Artificial Intelligence, now a popular topic, has its roots in the 20th century, specifically between the 1950s and 1960s, when efforts were made to model and develop mathematical and computational techniques with cognitive abilities similar to our own, capable of learning from data and applying that knowledge to solve simple problems. Just as synapses naturally occur in our brains to enable reasoning, the desire to model human reasoning, memory, and intelligence—and to create an artificial brain capable of synapses—has always been a goal for scientists, researchers, mathematicians, and neurophysiologists. After all, having an intelligent machine as an ally could, in theory, solve numerous challenges that human abilities alone could not.
In the early 2000s, Artificial Intelligence was still a distant concept, mostly discussed in academic research circles or seen as a topic for science fiction, largely due to the limitations of processing power and the prohibitive costs of high-performance computers capable of running machine learning algorithms with massive data sets and complex models. Over the next 20 years, significant advances in accessibility and reduced processing costs—especially with the popularization of cloud computing—made it feasible to apply more complex machine learning algorithms, such as Deep Learning, which in turn has increasingly democratized the benefits of Artificial Intelligence.
Today, AI is so prevalent in popular culture that it has evolved from the SkyNet of the Hollywood movie Terminator to being seen as a job killer, as exemplified by the widespread use of OpenAI's ChatGPT, which sends chills down the spine of some professions. However, successful AI implementation faces challenges ranging from proper, bias-free training to overcoming the cultural barriers of organizations that prefer not to evolve processes they perceive as functioning well, even if those processes lack efficiency or productivity.
In the realm of asset management, a notable success story is the application of AI and Computer Vision for mapping electrical energy assets, a project being implemented in various energy distribution companies in Brazil. The solution involves equipment comprising high-definition cameras, GPS, and other hardware mounted on top of vehicles. These vehicles drive through city streets, capturing images of the entire energy distribution network, including poles, transformers, cables, and lamps. AI-powered software then automatically processes these images and identifies all the assets in the electrical grid, down to the wattage of each streetlight in an entire city. The final result of this mapping is an inventory that is completed up to 100 times faster and more accurately than traditional methods for an energy distribution company’s assets.
However, not every problem requires an AI solution; traditional statistical models can—and should—be applied with good results. There is a buzz that companies need to apply Artificial Intelligence as if it were a requirement to avoid being labeled as obsolete, as if AI were the Holy Grail that will either save or threaten humanity. In simpler cases, having efficient processes and clear performance indicators is already a significant improvement. For AI to truly add value, some basic requirements need to be mapped, such as clearly defined objectives, the technology(ies) to be used, the impact on the current process, and the return on investment. Often, decisions to invest in an AI project are driven by innovation, but with questionable economic returns.
It is also important to mention the commitment of senior leadership in organizations seeking to harness the value-generating potential of AI, clearly explaining the impact on processes and how the new configuration will affect both the processes and the people involved, to avoid generating insecurity about being replaced by AI.
Paraphrasing Physics and its concept of inertia, which states that any object at rest or moving in a straight line will remain at rest or continue its motion until acted upon by an external force, it is easier for a company to do nothing or maintain a modus operandi that has been in place for years than to pursue evolution (or revolution) through AI implementation. For Artificial Intelligence to generate value, a strategic decision, as well as a company-wide movement with a well-crafted change management plan, is desirable. After all, in this case, innovation means making processes more efficient and generating value for the market.
André Sih, Managing Partner of Fu2re Smart Solutions, and Jardel Pinto, Maintenance Strategy and Equipment Reliability Manager at MODEC & VP of Value Assurance at Shape Digital.
Comments