What are the pros and cons of integrating artificial intelligence applications into municipal operations?
Almost every application uses machine learning in one way or another, from Amazon recommendations to autopilot technology. Everyone seems to love these advancements in daily activities, but what can you expect when integrating artificial intelligence applications into municipal operations?
The current processes used to ensure the integrity of sewer networks are resource-intensive. Artificial intelligence can optimize this process, but it is important to have realistic expectations when integrating applications into operations. To maximize the benefits of leveraging machine learning practitioners, we need to be aware of the limitations of such applications.
Pros:
Machine learning applications can do everything from recognizing patterns, predicting events, and understanding speech. With the powerful applications of machine learning spanning industries, Amb.AI looks into how this can help infrastructure maintenance operations, particularly storm and sanitary sewers.
Faster, cheaper, and more accurate detections
Detecting defects is probably the most obvious area where machine learning algorithms can lend a very capable hand. Applications utilizing computer vision algorithms can rapidly plow through hundreds of hours of video footage, such as sewer CCTV videos, to automatically identify areas of interest or concern and flag them. This speed can cut video inspection time by up to 90 percent, liberating technicians from the burden of spending hours staring into endless tunnels on screens. This significant reduction in inspection time can be easily translated into a reduction in cost, both directly and indirectly. The direct cost savings are associated with decreased resources required to perform the inspection. At the same time, indirect saving spans several aspects, such as reducing the need to rewatch, increasing job satisfaction and productivity, and reducing the potential risks associated with errors. And while computers can make errors, they are less susceptible to making mistakes out of boredom, fatigue, or distractions. It is possible to run a computer-vision-powered system 24/7 and get the same level of accuracy.
Time-based detection
It is no surprise that defects develop in sewer pipelines as we operate them. As a result, it is important for all municipalities to regularly assess their condition, as understanding the rate of deterioration is essential for maintenance planning efforts. By analyzing the videos of the same pipe section over the years, the algorithm we developed at Amb. AI can determine which parts have developed new defects, which defects are in stable conditions, and which are actively deteriorating. Better still, the algorithm can quantify the deterioration.
Failure Pattern Recognition
The determination rate, along with other operation information cross-referenced with historical data, can help reveal valuable insights into the failure patterns in a given system. This allows for the development of accurate models that can simulate and predict the behavior of the sewer network over time to keep sewage systems where they belong - out of sight and out of mind.
Increased planning and operations efficiency
When models can predict the structural aging of sewer pipes, planning operations and maintenance can become much more efficient. Data from predictive models and another class of algorithms called optimization algorithms can result in maintenance operations being optimized and streamlined to reduce waste and ensure that available resources are being used in the most efficient manner possible. This helps ensure municipalities are being the best stewards of public dollars.
Cons:
Artificial Intelligence has been through waves of popularity and acceptance throughout the last 10 years. While buzz can come and go, expectations remain high with AI operations. It is important to temper these expectations and plan for developing an environment where we optimize the integration of machine learning applications properly into our work ecosystem.
Error risk
Like people, computers have limitations. It is unrealistic to think that computers make no mistakes. Many factors can affect the judgment of computers. For instance, in the context of sewer condition assessment using CCTV data, small training datasets, poor video quality, and completely unfamiliar data can reduce the accuracy. AI-powered applications can do what they are programmed to do.
No abstract-thinking ability
AI cannot do everything that humans, as complex abstract thinkers, can do. AI-powered applications are still in their infancy at large and are not able to entirely replace humans. Appreciating this fact helps temper expectations and reduces the backlash when the application does not perform to unrealistic standards. A lot of negative narratives around AI come from the fear of machines replacing people. At Amb.AI, we see artificial intelligence as an empowerment tool that liberates people from the shackles of dull tasks and allows them to focus on creative and abstract tasks. AI is best applied in tandem with people, not independent of or instead of them.
Algorithm training time
In the same way, you wouldn’t expect your new hires to perform 100% as planned the first day they start; you should also not expect AI applications to. Sewer networks are extremely diverse on many levels, such as types of construction materials, operation conditions, intended use, and so on. Training and familiarizing the algorithm with your local operational conditions before putting it into service are advised to avoid the frustration associated with lower-than-promised performance.
While keeping the shortcomings of the technology in mind utilizing artificial intelligence can be a great aid to increase the efficiency of processes while cutting on resource use.
Want to learn more?
Email our CEO directly to hear it best from him - samer.buhamdan@amb-ai.com.
