QpiAI Founder Argues for Shift to Intelligence Modelling Over Traditional AI

Nagendra Nagaraja, founder and CEO of QpiAI, stated that current artificial intelligence models are fundamentally limited because they rely on existing human knowledge rather than generating new insights. Speaking at the TechSparks event, Nagaraja proposed a shift toward what he terms “intelligence modelling” to overcome the constraints of modern computing.

Key Takeaways

• Intelligence definition: Nagaraja defines intelligence as the capacity to discover new information rather than automating known tasks.
• Silicon limitations: Classical microprocessors and GPUs face physical constraints, including heat generation and power grid strain, as they attempt to scale.
• Data saturation: Large language models are experiencing diminishing returns because they rely solely on digitized human text.
• Quantum potential: Quantum computing, utilizing qubits, can process multi-dimensional data from nature that current classical systems cannot handle.
• Industry focus: Energy and compute represent the two foundational industries that underpin all other economic sectors.

The Limits of Classical Computing

Nagaraja argued that the current boom in AI is constrained by the reliance on classical microprocessors. These chips, which operate on binary bits, face significant physical barriers as performance requirements increase. Electrical resistance creates heat management issues, forcing companies to adopt complex strategies like 3D chip stacking. Furthermore, the reliance on GPUs has created an economic dependency, which Nagaraja described as an “NVIDIA tax.”

The CEO noted that large language models are reaching a point of data saturation. Because these models are trained almost exclusively on human-generated text, they represent only a tiny fraction of available information in the universe. Most data, according to Nagaraja, exists in unquantized states within molecules, DNA, and crystalline materials, which classical hardware cannot process effectively.