#5月财经新势力 #
IBM-Generating ROI with AI
Compiler: Zhicheng Enterprise Research Institute Cui Shuai
Recently, AI has suddenly become popular again: generative AI has taken the business world by storm, with large language models (LLMs) such as OpenAI's ChatGPT, Baidu's ERNIE, Google's LaMDA, and Facebook's LLaMA dominating the headlines. At the beginning of 2023, the number of mentions of artificial intelligence on earnings calls increased by 77% year-on-year.
At the same time, capital investment has also increased, and artificial intelligence is becoming an increasingly large part of IT budgets. Global spending on AI-centric systems is forecast to reach $154 billion this year, up 27% from 2022.
But will businesses use these resources wisely? As AI models become faster, smarter, and more reliable, and organizations are racing to take advantage of them, are their return on investment (ROI) expected to be?
The answer is yes – but only if the organization takes a rigorous approach to investing.
IBM's findings reveal a huge achievement gap between AI projects, with few companies delivering the financial value that shareholders expect. In fact, the average ROI of a business plan is only 5.9%, which is well below the typical 10% cost of capital. However, as AI maturity continues to increase, this figure has improved significantly – top-notch companies have achieved an enviable 13% return on investment.
So, what makes them special? How can leaders across industries learn from their successes? The report will elaborate on the following three aspects:
- Why ad hoc AI projects provide less value than strategic projects
- The impact of trusted data and the virtuous cycle of AI-data symbiosis
- Six key competencies for top organizations.
Building a full-fledged AI organization is necessary to unlock the full potential of AI, and companies that do this are creating tremendous business value, not just media hype.
Download the full PDF report, follow the official account of "Zhicheng Enterprise Research Institute", and reply in the message dialog box: ROI
01 Beyond opportunistic AI
Businesses have been betting on AI for nearly a decade, and its learning curve has been steep.
However, some companies are attracted by the "surprise factor" that technology brings and forget to combine projects with strategy. Other businesses see AI as a hammer, where every business problem is a nail. Almost all are working to realize the application opportunities of AI beyond experimentation, proof of concept, and pilot.
The good news from the research is that many organizations have already weathered the storm and the adoption of AI is now more successful than ever, with more than twice as many executives saying their companies were effectively leveraging AI in 2021 (54%) than in 2020 (25%). They also expect AI investment to grow to 6.5% of IT spending by 2024.
Overall, the ROI of AI has been steadily increasing since 2020. For enterprise-wide AI initiatives, the average ROI has grown from just over 1% at the beginning of 2020 to nearly 6% at the end of 2021. This may be due to the coronavirus pandemic prompting businesses to invest in AI-driven solutions that will speed up remote work, enhance the user experience, and reduce costs.
To gauge whether AI ROI is keeping up with this trend, IBM surveyed more than 350 executives again in April and May 2023. The survey found that AI's ROI is continuing its expected growth trajectory, reaching an estimated 8.3% by 2022.
Still, these returns are below the cost of capital, which is typically 10% in most industries. Overall, less than a quarter of businesses in our survey said they achieved ROI above 10%.
Essentially, AI is following a "J-curve" pattern, typical of transformative technologies, and mass adoption of emerging technologies requires reimagining business models, workflows, skills, and many other aspects of business. In the process of team problem solving, the rate of return tends to stagnate; However, as capabilities mature, performance can ramp up quickly, and in this environment, companies need to have a strategic plan in place to amplify the impact of AI over time.
IBM's analysis reveals how ROI is improving in AI maturity, with the average enterprise pursuing opportunistic AI initiatives being laggards. In 2021, the ROI of companies interested in embedding AI into products, services, business units and functions climbed to more than 7%, and as maturity expanded further and AI was deployed as part of strategic business transformation, the return increased again to 8%.
As companies figure out where and how to deploy AI, a lot of investment translates into bigger and bigger gains. At the top of the curve are the top performers with an average ROI of 13%. By taking a stable, balanced approach to leveraging AI, which includes building data and analytics skills, developing a multidisciplinary approach, creating diverse teams, and training teams through AI centers of excellence, they have now developed comprehensive capabilities across the enterprise that pay off handsomely.
02 Data and Artificial Intelligence: A virtuous cycle of feeding learning
The first factor in building a world-class AI organization is how the data that is selected, collected, managed, and used — a rich but elusive resource that either enables AI or limits it.
Data is sometimes likened to oil: a precious resource that is expensive to extract, difficult to process, and if polluted, it pollutes entire ecosystems, but if used wisely, it is worth billions of dollars.
That's because reliable, representative, agreed-upon data is the foundation of trustworthy AI. People don't use AI solutions they don't trust, and companies that take AI ethics more seriously say their customers and employees trust AI more deeply.
In addition to this, reliable data can help close the ROI gap. Companies with high "data wealth," while not yet world-class, have a lot of high-quality data, are able to effectively monetize it, and say their data is trusted by internal and external stakeholders. IBM's analysis shows that these attributes drive above-average ROI and are more effective in making AI projects possible.
In fact, high-quality, high-value, trusted data only supports half the benefits of a company's ROI improvement. That said, data alone is not enough to realize the full potential of AI. While data quality, quantity, robustness, value, and trust are all important, how businesses leverage data has a greater cumulative impact on ROI than the data they have.
Today's top-performing chief data officers (CDOs) are focused on extracting value from an organization's data. In IBV's survey, only 8% of CMOs received more recognition of value than their peers and were able to achieve higher returns with less spending, and the key was how they used AI to improve their data. Three-quarters of them said applying AI to their data could help them make faster and better business decisions.
So, it's not just about using data to improve AI, AI can also help companies make better use of data. It's a virtuous circle. As Mirco Bharpalania, Senior Director of Multidisciplinary Solutions at Lufthansa Group, says, "AI is very important because it actually opens up the world of data in which we live. ”
03 Six key competencies to achieve world-class results
So, what enables some businesses to achieve world-class ROI outcomes from their AI investments? And how can they amplify the utility of high-quality, trusted data to unlock financial and business value?
To answer these questions, IBM found that best-in-class AI performers build capabilities in six key areas in a holistic, integrated way, with trust at their core.
Applying artificial intelligence, automation, or any other technology to processes that don't work will still yield subpar results. Only by evaluating core and non-core functions (e.g., customer service, marketing, supply chain, finance, etc.) and strategic investment plans for business units can leaders identify strategic opportunities to embed AI.
A well-thought-out AI strategy can drive transformational outcomes and increase the ROI of individual AI projects. According to IBM's research, companies that value the impact of AI on business strategy are 1.8 times more likely to have AI initiatives effective and achieve nearly twice the ROI.
Leaders also need to balance competitive differentiation with cost optimization. Some companies even leverage publicly available, open-source AI resources to deliver faster, cheaper, and more scalable solutions to the market. Ethical questions about how to train these tools will also play a role in the future development of AI, so companies will need to define their position before going too far.
Breakthrough AI is built on open innovation, however, only leading companies have learned how to dispel the myth that innovation can do anything. To align experimentation and implementation with strategy, companies must see AI as a discipline, and they must develop ethical principles, develop rigorous governance, and emphasize pragmatism rather than theory.
The first thing to understand is which AI operating model best aligns with your business needs (e.g., centralized, spoke, decentralized structure). The study found that companies with high data wealth, embedding AI operating models into the structure and culture of their organizations, were able to generate 2.6 times more ROI than their peers.
What does this look like? An example is about how a company creates a minimum viable product (MVP). Leaders should outline a clear process for applying AI, starting with identifying the business problem that the solution is intended to solve. Choose the most effective AI projects to advance by setting clear goals for experimental rollouts.
Artificial Intelligence Operations (AIOps) turns great ideas into reality and becomes a flywheel for operating models. It integrates people, processes, and platforms to apply AI quickly and at scale. A successful design process helps teams build, while also monitoring the performance of AI applications, can achieve a high ROI of up to 2.6x.
Engineering disciplines can accelerate AI flywheels to make them work efficiently. Just as many companies use agile DevOps and other software engineering methods to accelerate projects without sacrificing quality, AI Ops can help shorten development cycles, improve collaboration, increase operational efficiency, and deploy solutions more successfully. Standardization and structuring are essential to keep pace with innovation without sacrificing the ethical principles of AI.
Anyone can create proofs of concept, however, for AI models to be effective, useful, and trustworthy, they must be properly integrated into the operating system. What a company can do with AI depends largely on how it selects, manages, analyzes, and applies data across the enterprise, as humans make mistakes and teams need skills and processes to help ensure the right data is selected to support AI models.
This will also have a significant impact on the ROI of AI. In world-class organizations, data teams review the governance, management, ethics, literacy, and other frameworks that people need to access, understand, and trust enterprise data. IT teams evaluate infrastructure and processes to balance AI experimentation with industrial-scale expansion.
Since the birth of AI, the lack of skilled talent and technical expertise has been the biggest obstacle to implementing AI. To remain competitive in a tight labor market, companies must train their teams to use AI effectively and responsibly. Research shows that AI projects are more successful when companies help teams strengthen AI capabilities.
Organizations that actively encourage the sharing of AI knowledge across the enterprise and provide business and technical training to attract new talent have a 2.6x ROI of other organizations. HR and talent leaders – sponsored by businesses – are driving this effort. Between 2018 and 2021, the proportion of CHROs with active plans to retain and retrain employees increased significantly.
When companies face cost pressures, change management is often the first project to be cut. But such austerity can be dangerously short-sighted: small things fail to be big. The right culture — one that values trust — helps solidify AI's capabilities and maturity.
If people don't trust the work AI does, or the data it builds, adoption will lag and rewards will fall. On the other hand, our research shows that cultural maturity is one of the biggest contributors to best-in-class ROI. Organizations perform better when AI becomes part of a company's DNA and change management becomes a broad skill. In fact, project teams using standardized and documented methods, including value realization or benefit tracking, can achieve ROI of up to 2.5x.
Cover: Pixabay
—END—