Founder & CEO at SmartClick.
People used to associate the term “artificial intelligence” with images of science fiction without even thinking that one day it could come to life. The concept of AI has been elevated from the realm of sci-fi to reality. There are many ways AI is used behind the scenes in everyday life that we don’t even realize. These solutions are sometimes inaccessible, and they still need extensive training and expertise to become commonplace.
As an entrepreneurial executive with more than 20 years of work experience in the software development and technology industry, and as the founder and CEO of one of the leading AI companies in Armenia, which builds deep tech innovations based on artificial intelligence and machine learning, this is a topic I deal with on a daily basis.
Let’s go over a few examples of hidden AI I’ve seen during my time in this field.
There are many ways AI is efficiently used in the medical industry. One of the recent advances involved applying AI for drug discovery. This has been achieved by MIT. Using an AI model that had been trained with a broad range of medication and disease datasets, the researchers identified a new type of antibiotic called halicin that can kill many species of bacteria. The team performed classification modeling through a deep neural network and used a computer model that’s capable of screening over 100 million chemical compounds within a few days.
This is surely a great achievement in drug discovery; however, these models are often black boxes because they learn to predict molecular function with no assumptions about how drugs work and without the labeling of chemical groups. They learn new patterns unknown to human experts, which means they need further study, testing and approval from the FDA to be brought into clinical trials.
From this example, businesses should note that all AI, no matter its application, is most useful when it is explainable. If you can track how your system comes to a conclusion, you are that much more prepared to understand and implement the information it provides.
Human Muscle Simulation
Another interesting example of AI application in healthcare that’s still not developed for medical use is a muscle simulation experiment conducted by Seoul National University. The idea was to build a model that could reproduce realistic human movements. The researchers created a musculoskeletal model based on deep reinforcement learning, demonstrating predictive simulation of muscle conditions and human movements in case of muscle weakness, bone deformity and use of a prosthesis. Such simulations could be useful in studying human physiology from a new point of view and developing new ways of simulating human movements.
This could be a valuable tool for understanding human movement before and after surgery. Thus, surgery specifications could be adjusted, and potential issues could be foreseen. Simulations could also become a viable approach for determining how the muscles involved in physical activity could be used for therapy and sports. Furthermore, simulation technology could be applied to identify potential issues for a person with gait defects.
AI is also involved in different aspects of physics. With the ability to handle millions of particles within a virtually unbounded simulation domain, AI enables large-scale fluid simulations with advanced effects such as honey coiling and magnetic fluids. Tsunami simulations are also developed through the use of complementary AI models to help scientists predict the occurrence of tsunamis and how they’ll most likely behave. It’s also possible to use AI for creating smoke simulations to produce more accurate predictions about how smoke could move in case of fires to help minimize the effects of such hazards.
AI technology has great potential to become a tool for development and progress in science and other fields. The technology can be used to find new solutions by simulating scenarios instead of expensive and time-consuming testing.
For the above cases, still, the lack of AI transparency could make its application rather complicated, as the way AI technology makes decisions can’t be explained entirely, and the FDA needs to understand the approaches in medical and scientific fields before methods can be approved.
This is yet another reminder that any solution you implement in your business should be as explainable as possible. Being able to trace your results will prove invaluable in making decisions down the road.
Product Formula Making
Machine learning algorithms can be used to bring new approaches in determining formulas for products, such as fragrance making. A leading German producer of fragrances, Symrise, and the tech giant IBM efficiently use Philyra machine learning algorithm that studies the existing fragrance formulas and compares the ingredients to customer data, such as geography, customer age and preferences. Then it creates new formulas targeted to specific market segments.
Similarly, AI is integrated into beer-making processes to enhance production and improve the quality of the beer. IntelligentX, which has introduced the world’s first beer brewed by AI, receives customer feedback based on machine learning algorithms to create unique recipes tuned to customers’ tastes.
These solutions save significant resources; however, human senses — smell and taste, in this case — remain irreplaceable, which means the human touch is an important factor to depend on. There is another lesson to be learned here for businesses: Your AI should never be used as a means to replace all human contact with a process. You will still need human intuition and emotion to make successful choices in your systems.
The advancements of AI technology have been reshaping many industries. To thrive in this new environment of digital transformation and automation and handle these emerging opportunities, forward-looking corporations must invest in AI that is not only explainable, but that is also not a complete replacement for their existing workers. Marrying these two tips will help you successfully act on any insights your solution provides.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?