In the final quarter of 2022, Gartner released an article outlining the top ten strategic technology trends for 2023. The research divides these trends into three buckets for companies to:
In his December 6, 2022 article, Forbes contributor Steve Andriole challenges the concept of adaptive AI:
“To ‘neologize’: ‘to make or use new words or create new meanings for existing words,’ which may sum up this year’s technology trends issued by the Gartner Group. I’ve never seen such creative repackaging and such inexplicable reach.”
Read on as we explore:
- The definition of adaptive AI
- How adaptive AI works
- How adaptive AI differs from traditional AI
- Real world examples of adaptive AI
- Digital work trends
- Adaptive AI advantages
- Implementing adaptative AI
- What’s in a name?
What is Adaptive AI?
Adaptive AI is recommended as a technology for organizations to pioneer this year. Gartner defines it as supporting “a decision-making framework centered around making faster decisions while remaining flexible to adjust as issues arise. These systems aim to continuously learn based on new data at runtime to adapt more quickly to changes in real-world circumstances. The AI engineering framework can help orchestrate and optimize applications to adapt to, resist or absorb disruptions, facilitating the management of adaptive systems.”
Furthermore, Gartner states:
“Adaptive AI systems allow for model behavior change post deployment by learning behavioral patterns from past human and machine experience and within runtime environments to adapt more quickly to changing real-world circumstances.”
Andriole asks if it means “that supervised and unsupervised learning applications adapt in semi-real-time to changing behavioral patterns discovered post-deployment. Perhaps shifting in real-time from one to the other with some algorithmic dancing along the way? What?”
What distinguishes adaptive AI from traditional AI? Gartner says adaptive AI systems “change their learning dynamically and adjust goals.” This sounds like deep learning, but Gartner says adaptive AI differs from AI as we know it and that “traditional AI in that it can revise its code to adjust to real-world changes.”
Ultimately, AI should be designed to be adaptive to a certain level. This trend may be about delivering on the promises that AI made over the last ten years and failed to deliver.
What do you think? Is adaptive AI the same as AI, labeled differently? The nature of AI is already adaptive.
How Adaptive AI Works
We mentioned that adaptive AI changes its code and continuously retrains to learn and adapt as new experiences come up, has new observations and data is processed in production without needing developers to rebuild it.
Furthermore, it eclipses traditional AI because it is learning while interacting with data, offering more benefits and better learning capabilities.
Gartner states that adaptive AI is different from traditional AI. Here is what they say:
|What is the Difference?|
Adaptive AI Examples
When looking at adaptive AI examples, we ask ourselves if this “new” strategic technology is NLP/deep learning re-packaged and re-labeled.
Like traditional AI, there is a multitude of real-world applications for adaptive AI. Gartner found that brands like Amazon, Netflix and Google use adaptive AI to improve user experience. Also, cybersecurity organizations explore adaptive AI to create automated, self-sustaining protocols that learn and model themselves with recurrent iterations that fight cyber threats and attacks in real-time.
Brian David Crane, founder of the digital marketing fund Spread Great Ideas told CMS Newswire that adaptive AI is the next big thing in automation and AI, and it is an example of how brands use AI today. Crane believes adaptive AI has wide-ranging transformative implications in customer experience.
According to Crane: “By analyzing social, behavioral and past interactions, adaptive AI uses continuous interactions to predict and anticipate customer behavior and provide highly personalized solutions to improve the customer journey and deliver positive CX” when he explained adaptive AI’s focus on feelings, emotions and sentiment analysis that offer ideal interactions in real-time.
There are further uses for adaptive AI, such as assessing credit risks. John Fenstermaker, Equifax’s Head of US Customer Analytics, planted a flag in the realm. Fenstermaker developed a product based on customer struggles with making optimal business decisions. In 2019, Fenstermaker took on the challenge of helping an Equifax customer, a deep subprime lender. This customer sought a replacement for a core model it wanted to phase out. The lender faced challenges differentiating credit risks with their audience and sought to mitigate risks with AI. The goal of the new model was to capture more “good” risks than “bad” risks. Using complex non-linear attributes for deeper learning of consumer behavior.
According to a case study on Equifax’s website:
- Customer data provided by the lender was leveraged to automatically determine if a new applicant was a prior or current customer
- Equifax’s consumer credit database was used to reveal an applicant’s end trajectory by showing their financial decisions over 24 months
- Data from utilities, communication and pay television was used to provide insight into how people pay household bills
- Peak attributes offered a premium set of tri-bureau enabled consumer credit attributes to help the lender’s highly specialized needs. Interactive attributes use AI-enabled technologies to create “super attributes,” which fuse multiple attributes to create one more powerful and predictive attribute
Moreover, the case study mentioned that the results from the new model “helped the lender approver 92,000 more accounts without increasing losses.”
Furthermore, Equifax’s results included the following:
- $13.7 million in annual loss savings
- A 92 percent lift in the Kolmogorov-Smirnov (KS) statistic
- A 7 percent lift in the retailer’s scorable rate, a significant improvement when dealing with subprime consumers. Equifax found that lenders can’t score many applicants due to a lack of credit. The more customers the model can score, the more it can confidently approve and serve.
Our takeaway from this case study is that adaptive AI’s technical profile and value offer an opportunity to alter behavior after deployment by learning behavior patterns from prior human and machine experience. AI engineering offers foundational parts of implementation, operationalization and change management levels. Using real-time feedback, it dynamically learns and changes during unforeseen challenges.
To wit: is the re-packaging necessary?
What Gartner says about digital work trends:
The U.S. Food and Drug Administration (FDA) will implement a certification program for AI-enabled products in hospitals. The FDA expects to see widespread use throughout America.
Educators and learning institutions use adaptive AI technologies to alter class lessons and materials according to students’ needs and learning pace. They believe personalized educational journeys will increase high school levels, college degrees and course certificates.
Trust, risk and security management
AI learns and adapts. It detects minor alterations in online behavior a human may miss. AI works better than humans at detecting blind spots, safeguarding identities and organization-critical applications, finding cyber threats, responding to threats and delivering recovery solutions.
Advantages of Adaptive AI
The Gartner report states that adaptive AI “brings together a set of methods (I.e., agent-based design) and AI techniques [such as] reinforcement learning to enable systems to adjust their learning practices and behaviors so they can adapt to changing real-world circumstances while in production.” Building adaptability and resilience into design allows organizations to react “quickly and effectively to disruptions.
“Flexibility and adaptability are now vital, as many businesses have learned during a recent health and climate crises,” says Gartner Distinguished VP Analyst Erick Brethenoux. “Adaptive AI systems aim to continuously retrain models or apply other mechanisms to adapt and learn within runtime and development environments — making them more adaptive and resilient to change.”
As a technology to pioneer, adaptive AI will enable “business model change, reinventing engagement with employees and customers, and accelerating strategies to tap new virtual markets.”
- Create secure foundations
- Maximize value from data
The business value of operational AI lie is in its capability to “develop, deploy, adapt and maintain AI across different environments.” With engineering complexities and the demand for faster times to market, Gartner says it is important to develop more flexible AI engineering pipelines or create AI models able to self-adapt during production.
Adaptive AI Pioneers
Gartner states that adaptive AI speeds value and maintains AI’s alignment to organizational goals in real-time.
Let us present several use cases from Gartner. You can decide for yourself if adaptive AI is a “new” AI field:
- Dow, the U.S. chemical and materials manufacturer, deploys adaptive AI systems that use feedback on usage patterns and business value optimization to enhance enterprise analytics. It has resulted in a 320% increase in value generated by the analytics platform.
- Cerego, AI-based training software used by the U.S. Army, enables adaptive learning. The solution knows what to teach, how to measure progress and when to test, adapting its lessons to each individual’s learning progress.
- The Danish Safety Technology Authority (DSTA) has to monitor the safety of products sold in Denmark, regardless of where they come from. Its AI tool rapidly finds products and their manufacturers, thereby speeding up the detection of product problems. DSTA has created a spin-off product currently deployed in 19 other European countries.
Adaptive AI uses real-time feedback to learn and adapt, allowing for quick responses to changing environments, and contributions to training data, thereby personalizing generalized results.
What do you think?
Gartner’s steps to implementation:Technology leaders need to re-engineer various processes to build adaptive systems that can learn and modify behaviors based on circumstanceImplementing decision intelligence capabilities broadens decision-making capabilities and makes them flexibleAdaptive AI solutions create superior, faster user experiences by adapting to circumstances as they come up The Future of Adaptive AI.
As the tech industry develops AI, it grows more powerful. Business AI is an expected part of technology stacks in the workplace. It is essential to invest in AI technology to improve digital resilience and remain at the front of the technology disruption curve.
What is your opinion? Is adaptive AI a new, burgeoning field, or is it a result of AI maturing?
Stefanini and Technology Innovation
Stefanini realizes the potential of AI. Businesses must embrace new and innovative ways of working to ensure their future viability.
Be ready to meet what is coming in the digital future. Partner with us to create the right solution for your needs. Reach out to a Stefanini expert today.