Companies are making progress on internal ethics but training still lags.
Companies are starting to see a few benefits from artificial intelligence, but there are still many roadblocks to progress. In Living in an AI World: Achievements and Challenges of Artificial Intelligence Across Five Industries, KPMG looks at what's working and what isn't in five industries. The report explores how leaders in healthcare, financial services, transportation, technology, and retail are using artificial intelligence (AI) to change operations and workflows.
SEE: Managing AI and ML in the enterprise 2019: Tech leaders expect more difficulty than previous IT projects (TechRepublic Premium)
Two findings were not a surprise. A majority of leaders in four of the five industries report that AI is overhyped, ranging from 57% of tech leaders and going up to 69% among transportation industry leaders. Only financial services leaders said that the reality of AI was beginning to live up to its marketing promises. Another not so surprising finding was that C-level executives are much more likely than managers to think that workers are prepared for the AI transformation: 79% compared to 38%.
There were three surprises in the report about federal guidelines, ROI expectations, and which industries are staying on track with retraining efforts.
Leaders want federal guidance
The report found that a clear majority of respondents think the government should be involved in regulating AI technologies. Transportation insiders are the most in favor at 82% with 59% of financial services insiders favoring regulation. A significant number of leaders in transportation and tech think government leaders should be very involved at 41% and 39%.
The survey also showed that as professionals gain more experience with AI, their support for government oversight goes up. Ninety-one percent of respondents whose companies are using AI at scale, and 75% of those with high AI knowledge said the government should be involved in regulating the technologies.
Companies are doing this work internally with 44% reporting that their companies have already implemented an AI ethics policy code, and another 30% said their companies are working on one.
SEE: Artificial intelligence ethics policy (TechRepublic Premium)
AI ROI requires time and patience
Executives underestimate the time and effort required to see an ROI from AI, said Dr. Sreekar Krishna, principal, innovation and enterprise solutions at KPMG.
"Value ... comes from continuing to run the system, and as your processes are transformed by what the model is doing," he said. "... at the one-year mark, you'll begin to see tangible benefits."
Traci Gusher, principal, innovation and enterprise solutions, US lead, artificial intelligence, said that leaders also need to acknowledge the effort needed from the workforce to train AI efficiently.
"AI can be a bit like a toddler – it learns fast – but only through education and observation," she said.
Gusher also said that executives need to think about AI as more than just the latest tech tool.
"It's about using AI as a strategic lever to transform the business. And that requires building deep AI capabilities across the organization–both from the bottom up and the top down–across technology, people, data and process."
The retail industry's evolving approach to AI is a good example of how strategies must evolve. Gusher said retailers were early adopters of AI, using it for marketing-related functions, such as customer segmentation and analyzing customer churn. Progress stalled until companies started expanding the work to other areas of the business.
"It's only now that retailers have started applying AI to middle- and back-office functions that they are beginning to reap important benefits – particularly in cost reductions," Gusher said.
Training lags in most organizations
Industry leaders listed a lack of training as the second biggest challenge to integrating AI into the workplace, right behind a lack of understanding of the technology's capabilities.
The healthcare and retail industries report the lowest level of support – with only 47% of healthcare, and 52% of retail leaders saying their companies offer an AI training course. Financial services, transportation, and tech firms lead the way with 57% of financial services respondents and 65% of technology and transportation respondents reporting that their companies offered training.
Bob Parr, chief data officer, advisory, at KPMG, said that reskilling needs to happen at every level of a company to ensure that data literacy is a company-wide skill. Parr also said that organizations need to be deliberate about getting separate teams to collaborate by setting up projects that combine subject matter experts and data scientists.
"You need to educate teams about what each employee group brings to the table and create a framework in which people can engage with each other productively," he said.
The next phase of the AI transformation
As companies continue to implement AI, the report authors recommend taking these steps to get the most from the technology:
- Invest in R&D around AI and related capabilities to expand existing deployments into other areas of the business
- Build data literacy across the entire organization
- Create a data-first culture by increasing technical talent
- Prioritize privacy and security issues by designing AI algorithms that align with corporate values
- Modernize IT infrastructure to take AI into production and avoid common "failure to launch" challenges
The survey was conducted by Ketchum Analytics for KPMG. The findings in this report are based on the results from a survey of 751 US business decision-makers with at least a moderate knowledge of AI in their industry.
- Machine learning: A cheat sheet (TechRepublic)
- Telemedicine, AI, and deep learning are revolutionizing healthcare (TechRepublic download)
- Artificial inteligence ethics policy (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)