Evolution of AI: From diagnosing diseases to building skyscrapers

Neural networks in medicine: New opportunities for diagnostics and treatment

Over the past five years, neural networks have become an important tool in the medical field — not only in research, but also in applied projects. Artificial intelligence is used to process and interpret medical data, facilitate clinical decision-making, assist in diagnostics, automate administrative tasks, predict disease outbreaks, and much more.

Thus, during this time, artificial intelligence has succeeded in analyzing visual medical information: MRI, ultrasound, CT, X-ray, fluorography, and ultrasound diagnostics. In 2018, the news that algorithms detect diseases on chest X-rays as well as radiologists (and sometimes better) became a scientific sensation. In 2024, such publications are coming out regularly and are not surprising.

Artificial intelligence has also found wide application in the discovery of new drugs. In five years, scientists have moved from theoretical calculations to applied projects — for example, in 2023, the company Exscientia began the world’s first tests of a molecule created using artificial intelligence on people. If everything goes well, this molecule will become a medicine for the treatment of certain types of cancer.

Molecule GTAEXS617 — developed by Exscientia (Photo: apeiron-bio.com)

The use of neural networks has become especially important in connection with the global outbreak of COVID-19. Artificial intelligence has proven effective in the early detection of COVID-19, predicting hospitalization in the intensive care unit and finding people at high risk of COVID-19 infection — these developments will be used to prevent other pandemics.

The effectiveness of neural networks in medicine is expected to continue to grow in the coming years. This can lead to a revolution in the way we diagnose and treat diseases.

Neural Networks in Energy: From Consumption Forecasting to Power Plant Optimization

Over the past five years, the global energy industry has faced new challenges that artificial intelligence will have to address. First and foremost, this is the fight against global warming and climate change.

Global climate models have given climate scientists the ability to understand what might happen in the future  —  both to the Earth as a whole and to specific regions. Climate modeling involves using data sets and complex calculations to represent the interactions between the main components of the climate system, namely the atmosphere, land surface, oceans, and sea ice.

This analysis helps us understand how to most efficiently use new energy sources, such as solar panels or wind turbines. Neural networks can be used to predict the specific properties of new materials from which these energy facilities will be built, to find patterns in the use and production of renewable energy, as well as to develop energy policies and optimize energy management.

For example, Finland has created VTT EnergyTeller, a service based on artificial intelligence that allows for more accurate forecasting of energy demand, including energy from “green” sources.

The use of AI is also widespread in the oil and gas sector  —  oil and gas production has become more difficult over the years. That is why new technologies and approaches to production were required. Neural networks make it possible to simplify exploration, develop new fields, and optimize hydrocarbon production.

One of the world leaders in oil and gas production, Shell, has long been using AI in its work, including in the repair of production equipment and the prevention of its breakdowns.

In 2022, AI helped solve a long-standing problem with Shell’s Perdido platform, one of the deepest floating oil and gas platforms on the planet (located in the Gulf of Mexico). Its pumps responsible for separating oil and gas periodically failed. To predict such breakdowns, engineers began using artificial intelligence: it looked for patterns in the data returned by Perdido pumps (in temperature, pressure, and the chemical composition of oil and gas separated by the pump) that would be harbingers of breakdowns.

In the end, the Shell team managed to cope with the task: in 70% of cases, two days before the breakdown, the technical data of the pump changed slightly, in addition, the composition of the oil and gas that it separated changed. This time is enough to debug the pumps and prevent serious problems. And therefore, to prevent downtime in the work of Perdido.

Perdido platform (Photo: Wikipedia)

Neural networks in industry: increasing efficiency and optimizing processes

The use of artificial intelligence is becoming a global trend that determines the competitiveness of modern production. Leading consulting companies such as Gartner and McKinsey & Company have been emphasizing the revolutionary nature of these technologies and the high return on investment in them for the past five years. The use of AI opens up new opportunities for industrial companies to increase productivity, reduce costs and achieve sustainable development goals.

Now, one of the main components of success in industry is the ability to produce personalized goods in small batches on special order, taking into account customer requirements. Artificial intelligence technologies make it possible to analyze large volumes of data on customer needs to determine product characteristics. Following this trend, in 2021, the Machina Labs startup in the United States began to manufacture any industrial parts to order, and Nike created the Nike Maker Experience system, which makes it possible to create a personalized pair of sneakers in less than two hours.

Making parts at Machine Labs (Photo: techcrunch.com)

Another key change is the mass implementation of smart sensors in equipment and production lines. The integration of the Internet of Things (IoT) and artificial intelligence (AI) creates a system where AI acts as the “brain” and the Internet of Things serves as the “body”. IoT devices collect and transmit data from various sources, which are then analyzed, structured, and processed by AI. Based on the results, AI decides on the necessary actions. Thanks to IoT and AI, production becomes more transparent.

A similar technology was implemented by the automotive parts manufacturer Faurecia  —  it built a workshop taking into account the requirements of industrial IoT and automation, the integrated system ensures transparency of operations and carries out quality control of parts manufacturing.

In 2020, the average level of implementation of AI solutions in the world was 54%. It is important to note that, although neural networks are currently being implemented very unevenly, AI in industry is one of the most promising areas.

Neural networks in construction: from project development to facility delivery

It may seem unexpected, but over the past five years, neural networks have become widespread even in construction. This is evident from the numbers: in 2021, the global market for artificial intelligence in the construction industry was valued at $496 million, and is expected to grow to $8.6 billion by 2031.

Firstly, construction robots controlled by artificial intelligence have appeared and are actively developing. Among the well-known cases are the construction of a dam by robots in China or the construction assistant robot from Boston Dynamics, which was announced in 2023.

Secondly, technologies help collect large volumes of data about the construction site and use artificial intelligence algorithms, for example, for predictive analytics. Predictive analytics systems process large volumes of data and generate probable scenarios for construction duration, cost, possible risks, etc.

Thirdly, neural networks have facilitated the process of creating BIMs  —  digital interactive 3D models of structures at each stage of the life cycle of a construction project. A huge amount of data is required to create and correctly use a BIM model. In 2018, Autodesk developers came up with generative design. This design method using artificial intelligence allows you to quickly create BIMs — thousands of design options optimized for all parameters: from strength to sustainability.

Already, many technologies that seemed like science fiction are becoming part of the everyday life of engineers and builders. But a boom in AI implementation can be expected in the coming years: according to some forecasts, the global AI market in construction will grow by 34% per year from 2022 to 2031 and expand from just under $500 million in 2021 to $8.6 billion in 2031.

Neural networks in the transport industry: route optimization and driverless driving

Five years ago, it was expected that by 2023, the global AI market in transport would reach $3.5 billion. It is difficult to say whether the predictions came true, but it is undeniable that during this time, artificial intelligence technologies in transport have made great strides.

One of the most revolutionary applications of artificial intelligence has become driverless vehicles. Autonomous taxis have been operating in Tokyo since 2018, and the United States is introducing driverless trucks to reduce logistics costs.

Another transport problem that people face every day is traffic jams. Now, AI is ready to solve this problem as well. Sensors and cameras embedded everywhere on the roads collect a large number of traffic details. This data is then sent to the cloud, where analysis and detection of traffic patterns will be performed using big data analytics and an AI-based system. Similar systems are used, for example, in different states of America, Singapore and other countries and cities of the world.

Another pressing problem that AI has solved is traffic accidents. Scientists from the Massachusetts Institute of Technology have proposed their solution  —  they have created a neural network that predicts the occurrence of accidents. Similar solutions will probably soon be implemented around the world.

The dataset used to create the crash risk maps covered an area of 7,500 square kilometers from Los Angeles, New York, Chicago, and Boston. Among the four cities, Los Angeles was the most unsafe, as it had the highest crash density, followed by New York, Chicago, and Boston (Photo: mit.edu)

illuminem Voices is a democratic space presenting the thoughts and opinions of leading Sustainability & Energy writers, their opinions do not necessarily represent those of illuminem.

Curious how major companies measure up on climate? On illuminem’s Data Hub™, explore verified emissions data, net‑zero targets, and sustainability performance of thousands of firms — from industry leaders to emerging innovators.