Generative AI (genAI) is rapidly becoming an essential of enterprise technology strategies. According to Gartner Research, by 2026, a staggering 80% of enterprises will have integrated genAI APIs, large language models (LLMs), or genAI-enabled applications into their production environments.
A dramatic increase from the less than 5% adoption rate seen in 2023 clearly shows the growing recognition of genAI’s potential to drive innovation, efficiency, and competitive advantage across various industries.
Currently, 9% of companies are already using genAI to transform their business models and create new opportunities.
Organizations are making use of genAI to automate repetitive tasks, generate actionable insights from vast datasets, and improve customer experiences through advanced personalization. As more companies witness these benefits, the adoption rate is expected to continue its upward trajectory.
When generative AI projects crash
Despite the promising outlook, nearly a third of genAI projects are expected to be abandoned by the end of next year.
According to a Gartner survey of 822 corporate leaders and board directors, the primary reasons for project failures include poor data quality, inadequate risk controls, escalating costs, and unclear business value.
High rates of abandonment clearly shows the major challenges that organizations face when implementing this technology. Key issues include:
- Poor data quality: Leads to inaccurate models and unreliable outputs, undermining the entire project.
- Inadequate risk controls: Expose organizations to potential security breaches and compliance issues, which can be particularly damaging in industries like finance and healthcare.
- Escalating costs: Associated with genAI projects, often ranging from $5 million to $20 million, can quickly become prohibitive, especially when the business value is not immediately apparent.
- High rate of project abandonment: Poor data quality and inadequate risk controls collectively contribute to this, indicating the need for more comprehensive planning and execution strategies.
The high cost and complexity of deploying Generative AI
Implementing genAI is complex and costly. The financial burden of deploying AI solutions can range from $5 million to $20 million, making it a significant investment for any organization. Costs like these cover the development and deployment of AI models, integration with existing systems, and ongoing maintenance and updates.
Gartner predicts that by 2028, more than half of the enterprises that have built LLMs from scratch will abandon these projects. Reasons for this abandonment include the high costs, complexity, and accumulated technical debt associated with deployments.
Technical debt, which refers to the implied cost of future rework caused by choosing an easy solution now instead of a better approach that would take longer, can escalate quickly in AI projects.
As models become more complex and the infrastructure supporting them grows, maintaining and updating these systems becomes increasingly challenging and expensive.
The rollercoaster of generative AI investments
Unlike traditional technology investments, where costs and benefits can be more easily forecasted, genAI projects are influenced by a wide range of factors, including the specific use cases, the level of investment, and the deployment approaches taken.
The costs and returns from genAI investments can be highly unpredictable.
A company investing heavily in custom AI models might see higher upfront costs but also greater long-term returns through more tailored and effective solutions.
Conversely, organizations that opt for off-the-shelf models might save on initial costs but could face limitations in customization and integration, potentially reducing overall benefits. Variability makes it key for companies to carefully assess their specific needs and contexts when planning genAI investments.
Understanding these nuances and preparing for potential fluctuations in costs and returns can help organizations better navigate the complexities of genAI implementation and maximize their chances of success.
How early adopters are winning with Generative AI
Despite the challenges, early adopters of genAI are reporting substantial benefits. On average, companies have seen a 15.8% increase in revenue, a 15.2% reduction in costs, and a 22.6% improvement in productivity.
Companies using genAI for customer service automation have reported faster response times and higher customer satisfaction rates.
In manufacturing, genAI is being used to predict maintenance needs and optimize production schedules, leading to reduced downtime and increased efficiency. Early success stories demonstrate that while the road to genAI adoption may be fraught with challenges, the rewards can be substantial for those who navigate it successfully.
Expanding the scope of AI usage
In 2022, 55% of organizations experimented with generative AI (genAI) in their workflows, according to McKinsey & Co which shows an interest in exploring the potential of genAI, yet the application remains limited.
Less than a third of these enterprises employed AI for more than one function.
Limited scope of use shows that while organizations are interested in genAI, many are still in the pilot stages, testing its capabilities in isolated parts of their operations before considering broader implementation.
A cautious approach is likely due to the nascent stage of the technology and the uncertainties surrounding its integration and benefits.
Trends in Generative AI investment
According to LucidWorks’ second annual GenAI Global Study, 63% of global companies plan to increase their AI spending in the next 12 months which is a notable decrease from the 93% who had planned to increase spending in 2023, signaling a more cautious approach towards AI investments.
LucidWorks’ study also explains a specific trend within the financial services sector, where only a quarter of the AI initiatives planned for 2024 are being deployed. This hesitation can be attributed to several key concerns.
Financial services and AI
In the financial services industry, data security remains the top concern, cited by 45% of the leaders surveyed. This is followed closely by accuracy (43%) and cost (40%).
Concerns among financial organizations clearly show the high stakes in financial services where data breaches can have severe consequences, accuracy is paramount for decision-making, and costs can quickly escalate.
The industry’s cautious stance suggests that while there is recognition of AI’s potential, the risks and investment required are leading to a more measured approach.
The competitive drive behind AI Investment
Companies are often driven to deploy genAI tools due to competitive pressures. As businesses strive to stay ahead, genAI becomes a strategic investment to avoid falling behind.
By 2030, it is projected that companies will spend $42 billion annually on genAI projects, encompassing applications such as chatbots, research tools, and content summarization. Significant investment clearly shows the belief that genAI is a technological trend and a competitive necessity.
Businesses aim to use AI to simplify operations, increase customer engagement, and drive innovation.
The elusive benefits of Generative AI
Measuring the return on investment (ROI) for genAI projects remains a challenge for many organizations. According to surveys, 42% of companies have not yet seen significant benefits from their genAI initiatives. While the tech and retail sectors report higher deployment and realized gains, most industries struggle to move beyond pilot programs.
Rising concerns about security and costs, which have surged 14-fold and five-fold respectively in the past year, further complicate the ROI assessment.
How to measure ROI in Generative AI
Determining the ROI of genAI is complex due to its indirect and non-financial impacts. Automating code generation can increase a software developer’s productivity, leading to faster product development and improved customer satisfaction.
While impactful, these benefits are not immediately quantifiable in financial terms.
To accurately measure ROI, companies should view it as a financial ratio of the investment’s gain or loss relative to its cost which involves tracking direct financial gains and considering longer-term impacts on efficiency and innovation.
How leaders can win with Generative AI
- Your roadmap to Generative AI success: Executives aiming to pursue genAI projects need a clear strategy. Gartner outlines several crucial steps for understanding this complex technology.
- Aligning AI with business goals: First, it’s important to determine the potential business value derived from genAI by aligning it with strategic business adjustments. This means identifying areas where genAI can drive the most significant impact and integrating these insights into the broader business strategy.
- The true price of Generative AI: Next, calculating the total costs of genAI deployment is critical which includes the direct expenses of implementing the technology and the associated costs of necessary business adjustments. A comprehensive cost analysis helps in understanding the full financial commitment required.
- Evaluating ROI in AI projects: Assessing ROI involves estimating the financial returns from genAI initiatives and comparing these returns to the total costs. This step is key for making informed investment decisions. If the ROI meets or exceeds expectations, it’s an indicator to expand investments by scaling genAI across more business divisions or a broader user base.
- Deciding when and how to scale AI: Finally, if the ROI falls short, it’s important to reconsider investments and explore alternate scenarios. A flexible approach makes sure that resources are allocated efficiently and that genAI initiatives align with overall business objectives.
Why small AI models are gaining popularity
Markets show a clear preference for commercial large language models (LLMs), with nearly 80% of companies using these models. Yet, there is a growing interest in customized small models, with 21% of organizations opting for open-source solutions exclusively.
Shifting towards customized models is driven by the desire for greater control and specificity in AI applications. Small models tailored to specific organizational needs can offer more precise and relevant insights, though they often require more significant initial development efforts.
