The robotics industry stands at a transformative inflection point, where machines are evolving from rigid, pre-programmed tools into adaptive systems enabled by liquid neural networks.

This shift represents a fundamental breakthrough in solving what has long been considered the “common sense gap” – the challenge of creating robots that truly understand how the physical world works, Daniela Rus, professor at MIT, director of the CSAIL Research Center, and ROBO Global Indexes strategic advisor, explained during a recent webcast. 

Key Takeaways

• Liquid neural networks offer a more adaptive alternative to traditional, static AI models for physical applications.
• New architectures achieve steering tasks with 19 neurons compared to the 100,000 required by traditional systems.
• Diversified ETFs like ROBO and THNQ allow for exposure to this technological evolution.

Solving the Common Sense Gap in Physical Robotics

Traditional transformer-based AI models, while revolutionary for digital applications, face critical limitations in physical environments. These systems rely heavily on statistical correlations and struggle with the spatial-temporal reasoning essential for real-world interactions, according to Rus. 

Critically, they often lack basic common sense understanding. Rus cited an example of how a humanoid robot could effectively water a plant. However, it attempted to water a person’s expensive shoes after being asked to “water my friend,” having learned to water plants but not understanding the distinction.

The emergence of liquid neural networks represents a paradigm shift in addressing these challenges. Unlike conventional AI that remains frozen after training, these architectures continue to adapt and improve based on new inputs. This makes them ideally suited for dynamic physical environments. 

The efficiency gains are remarkable: Where traditional systems require 100,000 neurons and half a million parameters for vehicle steering tasks, liquid networks achieve the same functionality with just 19 neurons and 2,000 parameters. According to Rus, these networks deliver hundreds of times more energy efficiency.

This technological breakthrough extends beyond mere computational efficiency. Liquid networks can perform reliable spatial-temporal correlations, causal reasoning, and learn transferable skills rather than context-dependent behaviors. This enables robots to understand physics-based interactions, such as recognizing the different force requirements when pushing a suitcase on smooth floors versus uphill sidewalks, Rus said.

See more: Purpose-Built Humanoids for Specific Jobs

Investment Implications of the Next Wave of Robotics

As physical intelligence matures — driven by advancing data, safety, and hardware — the market is preparing for widespread deployment in manufacturing, healthcare, and logistics.

The convergence of these technologies suggests that investors who recognize this inflection point today may benefit from the next major wave of automation and intelligent systems adoption. Navigating that complexity is where structure matters.

Diversified ETFs offer a streamlined approach to a complex subsector. The ROBO Global Robotics and Automation Index ETF (ROBO) provides exposure to the global value chain of robotics, offering access to the hardware side of this theme. Additionally, for focusing specifically on the intelligence layer, the ROBO Global Artificial Intelligence ETF (THNQ) targets the infrastructure and software powering these autonomous systems. 

Looking for regular updates? Subscribe here for weekly insights on robotics, AI, and healthcare technology, delivered straight to your inbox.

For more news, information, and analysis, visit the Artificial Intelligence Content Hub. 

vettafi.com is owned by VettaFi LLC (“VettaFi”). VettaFi is the index provider for THNQ and ROBO, for which it receives an index licensing fee. However, THNQ and ROBO are not issued, sponsored, endorsed, or sold by VettaFi. VettaFi and its affiliates have no obligation or liability in connection with the issuance, administration, marketing, or trading of THNQ and ROBO.