Artificial Intelligence has become a big deal in today’s world. Companies, governments and investors have put a lot of money into AI systems, hoping to make things more efficient, save money and create chances to grow. There are kinds of AI tools, like chatbots and generative AI and also things like data analysis and machines that can make decisions on their own. All these AI tools are changing how businesses work, but even though AI is developing fast and lots of people are using it, there's a growing worry: many AI projects aren't giving organisations the results they were expecting. However, this doesn't mean that AI itself is not working. AI is still a tool that can help organisations in many ways. The problem might be with how AI projects are being done, not with AI itself. Many organisations are still figuring out how to get the most out of AI and they are learning what works and what doesn't. So, while AI might not be living up to expectations now, it's still an important technology for the future. AI systems are being used everywhere. Businesses use AI systems to make their operations smoother, Investors put money into AI systems, and governments also use AI systems. Generative AI tools and chatbots are types of AI systems; there are also predictive analytics and automated decision-making machines are also AI systems too. AI systems help organisations operate better. Many AI projects do not give organisations what they want. AI projects struggle to give value, and organisations expect a lot from AI projects. AI projects should give organisations what they expect. AI itself is not failing. AI can still help organisations.

However, this does not mean that AI itself is failing. On the contrary, AI technologies continue to improve at an extraordinary rate. Large language models are becoming more sophisticated, machine learning systems are solving more and more complex problems, and new applications are emerging almost every day. The challenge lies not in the technology’s capabilities but in translating those capabilities into sustainable economic and organisational value. As businesses move beyond just experimentation and begin to evaluate real-life outcomes, the gap between AI’s potential and its practical value has become more and more visible.

One of the primary reasons for this gap is the difficulty of creating measurable business value from AI investments. Organisations often adopt AI with expectations of increased efficiency, innovations and profitability. Chen et al. (2025) have argued about the value generated by AI that depends heavily on the interaction between organisational data resources and specific business applications. In other words, AI cannot create value in isolation as it actually requires a supportive environment, high- quality data, and clearly defined objectives. Many organisations invest in AI because of competitive pressure or technological enthusiasm without fully understanding how the technology fits into their existing workflows. As a result, projects may succeed technically while failing economically.

Moreover, research also suggests that the benefits of AI adoption vary significantly across organisations. Shi et al.(2025) found that AI can enhance enterprise value, but the success of implementation depends largely on managerial capabilities and the effective use of data assets. Firms with stronger leadership and clear strategic planning are more likely to achieve positive outcomes. This finding highlights that AI is not a magical solution to organisational problems. Rather, it is a tool whose effectiveness depends on humane decision –making, governance, and management practices.

A Useful example can be seen in AI departments in radiology deployments within radiology departments using vendor systems like Aidoc and Zebra Medical Vision. These tools are designed to detect abnormalities in medical imaging, such as CT scans and X-rays. While they perform well in controlled environments, real- world hospital interrogation has proven to be rather complex. Radiologists often report workflow disruptions, inconsistent false positive rates, and difficulties aligning AI outputs with clinical judgment. In many cases, hospitals continue to rely on traditional diagnostic processes because the cost of restricting workflow and restraining staff outweighs the marginal efficiency gains from AI assistance.

Another factor contributing to the declining perceived value of AI is its high economic cost. Public discussions often focus on AI’s capabilities while overlooking the substantial infrastructure and maintenance. According to d’Orgeval et al. (2026), generative AI models rely on highly energy-intensive data-centres whose environmental and economic impacts extend far beyond the servers themselves. As AI usage expands, so do the costs associated with electricity, cooling systems, and infrastructure development.

Recent scholarship has emphasised the material foundations of AI. Chang and Jiao (20260 argue that digital 9intelligance is fundamentally dependent upon energy systems and physical infrastructure. Although AI is often presented as virtual or immaterial technology, its operation depends on a vast network of power, transmission, and data processing facilities. Consequently, not only software development but also practical concerns, such as availability and infrastructure capacity. These physical limitations complicate the assumption that AI can expand indefinitely without significant economic consequences.

The world of AI and money has changed a lot. At the beginning of the AI excitement, investors were good with paying for what might happen, based on what was happening now. They gave a lot of expenses to companies because they thought AI would change industries and make lots of money. Now that investors have seen many AI projects, they want to see real results. Companies must show that their AI projects are really helping with things, like getting work done, making more money or working better. AI projects must make a difference. Companies are working hard to make their AI projects successful.

This shift reflects a broader transition from technological excitement to economic accountability. While organisations continue to invest heavily in AI, they are becoming more conscious about where and how those investments are made. Instead of pursuing AI projects simply because the technology is popular, companies are increasingly focusing on use cases that offer clear strategic benefits. Studies of AI adoption suggest that successful implementation requires alignment between technological capabilities and organisational goals rather than blind enthusiasm for innovation (Davis & Li, 2024).

Furthermore, the social perception of AI plays an important role in determining its value. Public concerns regarding privacy, job displacement, misinformation, and algorithm bias have contributed to growing scepticism about AI technologies. Trust is a crucial component of technological adoption. Even highly advanced systems may struggle to achieve widespread acceptance if users perceive them as unreliable, unethical or harmful. Consequently, organisations must address not only technical challenges but also social and ethical concerns when implementing AI solutions.

Nevertheless, it would be premature to interpret these challenges as evidence of AI’s decline. Historical examples demonstrate that transformative technologies often experience periods of adjustment before their full value becomes apparent. The current problems are about fitting a new technology into existing systems, not that the technology itself is failing.

In the end, the seeming loss of AI value isn't because the technology is getting worse. This is because people are realising that just having technology doesn't guarantee success. Many companies must deal with cost, infrastructure, governance, strategy and public trust to make AI valuable. As AI keeps changing, its long-term effect will depend not on what AI can do but on how businesses and societies learn to use it. The future of AI will likely belong to those who can turn possibilities into real and lasting results, not just those who invest the most.

References

1. Chen, Y., Zhang, Y., Wang, F., & Zhang, C. (2025). Creating Business Value from Early Generative AI Adoption: A Contingency and Configuration Approach. Academy of Management Proceedings.https://doi.org
2. Chang, A., & Jiao, J. (2026). Energy creates intelligence: The infrastructural geography of global AI. AI & Society. https://doi.org
3. Davis, J. P., & Li, J. B. (2024). Early Adoption of Generative AI by Global Business Leaders: Insights from an INSEAD Alumni Survey. arXiv. https://arxiv.org
4. d'Orgeval, A., Sheehan, S., Avenas, Q., Assoumou, E., & Sessa, V. (2026). Generative AI impact assessment through a life cycle analysis of multiple data centre typologies. Applied Energy, 406, 127288. https://doi.org
5. Shi, Y., Wu, T., Qin, C., & Liu, B. (2025). The value-creating potential of AI: A multi-dimensional analysis of effects and mechanisms. International Review of Financial Analysis, 108, 104694. https://doi.org

.    .    .