At the top of the list of 2019 buzzwords is AI (artificial Intelligence). A concept that’s been around for decades has taken off, as the computer power to actually do AI has become available in usable forms—no mainframe needed. But computer power is only the start, the programming has to be in place to do actual work, to perform viable functions, and to make a difference. To do that requires a way to train the AI so it can perform the functions it was designed to do.
Recurrent Dynamics, a Toronto-based AI start-up, has achieved a computational breakthrough that makes it viable to train artificial intelligence continuously, at the edge, on devices such as cellphones, tablets, and any of the 25 billion connected devices worldwide.
Advancement in artificial intelligence has been hindered by the standard approach to training. This involves acquisition of data, and training on centralized cloud infrastructure, typically provided by Amazon Web Services, Google Compute Cloud, or Microsoft Azure. The huge upfront cost and complexity of this has meant that only very well-funded companies can advance AI capabilities.
Recurrent Dynamics intends to change that by enabling AI to be trained on small devices such as cellphones and tablets rather than expensive cloud infrastructure. If the AI model runs on a device, it can now be trained on the device.
What this will mean, in the short term, is that AI will become significantly more capable, in less time due to dramatically faster prototyping and larger scale training. In addition, there will be a growth in practical applications of AI because the new paradigm of training at the edge avoids the huge upfront costs of centralized training in the cloud. Millions more developers can now participate in advancing AI solutions. Because training can be coordinated between devices using the IoT (Internet of Things), the cloud infrastructure will have a diminished role.
One of the early applications of AI in the construction industry is for training workers and improving their skills. To do that, AI will require technically trained programmers and others familiar with the robotic processes that are the basic building blocks of AI. To that end, Automation Anywhere, and its education and certification division, AAU (Automation Anywhere University), has trained more than 350,000 developers, business analysts, partners, and students in RPA (robotic process automation). The program is expanding with more than 65 authorized training partners, 300+ academic institutions, continuing education programs, and professional associations.
According to the McKinsey Global Institute, automation and advances in artificial intelligence are anticipated to encourage as many as 375 million workers, or roughly 14% of the global workforce to reskill themselves by 2030. RPA ranks as the third fastest growing technology for reskilling in the U.S. freelance job market, according to a survey by Upwork.
The Recurrent Dynamics program provides role-based learning and courses ranging from beginner to expert level and offers one of the most comprehensive RPA training. Within the next five years, the company anticipates certifying more than one million individuals for the future of work and is continuing to build partnerships with training partners, professional communities, and academic institutions globally.
While AI for training workers moves forward, AI for lower education levels is being implemented at a brisk pace worldwide. For example, this year the Chinese Ministry of Education and State Council issued a Work Focus Mandate to modernize China’s 400,000 secondary school campuses with the latest AI technologies for campus safety, student attendance, and interactive, performance-driven learning.
Remark Holdings, Inc. has deployed KanKan AI’s Smart Campus solution in the Hangzhou Primary School System in Hangzhou, China. KanKan AI developed its Smart Campus Solution in response to such mandate. As implemented in Hangzhou, the Smart Campus Solution is comprised of integrated hardware and software and provides a professional channel subscription option on WeChat for parents wanting to view real-time video of their child.
The Smart Campus Solution uses facial recognition and object recognition technology to automate student check-in and check-out at the school’s entrance and exit points, control access to dormitories, laboratories and libraries, alert school administrators of unauthorized persons who have trespassed on the school campus, ensure students are released only to parents or other pre-approved persons, and monitor for unauthorized objects, such as weapons, brought into school buildings.
Cameras installed at school entrances, in passageways, and in campus buildings allow the system to monitor campus locations without using an RFID (radio-frequency identification)-based system, turnstiles, or other less-effective or inconvenient methods.
By combining AI, facial recognition, and IoT technologies, education, both in standard academic and special training applications, is moving rapidly into the next stage. Bits and pieces of the technologies can be applied randomly but when combined, they are a formidable way to ensure the leaders, workers, and trainers for tomorrow are being educated today.
But education comes in many forms. Training workers to do a better job is important but training them to stay alive is just as important. Construction depends on materials and materials depend on delivery, on time, on location. Fleets of trucks moving along the nation’s highways are the most common delivery method for construction supplies. Now AI is being used to prevent accidents and train drivers to be aware of their limits.
Lytx, a provider of machine vision and artificial intelligence-powered video telematics solutions for commercial and public sector fleets, has data showing drowsy driving and falling asleep behind the wheel are dramatically declining among commercial drivers who use the Lytx Driver Safety Program.
Lytx data reflects a 39% reduction in drowsy driving events among Lytx clients from June 2018 to June 2019, and a 66% reduction in drivers falling asleep behind the wheel. The data also shows morning hours between 5-8 had the highest concentration of events recorded for drowsy driving, falling asleep behind the wheel, and collisions, while 6-9 pm had the fewest instances.
A Lytx client, Hogan Transportation, is one of the nation’s largest and fastest growing transportation service providers and is committed to raising awareness of fatigue-related incidents in its fleet. The Lytx Driver Safety Program provides the company with an accurate picture of what is happening in its fleet and delivers a level of visibility and support to help Hogan make meaningful changes and improvements to reduce driver fatigue. Having hard data and video evidence around driver fatigue and its correlation to dangerous driving has helped Hogan prioritize its commitment to decreasing fatigue amongst drivers, and it is seeing positive results.
Lytx’s machine vision and artificial intelligence algorithms are constantly advancing, increasing its ability to identify and alert for risky driving behaviors. The company released its first solution for distracted driving in 2015, and today has more than 200,000 devices in the market, with an additional 10,000 added each month.
Practical applications of AI are starting to show results, in construction and its supplier base. Leading-edge companies are testing these waters and will soon be profiting from their experiments.
Want to tweet about this article? Use hashtags #construction #IoT #infrastructure #AI #artificialintelligence #machinelearning #bigdata #digitaltransformation #cybersecurity #blockchain #FutureofWork