Although AI is currently a mainstream technology, it continues to evolve as engineers advance different AI categories to address specific needs.
One such category is adaptive AI, which research firm Gartner has named it among the top 10 strategic technologies of 2023. Adaptive AI is a type of artificial intelligence that not only learns from ingested real-time data, but also modifies itself in response to that data. .
“It's built to learn almost autonomously on a regular basis,” said Erick Brethenoux, analyst vice president at Gartner. These characteristics make adaptive AI a promising technology that is expected to significantly improve various processes across multiple sectors.
”[Adaptive AI is] It's useful for rapidly changing situations, such as healthcare and finance, where real-time analysis based on changes occurring within a dataset is required,'' says Chief Architect of AI and Machine Learning (ML) at IT firm UST. says Adnan Masood. Services and Consulting Company: “This is where this technology really shines.”
What is adaptive AI?
Adaptive AI utilizes a variety of techniques such as ML, reinforcement learning, neural networks, agent-based modeling, and evolutionary algorithms. The fact that it is a combination, or composite, of different AI technologies places adaptive AI in the broader category of composite AI.
This combination of technologies allows adaptive AI to change its code in response to changing circumstances and experiences over time. So you can improve your own performance and accuracy as you work.
“Adaptive AI involves operating systems that can learn from data and change their behavior over time,” Massoud said. “Constantly updated based on incoming data.”
This is in contrast to other AI systems. For example, static AI systems have algorithms that are trained and tested during the development phase, and once proven accurate and efficient, are frozen before deployment.
Other types of AI systems are trained on historical or static data and do not incorporate information beyond the data cutoff date. Even ChatGPT faces this limitation, with its maker OpenAI stating that its GPT-4 model “generally lacks knowledge about events that occurred after the majority of the data was cut off (September 2021). “They have not learned from the experience.” ”
“The basic idea of adaptive AI is to add more up-to-date knowledge. It literally becomes more adaptable as the data set changes,” said IEEE senior member and co-founder of consulting firm Xmark Labs. said Nicholas Knapp.
Additionally, adaptive AI combines context awareness with other capabilities such as risk scoring, making it a particularly powerful technology. Brethenoux called it “the most advanced form of AI implementation on the market today.”
Adaptive AI use cases
Adaptive AI is particularly suited for use in chaotic, complex, and dynamic situations where it is impossible to predict or plan for every possible scenario. Areas where adaptive AI is already being used, is starting to take hold, or could be applied in the future include:
1. Industrial monitoring
Energy companies, utilities and similar organizations are using adaptive AI to monitor facilities that are difficult to reach and inspect, such as offshore wind turbines, Bretenou said. “They are adaptive systems because their behavior is based on what they observe,” he says. “You can't preprogram it.”
This use case uses adaptive AI to direct activities and take actions based on different situations. For example, the technology could direct drones to inspect when it detects ideal weather conditions, process information from those initial inspections, and perform additional inspections or tests based on the results. You can give instructions to the drone.
2. Agricultural monitoring
Similarly, adaptive AI can help scan agricultural areas. Drones and other equipment collect data, such as images and soil samples, that adaptive AI systems can analyze to determine the best next action to take.
3. Fraud and anomaly detection
Manjeet Rege, professor and chair of the Department of Software Engineering and Data Science at the University of St. Thomas, said the adaptability of adaptive AI greatly improves the accuracy of the system when conditions change rapidly, helping with anomaly detection. It says.
As an example, consider an adaptive AI system designed to detect credit card fraud. At the start of the COVID-19 pandemic, the sudden shift of credit card users from minimal online shopping to online-only shopping, incorporating information about lockdowns, is unlikely to be a sign of fraud. I should have recognized it.
4. Investment, trading and other commercial financial activities
Weiguo Patrick Huang, a business analytics professor at the University of Iowa's Tippy School of Business, cited commercial finance as an area that could benefit from adaptive AI systems. Adaptive AI receives and categorizes a myriad of signals and data sets, from changing financial markets to U.S. Federal Reserve policy announcements to social media trends, to detect unexpected and unexpected developments. can respond accurately.
5. Training and education
The U.S. military is already using adaptive AI to train personnel, and the technology's various benefits are expected to make its way into other education and training programs, Huang said. He explained that the technology allows him to assess student responses, learn what works best for different types of students, and use that information to fine-tune the lessons he offers.
6. Self-driving cars
Self-driving cars, also known as self-driving cars, already use adaptive AI to navigate unpredictable environments and instantly react to changes in conditions that they were not originally programmed to anticipate. doing.
Perhaps more importantly, adaptive AI can learn from these actions and continuously improve its navigation capabilities and response to ever-changing driving conditions. “We can analyze what we observe, make decisions very quickly, and adjust based on feedback,” Huang said.
7. Robotics
In robotics, adaptive AI can provide functionality similar to self-driving cars. Adaptive AI can use current and historical data to figure out how to best respond in the moment, and then use that response to improve responses as conditions change, Huang said.
As an example, Huang pointed to a company working to develop companion robots to support the elderly. Adaptive AI allows these robots to tailor interactions to each individual based on that person's interests and adjust their behavior in subsequent interactions to create the optimal experience.
8. Personalized recommendations and services
According to Bretenou, adaptive AI is being used to create hyper-personalized services for customers. It also helps recommend content, such as content from streaming services, and products from retailers by incorporating up-to-date data about customer preferences and context.
“I'm always learning about what I like,” said Nicholas Avila, chief technology officer for North America at digital consulting firm Globant.
9. Chatbots for better customer service
Legge said adaptive AI is expected to improve the effectiveness of customer service chatbots. Customer service chatbots have historically shown mixed results when it comes to providing helpful and relevant responses to customer questions.
But with adaptive AI, bots can continuously learn from customer interactions. This allows the bot to more accurately assess what a customer wants, even if the customer asks an awkward or unusually worded question, Legge said. With increased accuracy, your bot's responses are more likely to meet your customers' needs.
10. Supply chain improvement
Adaptive AI can improve supply chain resiliency by monitoring changing variables, more quickly identifying potential disruptions and other changes in market conditions, and making adjustments to compensate for those factors. Legge says that there is a gender.
Additionally, adaptive AI can analyze all these variables to move the right product, at the right time, to the right place, in the right quantity, at the best price to meet customer demand. Taken together, “this all means minimizing supply chain costs and better matching supply and demand,” Legge said.
11. Healthcare diagnosis
Looking to potential future applications, Rege said adaptive AI could improve diagnostics in medical settings. “This is a futuristic goal, but it has the potential to greatly benefit humanity,” he added.
Adaptive AI is better than other technologies at identifying subtle changes in a patient over time and learning how those changes are manifested in the early stages of disease, says Rege. I explained. Similarly, adaptive AI could help in areas such as effective patient monitoring, preventive care, and drug discovery.
12. Medical Claim Processing
Masoud said he sees the potential for adaptive AI to process health insurance claims more efficiently while also improving the ability to root out fraudulent claims. He explained that technology's ability to understand and learn from complex situations could eliminate the manual labor required to review claims that don't fit predefined rules.
13. Knowledge Management
Adaptive AI has the potential to understand complex organizational knowledge in any organization, Knapp said. “There is a lot of information that is siled and not available within the enterprise,” he said. “Adaptive AI has the potential to significantly solve this problem.”
14. A better digital twin
Organizations can apply adaptive AI to digital twins to better understand how changing scenarios impact the systems and people represented by the digital twin.
“When you publish a digital twin and get a wide range of information, you can't predict everything. It's chaotic,” Bretenou said. “That’s where the adaptive part comes in.”
Future impact of adaptive AI
Adaptive AI is gaining increasing interest as it has proven its value in a variety of fields. But most executives are still learning about this technology and how it differs from other types of intelligent systems, such as generative AI.
Adaptive AI is already being used in energy and utilities, agriculture, aerospace, military, and traffic management. Bretenou said he expects the types of industries that use adaptive AI to expand rapidly, from financial services to the technology's ability to bring order to complex, unpredictable, and even chaotic situations. He pointed out that it is in a unique position to help in all areas, including crowd control.
“There's been a lot of interest from companies thinking there might be something useful,” Avila said. “Companies have large amounts of data that cannot be monetized or activated, but this combination of AI could be the key to unlocking the value of the data they collect.”
