information they need to refine and improve solutions so that they can be rolled out to broader applications that drive results.
Along the way, do not forget the importance of maintaining an experimental mindset. Success in AI is often fueled, at first, by the learnings of initial failures. After all, when 16% of organizations report not taking solutions beyond the initial piloting stage, it’s clear that both patience and perseverance are key to tapping into AI’s awesome potential.
Conclusion
Over the course of our three-part series on AI and innovation, we’ve covered everything from the basics of AI to building a scalable solution.
Revisit part one of our series to brush up on the basics of AI and its applications.
Dig back into part two to explore practical strategies and methodologies that can power your own AI journey.
Altogether with the insights above, these resources can serve as your comprehensive AI playbook-and a source of end-to-end support as your organization moves from AI-ready to AI-powered.