We are on the brink of an AI revolution and arguably in the middle of the data revolution. Many might compare this to the Industrial Revolution, which Leeds was at the heart of, and through which grew massively. The Industrial Revolution enabled mass production and manufacturing of physical goods and materials with new machinery at its core. As we are set to embrace the AI and data revolution, is the output pure data, or can we hope to see this revolution bridge physical worlds with data?
We are reminded of our slightly dry ICT classes with one of the first concepts, talking about Input and Output devices. The mouse and keyboard revolutionised how we interact with ‘compute’. We are now not just limited to these, modern robotics that once were based on primitive motors and sensors.
We have seen many robotics competitions in universities and countries such as Japan. There is even a ‘Robotics Olympics‘ which was first held in 1990. Many teams would use ingenious ways to manipulate robots to achieve various skills. Interestingly one team from England built a wall-climbing robot that was disqualified for veering out of lane and demonstrating inappropriate behaviour in front of children. (Photo Credit: Petermowforth – Own work, CC BY-SA 4.0)
We have come a long way and now we have so much in our world that were once mere input and output devices are now connected to the internet.
Being connected to the internet alone does not make a device ‘smart’. We now have the capabilities to learn, make decisions, and probably to a limited degree problem-solve using AI technologies.
When asking ChatGPT what the difference is between smart and reactive, it suggested the term “Smart” usually refers to systems, devices, or approaches that possess advanced capabilities of decision-making, problem-solving, and learning. “Reactive” refers to a more straightforward response mechanism that reacts to stimuli or changes in the environment without necessarily learning from past experiences or making complex decisions. This is where Data and AI come in.
At BJSS, we are a Databricks partner At one of their keynotes recently they suggested: “The competitive advantage of ML and AI will be a company’s proprietary first-party data“. What does this mean? We have sometimes jumped to second base; we have spent time focusing on IoT-connected devices to control output. However, if can use them first to primarily measure how machinery operates, and human interaction is achieved, then we have access to training data.
Meaningful, representative training data is essential to help create models that can playback operations and outcomes to enable data to bridge back to real-life ‘physical’ scenarios and interfaces. We must be very careful with safety, ethics, personal and corporate responsibility whether we should be creating models in this way, or how we have created them.
There may be contextual information that is not collectible easily as data that might affect any given model’s accuracy. Making sure datasets, are complete, and in the right, consistent format can be a challenge. This is especially true for IoT devices that can often drop connections for a whole array of reasons, including interference, Wi-Fi connectivity, and even other mundane boring problems such as batteries needing replacing.
Making meaningful real-world models requires getting the right data in the right hands. Data Governance used to have connotations of ‘Computer Says NO!’, this is no longer the case. We can enable the right data experts to have the data that they need to make models that they need. So what are the amazing applications that we are seeing now and what more could we see in the future?
At BJSS we have helped Care Fertility using imagery over time of the formulation of cells, to improve the process of embryo selection during IVF. We have also helped National Highways become a data-driven organisation, providing safe, reliable, and better journeys for their customers. With applications including using AI to predict and to help respond to road incidents and urgent repair needs. But there are thousands of applications including managing food manufacturing processes, and anomaly detection in machinery that can visualize when there might be impairments to operations that need to be resolved.
The automatic restacking of shelves and retail solutions that now are enabling the next generation of un-staffed convenience stores. Many smart cities are using digital twins to bridge the physical and virtual worlds, real time simulations bring real-world scenarios to life, helping to optimise planning, operations, finance, and emissions reduction.
IoT-enabled components are very inexpensive, and data platforms and models are so accessible, now is a great time to dive in and start experimenting. I am sure that many product folks will be thinking about how your organisation might want to bridge the digital world and physical world using data and AI and, in the process, bring many improvements to the manual ways we have been working.