Among the many business disruptions caused by covid-19, one has been largely overlooked: the impact of artificial intelligence (AI).
As the pandemic started to disrupt the world last year, companies began to use a variety of available tools (including artificial intelligence) to solve challenges and serve customers safely and effectively.in A KPMG survey in 2021 In a survey of U.S. corporate executives between January 3 and 16, half of the respondents said that their organizations have accelerated the use of AI in response to covid-19 – including 72% of industrial manufacturers, 57% of technology companies and 53% of retailers.
Most people are satisfied with the results. 82% of respondents agreed that artificial intelligence would help their organizations during the pandemic, and most said that it provides more value than expected. More broadly, almost everyone said that the wider use of artificial intelligence will enable their organizations to operate more efficiently. In fact, 85% of respondents want their organizations to accelerate the adoption of artificial intelligence.
Nevertheless, emotions are not entirely positive. Even if they are seeking to refuel, 44% of executives believe that their industry is developing faster than expected in artificial intelligence. What’s more surprising is that 74% of people believe that using artificial intelligence to help companies is still more hype than reality — a sharp increase in key industries since our artificial intelligence survey in September 2019. For example, in the financial services and retail industries, 75% of executives now believe that artificial intelligence is over-hyped, higher than 42% and 64%, respectively.
How to align these seemingly opposing views with KPMG’s so-called AI spurs? Based on our work to help organize the application of artificial intelligence, we have seen several explanations for the hype. One is the simplicity and novelty of the technology, which allows people to misunderstand what it can and cannot do, how long it will take to achieve enterprise-scale results, and what mistakes organizations might make when trying artificial intelligence without the right foundation .
Although 79% of respondents said that artificial intelligence has at least moderate functionality in their organization, only 43% of respondents said that it is fully functional at scale. It is still widely believed that AI is something that can be purchased—like a new machine—that can produce immediate results. Although they may have some success in artificial intelligence—usually a small proof of concept—many organizations have learned that extending them to the enterprise level can be more challenging. It requires access to clean and well-organized data; a powerful data storage infrastructure; subject matter experts to help create labeled training data; sophisticated computer science skills; and purchase from companies.
Of course, it is no exaggeration to say that the proponents of artificial intelligence may from time to time exaggerate its potential or underestimate the effort required to realize its full value.
As for why executives are at odds with the speed of AI adoption, we believe that basic human nature is at work. For starters, it is always easier to believe that the grass on the other side is greener. We also suspect that many people are worried that their industry is developing too fast, mainly because their own organizations have not kept up with this speed. If they have experienced early-stage small problems in artificial intelligence—especially last year, when the world witnessed the achievements of artificial intelligence, such as the development of a covid-19 vaccine at a record speed—may easily succumb to these fear.
We see another factor that causes people to have a complex feeling about the potential of artificial intelligence—the lack of established legal and regulatory frameworks to guide its use. Many business leaders do not know what their organizations are doing to manage artificial intelligence, or what new government regulations may be in the future. Understandably, they are worried about related risks, including use cases developed today that may be suppressed by regulators tomorrow.
This uncertainty helps explain another seemingly contradictory finding in our investigation. Although corporate executives are generally skeptical of government regulation, 87% said that the government should play a role in regulating artificial intelligence technology.
Moving on from AI whipping
Although every organization needs its own manual to recover from the impact of AI and optimize its technology investment, a comprehensive plan should include five components:
- Strategic investment in data. Data is the raw material of artificial intelligence and the connective tissue of digital organization. Organizations need to tag clean, machine-digestible data with the help of subject matter experts to train AI models. They need a data storage infrastructure to transcend functional silos within the business and be able to deliver data quickly and reliably. Once the models are deployed, a strategy and method of collecting data is needed to continuously adjust and train them.
- The right talent. Computer scientists with AI expertise are in great demand and difficult to find, but they are essential for understanding the AI landscape and guiding strategies. Organizations that cannot build a complete team of scientists in-house will need external partners to fill in the gaps and help them organize the ever-expanding AI suppliers and products.
- Long-term AI strategy guided by business. By thinking about finding solutions to problems, rather than buying technology and finding ways to use it, organizations can get the most from artificial intelligence. They let the business, not the IT department, drive the agenda. When there are problems with AI investments related to business-led strategies, they will become opportunities for rapid failure and learning, rather than rapid and burning opportunities. But even if companies iterate quickly, they need to do so according to a long-term AI strategy, because the biggest benefits are realized in the long-term.
- Culture and employee skills upgrade. Without the support of the labor force and a culture of investment in artificial intelligence success, few artificial intelligence agendas will gain attention. Winning the promise of employees requires at least a preliminary understanding of technology and data, and a deeper understanding of how it will benefit them and the company. It is also important to improve the skills of the workforce, especially when artificial intelligence will take over or supplement their existing responsibilities. Embracing a data-driven way of thinking and instilling deeper AI literacy into the organization’s DNA will help them scale up and succeed.
- Commitment to the ethical and fair use of artificial intelligence. AI has a bright future, but if organizations use it in a way that customers don’t like or discriminate against certain groups of people, it can also cause harm. Every organization should develop an AI ethics policy and develop clear guidelines on how to deploy the technology. The policy should mandate measures and become part of the DevOps process to check for problems and imbalances in the data, measure and quantify unexpected deviations in machine learning algorithms, track the source of the data, and identify who trains the algorithms. The organization should continuously monitor the deviation and drift of the model and ensure that the interpretability of the model decision is in place.
The goals of executives for AI investment in the next two years will vary from industry to industry. Healthcare executives said their focus will be on telemedicine, robotic tasks and patient care services. In the life sciences field, they said they will seek to deploy artificial intelligence to identify new revenue opportunities, reduce management costs, and analyze patient data. Government executives said their focus will be on improving process automation and analysis capabilities, as well as managing contracts and other obligations.
The expected results also vary from industry to industry. Retail executives predict the greatest impact will be in the areas of customer intelligence, inventory management, and customer service chatbots. Industrial manufacturers see it in product design, development, and engineering; maintenance operations; and production activities. Financial services companies want to do better in fraud detection and prevention, risk management and process automation.
In the long run, KPMG believes that artificial intelligence plays a vital role in reducing fraud, waste and abuse, and helping companies strengthen sales, marketing and customer service operations. Ultimately, we believe that artificial intelligence will help solve basic human challenges in areas such as disease identification and treatment, agriculture and global hunger, and climate change.
This is a future worth striving for. We believe that both the government and the industry can play a role in achieving this goal-jointly formulating rules to promote the ethical development of artificial intelligence without stifling the innovation and motivation that have been carried out.
Learn more at KPMG “Booming in the Artificial Intelligence World” report.
This content was produced by KPMG. It was not written by the editors of MIT Technology Review.