Reevaluating AI development: Maximising intelligence against practical uses

The Artificial Intelligence (AI) Action Summit in Paris was a clear demonstration of France’s commitment to AI as a national strategy. Since 2018, the country has directed substantial resources toward AI research, talent cultivation, and infrastructure development. The event, which gathered global technologists and business leaders, saw the announcement of hundreds of billions of euros in AI and related infrastructure investments, underscoring Europe’s growing role in shaping the future of artificial intelligence.

The strategic foresight of President Emmanuel Macron’s government in fostering an innovative AI ecosystem is evident. France has positioned itself as a hub for AI research and start-ups, attracting global investors and industry leaders. This aligns with a broader European ambition to play a crucial role in AI development, ensuring that the continent is not left behind in a field dominated by the United States and China.

The recent rise of DeepSeek has sparked intense discussions within the AI community. The Chinese company’s ability to launch open-source models and publish research detailing engineering innovations has disrupted conventional approaches to AI training and inference. DeepSeek’s methodology suggests that large-language model training can be achieved at significantly lower costs than brute-force methods, challenging the traditional notion that AI development is exclusively for companies with deep financial resources.

Historically, the AI race has been framed as a competition to develop the most intelligent frontier models, with companies investing billions in computing infrastructure to gain an edge. Benchmark comparisons focus on capabilities in mathematics, coding, and reasoning, reinforcing the belief that AI supremacy is determined by computational power. However, this perspective limits the broader impact of AI, as only a handful of resource-rich entities can afford to participate in this high-stakes race.

A human analogy helps illustrate this imbalance. Training frontier models is akin to the expensive and time-consuming process of educating children to become Nobel Prize winners. Only a select few parents can afford the best resources to develop such exceptional individuals. But in the real world, societies do not function solely on the contributions of Nobel laureates. Instead, they thrive when a diverse population is equipped with practical skills and knowledge to contribute meaningfully. This shift in thinking is crucial for the AI industry.

The DeepSeek moment has significant implications. Firstly, it compels us to rethink the ultimate goal of AI development. The relentless pursuit of the most advanced AI models, independent of real-world applications, is unsustainable. Innovations that enhance efficiency in training and operation indicate that the pursuit of intelligence in isolation will eventually yield diminishing returns. The future lies in applying AI to drive economic impact, accelerating the proliferation of task-specific and specialist AI models. The laws of economics dictate that capital will flow toward commercialized applications that generate revenue, modifying the singular focus on scaling law the principle that increasing model parameters and compute resources leads to stronger AI intelligence.

Secondly, the rise of open-source models will democratise AI, allowing smaller companies to participate in the development and deployment of AI applications. Open-source AI provides accessibility, enabling developers to deploy models on a variety of infrastructures, from data centers to personal laptops. This breaks the monopoly of tech giants and fosters a more inclusive AI landscape. Smaller parameter models will proliferate, allowing AI to be tailored to specific business needs without requiring excessive computational power.

For instance, in e-commerce, an AI-driven shopping assistant does not need to be powered by a trillion-parameter model. A human store assistant does not require a PhD in mathematics to provide excellent customer service; instead, emotional intelligence and domain knowledge suffice. This illustrates how AI applications can be effective without excessive complexity. The future of AI development will prioritise usability and economic viability over raw intelligence.

This shift in perspective leads to a fundamental conclusion: AI’s value lies in its practicality rather than its theoretical intelligence. Just as society benefits more from well-rounded individuals contributing across various industries rather than a handful of Nobel laureates, the AI industry will gain more from the widespread adoption of useful and productive AI applications rather than an exclusive focus on frontier models. Economic incentives will guide companies to innovate at lower costs, making AI more accessible and cost-effective.

Entrepreneurs and businesses will increasingly focus on specialised AI models that drive economic impact. This shift is beneficial for innovation, as it reduces barriers to entry and encourages diverse participation in AI development. When AI becomes more affordable, smaller players can compete, fostering a more dynamic ecosystem. Consumers, in turn, will benefit from the widespread availability of AI-powered applications that enhance everyday experiences.

The AI Action Summit in Paris highlighted Europe’s aspirations in the AI landscape, and the discussions reinforced the idea that AI should be developed for widespread benefit rather than limited to elite competition. The DeepSeek phenomenon has sparked necessary conversations about the future direction of AI, emphasizing the importance of efficiency, accessibility, and real-world applicability. As the industry moves forward, success will be defined not by the smartest AI models but by those that create tangible economic and social value.

India has also made significant strides in AI development, positioning itself as a key player in the global AI ecosystem. The Indian government has launched several initiatives, such as the National AI Strategy and the establishment of the National AI Portal, to drive innovation and research. Additionally, India’s strong IT sector, coupled with its vast talent pool, makes it well-suited for AI-driven economic growth. The country is investing in AI-driven solutions in healthcare, agriculture, and governance, ensuring that AI development benefits a broader section of society. With growing collaborations between Indian start-ups, multinational corporations, and academic institutions, India’s AI landscape is poised for significant expansion, further contributing to the democratization of AI on a global scale.