When something is improved, it ends up being used even more: The Jevons paradox and its impact on AI
In relation to the previous article on AI and Moravec’s paradox, another antinomy that has been mentioned in recent months following the emergence of DeepSeek (new Chinese AI), with its ambitious proposal, and the announcements about the Stargate project (an initiative that focuses on the development of artificial intelligence infrastructure in the United States), is the Jevons paradox, recently mentioned by Satya Nadella (CEO of Microsoft) in one of his messages through X on social networks.
This paradox dates back to 1865, when British economist William Jevons made an almost intuitive discovery and substantiated it in his book “The Coal Question”: when steam engines became more efficient and consumed less coal per unit of energy, total coal consumption didn’t fall, but rather soared. Why? Because efficiency reduced costs and expanded its adoption across multiple industries. This is known as Jevons’ paradox: rather than decreasing demand for a resource, efficiency improvements can actually boost it even further.
In 2025, DeepSeek’s new AI model raised concerns about a potential drop in demand for high-performance chips and cloud computing. But history suggests otherwise: efficiency gains don’t eliminate demand for raw materials; they accelerate them. In other words, when something is improved, it ends up being used even more.
To better understand this paradox, we can relate it to an everyday example: low-energy light bulbs, specifically LEDs (Light Emitting Diodes). These bulbs are widely known for their energy efficiency and longer lifespan compared to traditional bulbs. Our logic suggests that, since LEDs consume less energy, we should see a reduction in overall energy consumption. However, although the electricity bill may decrease initially, over time it tends to stabilize or even increase. This is because, perceiving that LEDs consume little energy, we stop using them rationally and increase their use indiscriminately.
Therefore, the reduction in costs associated with language models (LLM) and other emerging artificial intelligence models, such as DeepSeek, should not be interpreted as a decrease in the use of underlying computing resources, such as NVIDIA GPUs. On the contrary, companies understand that the evolution of AI will entail an increase in demand for these resources. The CEOs of leading AI companies have repeatedly reiterated that the cost of AI will be reduced to negligible levels, enabling its omnipresence. The real key lies in access to processing resources and, to a large extent, energy. This perspective justifies OpenAI’s announcement of its participation in the Stargate project, as they need to invest in computing infrastructure and ensure preferential access to a large processing center, without relying exclusively on commercial alliances. This need for more computing is projected for Latin America, hence the importance of data centers like those Cirion Technologies is building and access to clean energy.
In Latin America, 80% of companies have already had contact with artificial intelligence, and large companies plan to allocate more than 15% of their budget to AI technology in the coming years, according to leading market analysts.
Because AI is essential in Contact Centers and their growing participation.
To embark on the journey toward transformation through artificial intelligence, it’s crucial to understand how customer, employee, and business demands are evolving, and what these trends mean for customer service operations. Artificial intelligence in contact centers not only represents an exciting opportunity but also essential for brands looking to solve current pain points and improve future experiences. With customer (CX) and employee (EX) expectations constantly growing, the impact of these issues will only intensify in 2025. Organizations have long faced obstacles that have prevented them from creating seamless, personalized, predictive, and proactive experiences, from unintegrated systems to inefficient processes and insufficient data. Providing outstanding customer service has always been a challenge. AI-powered virtual assistants, like those offered by Cirion Technologies, are extremely useful for handling frequent customer inquiries, freeing up human agents to focus on solving more complex and specific problems. However, as many CX leaders recognize, the true potential of AI in customer experience goes far beyond chatbots and their ability to answer routine questions. AI has the power to transform CX through predictive analytics, advanced personalization, and process automation, which can lead to significant improvements in customer satisfaction and loyalty.
Contact Center Transformation: From Costs to Data Value
Empowering business with actionable insights represents the surest way for a contact center to transform from a cost center into a value center. Modern AI analytics solutions are essential for capturing, unifying, analyzing, and democratizing such intelligence. A vast majority of contact center leaders aspire to make customer insights a more widely shared resource across the enterprise by 2025. Beyond fostering personalized service experiences, this data can drive overall business success, from more effective marketing campaigns to more user-friendly product design and smarter branding. As the gateway between the business and its audience, the contact center gains access to unprecedented intelligence about customer demands, expectations, behaviors, and sentiments.
What happens in the contact center should NOT stay in the contact center… until the next delivery.
Author:
Lic. Fernando Riedel
Voice and Collaboration Specialist
Cirion Technologies
fernando.riedel@ciriontechnologies.com
