Adoption of AI Systems

AI adoption is no longer a distant frontier. Many businesses are integrating AI systems to gain a competitive edge. But like any transformation, the journey comes with many hurdles. Common barriers include lack of expertise, financial constraints, data management complexity, and uncertainty about the regulatory landscape.

According to a McKinsey report, 72% of companies surveyed are now using AI in at least one business function, a significant jump from prior years. This trend is global, with 66% of respondents reporting AI use. The increased focus on generative AI (gen AI) is driving much of this momentum. Businesses are moving beyond traditional applications toward exploring gen AI solutions for automation, security, and AI governance.¹ Moreover, investments in gen AI were nearly nine times higher in 2023 than the previous year, reflecting a surge in interest and funding.² Yet for many businesses, the path to effective AI adoption is filled with challenges.

This post in our AI Governance Frameworks Comparison (AIGFC) series investigates the barriers that organizations face when implementing an AI system. We explore the latest trends in AI adoption and investment, then consider key challenges that organizations encounter, from planning and resource constraints to deployment and operational issues. By understanding these obstacles, businesses can develop strategies to overcome them and leverage the full potential of AI systems.

Contents

• Adoption and Investment Trends
• Signs of Growth
• Signs of Stagnation and Struggle
• Barriers to Effective Adoption
• Overview of Barriers
• Planning and Resource Barriers
• Deployment and Operation Barriers
• Discusison of Barriers
• Conclusion
• Related Posts

Adoption and Investment Trends

Signs of Growth

McKinsey's "The State of AI in Early 2024" indicates that most businesses have now adopted AI in some capacity. As shown in Figure 1.1, while adoption stood at around 50% from 2019-2023, 72% of businesses surveyed in early 2024 reported using AI in at least one business function. What's more, McKinsey's results indicate that this increased adoption in 2024 has been truly global. At least 66% of respondents from nearly every major global region report using AI.¹

The growing prevalence of AI among businesses is also indicated elsewhere. According to Stanford's "AI Index Report 2024," which broadly exploring recent developments in AI, AI was mentioned in the earning calls of nearly 80% of Fortune 500 companies in 2023, up 48% over the previous year.² A 2024 guide to the wide-scale business implementation of AI found that 87% of the 803 Western European C-level executives interviewed are at least actively exploring investment in AI tools.³

Various use cases are motivating businesses to adopt AI. Among companies currently exploring or deploying AI, a 2023 survey from IBM found that automation of IT processes was the most common use case motivating adoption (33%), followed by security and threat detection (26%) and AI monitoring or governance (25%). Research and development (44%) and reskilling/workforce development (39%) were the most common AI investments.⁴

By region, the US far exceeded all other regions in total 2023 private investments in AI. The US's total investment of $67 billion is a 22% increase on its 2022 investments and around three times as much as the investments of all other countries combined. In fact, US's total investment is 8.7 times higher than that of the next largest investor, China.²

McKinsey attributes much of the newfound adoption and investment in AI to the rapidly developing capabilities of generative AI. In their 2023 survey, 33% of respondents were using gen AI with adoption increasing to 65% in 2024.¹ Stanford's report supports this conclusion, showing that 2023 gen AI investments were around nine times higher than in 2022. Changes to investment in AI infrastructure, research, and governance for 2023, mainly in the major labs producing gen AI such as OpenAI and Anthropic, were even more extreme at around 20 times higher year-to-year from under $1 billion in 2022 to over $18 billion in 2023. One-third of 2023 US investments were into gen AI, making up around 90% of global gen AI investments.²

Many businesses are expressing optimism about the future impact of gen AI. KPMG's 2023 "Generative AI Survey" found that 80% of respondents believe gen AI will disrupt their industry, while 77% believe it will be the most impactful emerging technology for their business, well ahead of second-placed 5G at 40%.⁵

However, gen AI investment remains a lower priority than analytical AI for most industries. McKinsey found that while some industries allocated, on average, budgets of around 5% to gen AI and analytical AI each, most industries are reporting at least 20% more spending on analytical AI over gen AI.¹ This result is an important reminder of the high application value of analytical AI despite its lower coverage.

Signs of Stagnation and Struggle

A recent rise in global AI adoption, fueled by rapid gen AI development and high US investment, strongly indicates growth in AI adoption and investment. However, over the same period, many regions have seen declines in private AI investment while many businesses have struggled with AI deployment.

Figure 1.2 shows global corporate investment activity for the last decade. Before 2021, global corporate AI investment activity had grown rapidly year-on-year for a decade. However, overall global AI investment has declined since then. Investment activity peaked at $337 billion in 2021, declined to $235 billion in 2022, and then declined to $189 billion in 2023.²

Private AI investment in China, the EU, the UK, and even the US have seen declines and stagnation over the same period. Whereas private Chinese investors invested over $20 billion in 2021, it fell to just below $15 billion in 2022 and $8 billion in 2023. Meanwhile, EU and UK investments have remained fairly constant since 2021, holding steady at around $11 billion annually. And, despite its investment growth since 2022, US private investment in AI has thus far peaked in 2021 at $80 billion, 23% higher than 2023's investment of $67 billion.²

We also see a widespread struggle among businesses to adopt AI. Several surveys from IBM, largely of US-based investors, as part of a guide for CEOs on adopting AI, emphasize the impact of current barriers on adoption and investment rates. Fewer than 60% of US-based executives surveyed feel prepared for AI regulations, and 56% are delaying large investments until there's greater regulatory clarity. Furthermore, 72% of organizations will forgo gen AI benefits due to ethical concerns.⁷

Meanwhile, IBM's "Global AI Adoption Index 2023" found that Data Privacy (57%) and trust and transparency concerns (43%) were the most substantial barriers to adoption, according to organizations not exploring or implementing gen AI.⁴ Technical difficulties (39%), specifically integration with existing tools, are among the highest barriers to businesses adopting and scaling AI tools.³

Such barriers are plausibly preventing many businesses from thoroughly integrating AI. When asked how far along respondents' businesses were in using AI on a four-stage scale, software development company Retool's 2024 "State of AI in Production" found that only 30% of respondents thought their business was in either of the top two stages. What's more, the proportion of respondents self-reporting as being in the highest stage of use in 2024 was down to 9.8% from 13.4% in 2023.⁶

Barriers to adopting AI in 2023 and 2024 seem part of a broader trend. A 2020 survey from IT consultant company Capgemini of 993 companies with over $1B in annual revenue found that only 13% of respondents' organizations had rolled out multiple AI applications across multiple teams. Meanwhile, 72% of organizations had started piloting AI applications in 2019 and still hadn't been able to deploy a single application in production. Lack of mid-to senior-level AI talent (70%), lack of change management processes (65%), and lack of strong governance models for achieving scale (63%) were the top barriers to deployment.⁸

Reduced US venture capital investment in China and a weak innovation ecosystem in the EU are among the many factors proposed to explain recent investment declines.^{9, 10} Simultaneously, barriers to AI adoption seem persistent and widely impact businesses. Nonetheless, businesses can make progress on capturing the value potential of AI while mitigating related risks by understanding and addressing key barriers to adoption.

Barriers to Effective Adoption

Overview of Barriers

To structure this section, we have organized barriers as relevant to planning and resources or the deployment and operation phases of the AI system lifecycle.

Planning and Resource Barriers:

• Lack of Expertise:  when key roles lack knowledge or skills related to AI
• Cost and Investment:  resources necessary to deploy and operate AI
• Uncertainty about Business Case:  strategic uncertainty about the business use cases for and purpose of AI
• Uncertainty about Regulations:  strategic uncertainty about what future regulations for AI will involve
• Cultural Resistance:  where a large proportion of a business's staff are opposed to AI adoption

Deployment and Operation Barriers:

• Risk Mitigation:  AI risks and risk mitigation practices in alignment with standard definitions of AI risk, excluding risks explicitly related to data
• Data Management:  related expressly to data practices, including some overlap with risk mitigation
• Technical and Organizational Logistics:  barriers related to practically deploying AI on technical and organizational levels

Planning and Resource Barriers

Lack of expertise

A shortage of specialized skills, strategic understanding, and effective leadership seem to be the most substantial planning and resource barriers to AI deployment and scaling of initiatives.

One study of global IT professionals shows that 33% identify limited AI skills and expertise as a top obstacle to AI deployment.⁴ Among US business leaders, 24% cite a lack of leadership understanding or strategy, 28% report a shortage of skilled talent to develop and implement AI solutions, and 16% find gen AI challenging to use and learn.⁵ West European C-suite executives echo this mix of leadership and technical skill gaps, with 28% pointing to an internal skills gap.³

Further complicating matters, a comprehensive survey of organizations already using AI indicates that 35% face challenges related to talent when trying to capture value from gen AI.¹ Best practice assessments reveal that a vast majority fall short in several areas: 81% lack curated learning journeys tailored to build essential gen AI skills; 84% have not clearly defined the talent needed to execute their AI strategy; 83% do not maintain a talent strategy that supports effective recruitment, onboarding, and integration of AI-related talent; 79% have not appointed a credible, empowered leader for AI initiatives; and 78% of nontechnical personnel do not fully grasp the potential value or risks of gen AI.

Additional global perspectives reinforce these concerns. One survey identifies a lack of skilled staff as one of the three top barriers to AI adoption.⁵ Another survey points out that 53.7% of organizations struggle with a shortage of skilled personnel – such as data scientists, data engineers, or AI modelers.¹²

The data indicate an urgent need to invest in targeted skill development, refine leadership strategies, and educate non-technical staff about the opportunities and risks associated with gen AI.

Cost and investment

Many businesses struggle to afford the necessities of AI adoption. Survey data highlight that the investment cost is a key concern for US business leaders, with 23% identifying it as a primary barrier to implementing gen AI.⁵ Similarly, West European C-suite executives point to high implementation costs as a large obstacle, with 35% emphasizing the financial strain associated with new AI initiatives.³

Beyond the initial outlay, financial management practices play a crucial role. Among organizations that have deployed AI, 86% lack funding and budgeting processes agile enough to support the continuous delivery of gen AI solutions.¹ Responses from outside the US and Western Europe further underscore the cost challenge, as one survey reveals that 54.3% of organizations cite cost as a top impediment to AI adoption.¹²

These insights illustrate that cost and investment challenges extend well beyond upfront expenses. They point to systemic financial planning and resource allocation issues that must be addressed if organizations are to navigate the evolving demands of AI adoption.

Uncertainty about business case

Many organizations struggle to define and articulate the value that AI initiatives are expected to generate, leaving decision-makers hesitant to invest in transformative projects.

Global IT professionals report that the inability to identify the right use case and the lack of a clear business case affect 20% of respondents, indicating that a large portion of organizations face difficulty justifying AI projects from a strategic or financial perspective.⁴ Similarly, senior executives from West Europe reveal that nearly one-quarter of their peers harbor uncertainty about the return on investment, reflecting cautiousness when committing resources to AI initiatives.³

Organizations already deploying AI further illustrate these challenges. A considerable number report strategic hurdles – with 39% citing challenges in capturing value from gen AI. Even more concerning, a large share of best practice respondents indicate that senior leaders do not fully grasp how gen AI can create business value, and a majority have yet to develop an enterprise-wide roadmap that effectively prioritizes AI initiatives based on value, feasibility, and risk.¹

Global surveys also highlight that over half of organizations experience unclear decision criteria and ambiguous business cases or lack sufficient support from key business units. This persistent uncertainty permeates the entire AI adoption process, from initial planning to full-scale deployment.¹²

These insights point to a broader challenge: without a well-defined and compelling business case, the promise of AI remains clouded by ambiguity. Organizations must invest in developing rigorous frameworks that delineate the strategic benefits of AI, align initiatives with measurable outcomes, and secure the support necessary to drive transformative change.

Uncertainty about regulations

Uncertainty about regulations leads to fear of noncompliance and legal exposure when adopting AI. Data indicates that a substantial number of US-based executives feel unprepared for upcoming AI regulations, with 60% expressing concerns about their readiness and 56% delaying major investments until greater regulatory clarity is achieved.⁷

Among US business leaders, apprehensions about the evolving regulatory landscape also play a substantial role. About 30% rank these concerns among the top three barriers to implementing gen AI, while 21% worry about potential legal exposure that could complicate deployment and scaling efforts.⁵ In West Europe, the picture is relatively consistent – 23% of C-suite executives cite regulatory uncertainties as a primary obstacle to adopting and scaling AI tools.³

Businesses seem caught between the promise of innovation and the risks of navigating an unpredictable legal framework, which delays investment and slows overall progress.

Cultural resistance

Internal skepticism and resistance from employees is slowing the pace of adoption within businesses.

In one survey, 28% of US business leaders identified internal cultural resistance as a top obstacle to implementing gen AI.⁴ Similarly, West European C-suite executives report that 31% experience employee resistance as a major challenge when adopting and scaling AI tools.³

The data suggest that even the most advanced AI solutions may struggle to thrive without cultivating a culture of innovation and trust. Resistance may stem from uncertainty about new technologies, concerns over job security, or discomfort with shifts in established work processes. Such barriers highlight the need for organizations to address Change Management proactively by engaging employees, providing targeted training, and fostering open dialogue about the benefits and impacts of AI.

Deployment and Operation Barriers

Risk Mitigation

Recent surveys consistently underscore that risk mitigation remains a pervasive barrier to AI adoption, highlighting concerns that span ethical considerations, trust and transparency, data privacy, and the overall robustness of governance frameworks. These issues are not peripheral – they fundamentally shape how organizations deploy and scale AI systems.

A survey of global IT professionals revealed that ethical concerns are among the top inhibitors to AI deployment, with 23% of respondents citing them as a critical barrier. In the realm of gen AI, the challenges intensify: 43% of professionals reported that trust and transparency issues significantly impede adoption, while 57% flagged Data Privacy as a major concern.⁴ Complementing these findings, a separate survey targeting executives found that 72% expect their organizations to forgo the potential benefits of gen AI due to ethical issues.⁷

Further reinforcing these challenges, 21% of US business leaders in a targeted survey expressed a lack of confidence in the accuracy and reliability of gen AI tools.⁵ In a broader study examining the overall AI lifecycle, 35% of respondents identified risk and responsible AI as key hurdles to capturing value from gen AI.

Moreover, a considerable proportion of businesses appear unprepared to manage these risks: 64% do not consider risk awareness and mitigation as essential skills for technical talent; 76% lack clear processes to integrate risk mitigation into their AI solutions (e.g., by involving legal or compliance functions); 81% indicated that their gen AI models are not designed to facilitate audits, bias checks, or risk assessments; and 82% reported the absence of an enterprise-wide council or board dedicated to responsible AI Governance.¹

In addition, insights from surveys point to governance-related deficiencies as central obstacles. One survey identified the lack of AI Governance and risk management solutions as one of the top barriers preventing broader AI adoption.⁵ In contrast, another survey highlighted that 53.8% of organizations experienced shortcomings in their risk management frameworks, with 50.8% lacking tools to ensure fairness, explainability, and transparency and 52.6% expressing concerns regarding the adversarial robustness of their algorithms.¹²

Together, these findings illustrate that while investment and interest in AI continue to grow, risk mitigation remains a critical stumbling block. Addressing these challenges requires organizations to develop comprehensive governance frameworks, embed risk management practices throughout the AI lifecycle, and enhance technical capabilities that support continuous auditability and bias detection.

Data Management

Data Management as a barrier to AI adoption is characterized by data complexity, inaccessibility, inadequate governance, and security concerns.

A survey of IT professionals in managerial roles revealed that 25% consider excessive data complexity a primary obstacle to AI deployment.⁴ In parallel, another study involving business leaders in the United States found that 16% identified the inability to access and leverage data as a critical impediment to implementing gen AI.⁵ These findings suggest that organizations may struggle to realize the full benefits of AI innovation without streamlined data processes.

Further investigation into organizational practices uncovers deeper systemic issues. Respondents from various industries reported that 27% of organizations lack systems that provide data accessibility and seamless integration. Even more striking, 80% acknowledged lacking a robust Data Governance program to define and enforce essential policies and processes. Additionally, over 60% admitted to a limited understanding of their data sources and insufficient policies for managing diverse data types, while 23% noted the persistence of ungoverned data silos.¹³

Data Security and strategic data planning are also substantial Data Management barriers. In a survey among senior executives, 19% highlighted Data Security as a challenge in adopting and scaling AI tools.³ Moreover, 38% of organizations experienced difficulties in capturing AI-derived value due to data-related issues, with 83% lacking a defined Data Strategy and 76% failing to leverage data consistently to drive performance.¹ Finally, over half of the respondents cited inadequate volumes and quality of training data as a notable hindrance.¹²

These surveys strongly indicate that Data Management is not an isolated issue but a multifaceted barrier that intersects with technical complexity, organizational infrastructure, and strategic execution. For organizations to effectively adopt AI, they need to invest in integrated data systems, develop comprehensive governance frameworks, and implement robust data strategies.

Technical and organizational logistics

Logistical barriers involve a combination of technical constraints and organizational shortcomings. This barrier encompasses issues ranging from legacy systems' rigidity to infrastructure deficits and difficulties in integrating new technologies into existing operational frameworks.

Survey data underscores the multifaceted nature of these logistical challenges. In one study of business leaders, 28% identified an inability to pivot legacy applications and systems as a primary impediment, while 22% noted that insufficient technology infrastructure further restricts progress. Additionally, 16% of respondents described these systems as difficult to use and learn, highlighting a usability challenge that compounds the technical limitations.⁵

Further insights from a survey of senior executives reveal an even more pronounced reliance on external expertise, with 60% indicating that dependence on outside specialists creates a bottleneck in AI deployment. This external reliance is accompanied by hurdles in technical integration – with 39% of respondents pointing to difficulties in integrating AI tools with existing systems – and organizational issues, as 27% reported internal challenges that obstruct effective AI scaling.³

Additional analysis deepens the understanding of logistical constraints. A comprehensive study found that nearly half of respondents (47%) face challenges with their operating models, 43% encounter technology-related barriers, and 33% report difficulties with adopting and scaling AI initiatives. Notably, the integration of best practices is lacking, as evidenced by 82% of organizations that do not embed testing and validation in their model release processes and a striking 93% that have not established live monitoring systems to address issues quickly.¹

Complementing these findings, another study reported that 52% of organizations lack the necessary tools to monitor model performance for data and concept drift, and 51.6% struggle with establishing effective machine learning operations, underscoring the operational gaps that can undermine AI initiatives.¹²

The data illustrates that logistical barriers are not merely technical issues but also reflect broader organizational challenges. Addressing these barriers will require concerted efforts to modernize legacy systems, invest in robust technological infrastructures, and cultivate internal expertise.

Discusison of Barriers

Recent surveys regarding AI adoption reveal a complex range of barriers.

Planning and resource challenges add a significant layer of complexity. A notable shortage of specialized skills and leadership insight is evident, with around 33% of IT professionals and up to 28% of executives citing these deficiencies.^{4, 5} High costs and ambiguous business cases, coupled with regulatory uncertainties and cultural resistance – reported by 28% to 31% of respondents – create a multifaceted obstacle. Many organizations hesitate to invest heavily in AI due to unclear strategic benefits and fears of noncompliance with evolving regulations.^{3, 4, 5, 7}

Another prominent barrier is the suite of challenges associated with risk mitigation. Surveys consistently show that ethical concerns, trust deficits, transparency issues, and data privacy worries are not peripheral but central to AI's deployment. For example, 23% of IT professionals consider ethical dilemmas critical, 43% of those working with gen AI highlight trust and transparency issues, and 57% cite data privacy as a key stumbling block.⁴ This pervasive unease with deploying AI is compounded by many organizations lacking robust governance frameworks, with large portions admitting that risk awareness is neither prioritized nor systematically integrated into their AI strategies.

Data Management challenges further exacerbate the issue. Excessive data complexity and inaccessibility are key impediments, with 25% of IT professionals and 16% of US business leaders underscoring these hurdles.^{4, 5} Unresolved data silos, insufficient policies, and subpar training data quality further impede the effective deployment of AI.

Technical and organizational logistics round out the challenges. Legacy systems, inadequate infrastructure, and difficulties in integrating new technologies with existing frameworks continue to hinder progress. Survey data indicates that 28% of business leaders struggle with outdated applications, while reliance on external specialists and fragmented integration processes further slow down AI initiatives. Nearly half of businesses report challenges with their operating models and technology infrastructures, revealing that logistical constraints are as much an organizational issue as a technical one.

Conclusion

Overall, these investigations underscore that as interest in AI grows, its broad adoption is stymied by intertwined challenges across risk management, data handling, technical logistics, and strategic planning. Addressing these barriers will require a holistic approach – enhancing governance, modernizing infrastructure, cultivating specialized talent, and fostering a culture that embraces innovation.