AI is presented as “infiltrating” more and more areas of business, health, transportation, security, etc., but the reality is that sectors and industries have been experimenting and utilizing it for years. Whether you realize it or not, this article will reveal how AI is a part of your daily life. We’ll seek to de-mystify AI for you, and then we’ll look at a few areas and industries to see how AI is helping to find solutions for complex problems.
Since this is a significant topic, we are writing about it in a three-part series to keep to our standard of making these blasts 3 or 4 minute reads that provide you with substance. This article is to understand AI. In the first week of August, we will publish an article regarding the downsides of AI in terms of privacy, transparency/ethics, and energy usage. Lastly, we will provide you with an article about the expected evolution of AI.
There are 5 categories of AI: Analytic AI, Functional AI, Interactive AI, Text AI, and Visual AI. We’ll explain them and give you a (hopefully) easy-to-understand example.
Analytic AI
With its ability to evaluate and synergize massive amounts of data, Analytic AI is used to analyze and produce observations of data sets, even images. This is most exciting in terms of understanding patterns in datasets of information on a massive scale, particularly for medical research.
Functional AI
The Internet of Things (IoT) is an easy-to-understand concept, and one you may have experienced at perhaps an airport or a very nice office building. An AI-enabled HVAC system will monitor the facility’s temperature based on the climate outdoors, the current temperature indoors, and the capacity of people in the building, and will calibrate the proper temperature in real-time.
Interactive AI
Have you typed a question on a website and received an answer from its chatbot? If so, you have experienced Interactive AI. In the future, smart personal assistants are projected to help us in our daily lives. Interactive AI is may very well be a part of how we function in our own homes and offices in the future.
Text AI
Text recognition and speech-to-text conversion are tools many of us utilize: a prime example is the talk-to-text feature on your cell phone. Another is asking your Roku remote to find your favorite movie.
Visual AI
Generative AI creates images based on data sets and input from past data entries and past events. Computer Vision is AI that can derive information from visual inputs in real-time and then generate a recommendation. In transportation, it can analyze traffic density on freeways and “see” behavior at urban intersections. This can help analyze traffic flow and help compute routes that are better optimized for timing and fuel usage.
So how is this all really coming together in certain areas or industries?
AI in Cybersecurity
Threat Detection
We supply our clients with AI-powered software that analyzes significant amounts of data in real-time to find abnormalities that may be potential cyber threats. While some functions look for malware evidence, others note suspicious activities, and others guard against unusual network behaviors.
Containment
Other programs we employ are more mitigating. We utilize programs that identify and then contain a cyberattack. This reduces the response time for cyber threats and attacks. Often, our clients have no idea that an attempt was made in the first place because our tools have recognized and contained a cyberattack.
Advanced Security
We have installed some futuristic enhanced authentication software. We provide SOC (24×7 Security OPs Center) and EDR (Endpoint Detection and Response) that are backed by AI. This goes past AV in that it looks at network and endpoint events and is able to determine suspicious activity. We then review its findings and make the appropriate response.
AI in Health
AI Diagnosis Assistance
AI is proving itself to be extremely useful in the detection of different types of cancer.
In recent years, AI software has helped radiologists detect problems and diagnose this complex disease. Its strength is that the software can store and evaluate large datasets of images and identify oddities that a human radiologist might miss. Earlier detection is paramount in cancer care and the toll it takes on a patient.
The highly-regarded MASASI randomized control trial of 80,000 Swedish women found that cancer detection rates were 20% higher in women whose mammogram was read by a radiologist using AI compared to the women whose mammograms were read by two radiologists (the European standard of care) without any AI assistance. This has enormous implications in America and other countries where one radiologist reads a mammogram and that physician may or may not be a specialist in breast mammography.
AI in Transportation
Crossing that Bridge When You Come to It: AI and Bridge Predictive Maintenance
AI is improving the predictability of bridge maintenance. Considering all bridges are integral to transportation and about 80,000 of them need replacing in the US, this is a very influential area for the usage of AI. While the Department of Transportation mandates the evaluation of bridges and classifies them into an annual report called the National Bridge Inventory, professors at two universities found that report lacking to predict dangerous conditions or worse: bridge failures. Professor Kaijian Liu of the Civil Engineering Department at the Stevens Institute of Technology and Professor Nora El-Gohary of the Civil & Environmental Engineering Department at the University of Illinois Urbana-Champaign both saw the need for more data than just the NBI.
The professors gathered public data on factors like temperature changes, precipitation, various other weather conditions, report notes from the Washington State Department of Transportation, and bridge traffic data regarding cars and several classes of light and heavy trucks. After inputting all this data in a recurrent neural network AI (RNN) to learn about evolving conditions and after folding in multilayers of data, the AI assigned a future condition assessment range from 1 (imminent failure) to 9 (excellent). The result was impressive: the AI accurately flagged potential future issues about 90% of the time. This is 15% to 20% more accurate than existing methods. The professors noted that this method was particularly accurate at forecasting the future condition of aging bridges.
AI in Architecture
AI Assistance in Evolving Architecture
Architecture is evolving rapidly and algorithms are used to optimize the structural and operation aspects of projects. An intriguing example would be The Shed (The Bloomberg Building) which has a moving component that changes the usage of the front of the building. Hoisted up on bogies (similar to railway cars), the Shed’s moveable part either reveals a large terrace for up to 2,200 seated audience members or it can close and therefore instantly create a 17,000 square foot hall with light, sound, and temperature control, accommodating large-scale events and installations.
The benefit to humans is immense. We do not see AI as a technology to be frightened of yet proper parameters need to be put in place for personal privacy, cybersecurity, and the selling of information.
A Complete List of Links Included Within the Article Copy:
Artificial intelligence-supported screen reading versus standard double reading in the Mammography Screening with Artificial Intelligence trail (MASAI): a clinical safety analysis of a randomised, controlled, non-inferiority, single-blinded, screening accuracy study
https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(23)00298-X/abstract
CBS News: Mammography AI Can Cost Patients Extra. Is It Worth It?
https://www.cbsnews.com/news/mammogram-ai-cost-patients-is-it-worth-it
American Medical Association
Forbes
ProPublica
https://www.propublica.org/article/cigna-pxdx-medical-health-insurance-rejection-claims
Stevens Institute of Technology
https://www.stevens.edu/news/safer-bridges-an-ai-to-predict-structural-problems-before-they-worsen
The Shed
