Mastering data management and assembling a diverse team are two keys to success.
Many companies are stuck in dead-end pilots while an elite few have figured out how to make artificial intelligence (AI) work at scale, according to a new Accenture report.
"AI: Built to Scale" shows how difficult this transformation is as well as what it takes to do it successfully.
"In a nutshell, what our report found is that the majority of companies are really struggling to scale AI," said Bob Berkey, MD, Accenture Applied Intelligence. "They're stuck in the Proof of Concept Factory, conducting AI experiments and pilots but achieving a low scaling success rate and a low return on their AI investments."
Accenture surveyed 1,500 C-level executives across 16 industries to determine what makes AI projects successful. Accenture identified three phases in the AI evolution:
- Proof of Concept Factory - siloed work, no CEO attention, no scale
- Strategically Scaling - experimental mindset achieving scale and returns
- Industrialized for Growth - enterprise culture of AI, clear vision, metrics, and governance
Getting the AI transition right has significant financial returns. The Accenture analysts found a positive correlation between successfully scaling AI and three key measures of financial valuation: Enterprise value/revenue ratio, price/earnings ratio, and price/sales ratio. Companies that got this right saw an average lift of 32% on each of these metrics.
"Our research indicates that moving to the Industrialized for Growth stage will enable competitive differentiation, which is correlated with significantly higher financial results," Berkey said.
CEOs need to accomplish these three tasks to reach the growth stage of AI:
- Master the data set
- Make AI a team sport
- Focus on the I in ROI
Here is Accenture's advice on how to do that.
Master the data set
Strategic Scalers are experts at structuring and managing data from creation to custodianship to consumption. They recognize the importance of business-critical data and consider financial, marketing, consumer, and master data as priority domains.
"This can be a daunting area to tackle, and hard to pin specific ROI to, but getting this step right will pay dividends in terms of what an organization can accomplish when it gets this step right," Berkey said.
Berkey continued that the most successful companies invest heavily in data quality, data management, and data governance frameworks.
"Companies that are successfully scaling AI are more likely to wield a larger, more accurate data set, and they're integrating internal and external data sets as a standard practice," he said.
They use the right tools as well: Cloud-based data lakes, data engineering/data science workbenches, and data and analytics search.
Make it a team sport
Everyone has to be on board for AI projects to succeed. The first step is to make sure the work is front and center with company leadership. The next step is to form a diverse team to manage the work. Berkey said top performers outpace the struggling groups by 20 percentage points in this area alone.
"Companies that struggle to scale AI are more likely to rely on a lone champion within the technology organization to drive AI efforts, which our research found is not enough," Berkey said.
These teams should include data scientists, data modelers, visualization experts, machine learning, data and AI engineers, among others, and have clear sponsorship from the top.
Ongoing training is also critical to ensure employees have an understanding of how AI applies to their role, and understand and implement responsible AI.
"As important as it is to get the technology aspect right, companies won't be successful with adoption unless employees are primed to get on board with AI initiatives," Berkey said.
Focus on the I in ROI
Accenture analysts report that the companies that have made it to the third phase of the AI revolution have invested significant time and effort in laying the groundwork. These companies set longer timelines for AI projects, ranging from one to two years.Accenture reports that these leaders are "more intentional with a more realistic expectation in terms of time to scale and what it takes to do so responsibly."
Berkey said success also requires flexible business processes, and clearly defined accountability."They need to have structure and governance in place – including a clearly-defined strategy and operating model," he added.
The Accenture analysts also found that size does not matter. Small companies are just as likely to succeed as large ones, as long as the executive team installed the right AI mindset.
- Machine learning: A cheat sheet (TechRepublic)
- Telemedicine, AI, and deep learning are revolutionizing healthcare (TechRepublic download)
- Managing AI and ML in the enterprise 2019: Tech leaders expect more difficulty than previous IT projects (TechRepublic Premium)
- What is AI? Everything you need to know about Artificial Intelligence (ZDNet)
- 6 ways to delete yourself from the internet (CNET)
- Artificial Intelligence: More must-read coverage (TechRepublic on Flipboard)