A system's potential refers to its measurability and achievable measurement depth, while realizing that potential involves determining which aspects of the measurement results the system must send to the processor for further processing. Finally, sensor fusion refers to understanding how to combine measurements from different sensors in the correct proportions; the algorithm's intelligence and reasoning potential are key areas we explore. By enhancing sensor fusion through feedback loops, the algorithm can verify and correct its own logic, an essential component of machine learning.
These three attributes are crucial for understanding the depth of artificial intelligence, especially its unique capabilities. The more fundamental elements we uncover and calibrate, the better AI algorithms will perform in the long run. Having introduced the three areas we will explore, we will now delve into the first aspect—measuring depth—and its importance in building a robust foundation for high-performance AI algorithms.
Measuring depth
Metrology studies the science of measurement. Measurement depth plays a crucial role in building robust algorithms. The Gagemaker's rule (the 10:1 rule) states that the accuracy of a measuring instrument or device must be 10 times greater than that of the object being measured. Measurement depth is so important because it determines the level of accuracy achievable, limiting the algorithm's maximum potential. Therefore, the higher the accuracy when performing any given measurement, the greater the potential of the artificial intelligence algorithm.
Metrology focuses on gaining a deep understanding of a specific measurement. This measurement might be very simple and straightforward, such as voltage, ground, or temperature, or involve multiple modes like implementing aircraft control surfaces; it might also be very complex, like maximizing throughput on a production assembly line. Whether measuring a single parameter or multiple parameters, the depth of measurement determines the level of programmability we can achieve. For example, with a 3V voltage system, a measurement accuracy of only 1/10 V is incomparable to a measurement accuracy of 1/1000 V in terms of insight. Depending on what kind of system is being powered, the extra accuracy might be crucial for battery life, or it might be completely redundant. Fully realizing the potential of algorithms requires matching the depth of end-to-end measurement needs with the required depth. This is true regardless of what is being measured, even for potentially less intuitive data systems. Let's look at an example.
How to optimize measurement
An enterprise IT stack is a complex network of interconnected data systems, each needing to exchange information to coordinate organizational operations. These technology stacks comprise a suite of software, such as CRM, ERP, databases, order fulfillment, and more, each with its own unique data formats and custom application programming interfaces (APIs). Salesforce data shows that companies typically have over 900 applications in their technology stacks, many of which are cloud-based, and updates to these applications can have cascading effects. Discovering and isolating problems is like finding a needle in a haystack; optimizing the performance of multiple cross-application systems is even more challenging.
Within an enterprise's technology stack, each application software has a different responsible department, such as finance, human resources, sales, marketing, or supply chain. IT prioritizes the needs of the main organization. Each enterprise has its own customized workflows and integrates numerous applications and backend systems. The user's journey through the software involves various paths, with single linear journeys being very rare. Therefore, even if two enterprises use the same applications in their technology stacks, their mapping of all exchange points and end-to-end operational verification methods will be completely different. Applications requiring artificial intelligence have thus emerged. In this context, measurement locations could be data input points between systems, or data exchange and display points within systems.
To understand how AI algorithms operate in such a system, we first need to understand how they measure data at various points in the following three key areas:
1. Evaluate the user interaction methods with the application software, regardless of the operating system used. In some cases, when key operations are required, Robotic Process Automation (RPA) may also be involved.
2. Evaluate the data exchange between systems within a complex technology stack and the application programming interface commands that connect these systems to ensure they function correctly.
3. Evaluate screen information across all platforms (including desktop and mobile devices), such as images, text, and logos, to understand how they are presented.
Regardless of the operating system, software version, device, or interface mechanism used, evaluating measurement effectiveness must begin with measurement capabilities. The more situations that artificial intelligence cannot measure, the smaller its impact on operation.
in conclusion
When assessing the potential of something, we need to start with the basics. The foundation of an AI system is its measurement capability. The more conditions it can measure, the greater its potential impact. We need to understand what it can measure, and more importantly, what it cannot. The potential of an AI algorithm is limited by its sensing capabilities. Lord Kelvin's famous saying remains relevant today—"What cannot be measured cannot be improved." To understand the true capabilities of AI, be sure to begin by analyzing the breadth and depth of its measurements.
