One of the questions we hear a lot about artificial intelligence is, “Which algorithm or AI technique should I use?” For complex problems, the answer isn’t always one technique or another. Instead, the best answer to any problem is often a combination of multiple techniques and technologies.
We’re seeing with customers across industries that combining multiple AI techniques is what solves their problems. Maybe it’s forecasting and optimization. Or optimization and natural language processing. Regardless, it’s usually the combination of two, three or four AI technologies that provide the best benefit.
This ability to combine multiple AI technologies into a single solution is referred to by Gartner and others as composite AI. In the past, we might have called this multidisciplinary AI. But the new term better reflects how the techniques are integrated holistically to address unique challenges.
“SAS is pushing the boundaries of the larger composite AI movement that’s happening in the industry,” explained Bryan Harris at SAS Global Forum’s Opening session. “Accelerated digital transformation is demanding more sophisticated decisioning. This requires input from multiple analytic techniques, such as descriptive statistics, natural language processing, deep learning, audio processing, computer vision and more. Companies that can quickly harness these analytic techniques ultimately have a competitive advantage in their digital transformation.”
Later, at the conference’s Technology Connection, Harris highlighted the use of composite AI across multiple industries, including retail, medicine and banking.
This ability to combine multiple AI technologies into a single solution is referred to by Gartner and others as composite AI.
Composite AI in retail
For retailers, Harris brought in a team of SAS technical experts to demonstrate a composite AI solution for improving the process of returns. The solution uses conversational AI, predictive models and forecasting techniques to identify customers' preferences and to offer incentives for the customer to make more purchases during the return process.
When the solution identifies an online fashion influencer, for example, it automatically issues the refund and offers a custom discount for future purchases. The solution also tests out different promotions through a marketing campaign step called “market offer.”
“This shows how the power of combining composite AI with decision flows can be used to create a repeatable and elevated customer experience,” explained Shadi Shahin, Vice President of Product Strategy at SAS. The full solution, built on SAS Viya running on Microsoft Azure, is designed to improve customer satisfaction scores and reputation scores.
Composite AI in medicine
Next, Harris welcomed Dr. Geert Kazemier and Dr. Joost Huiskens from Amsterdam University Medical Center. Here, technologies like computer vision, data visualization and machine learning are being used to evaluate the efficacy of treatments for cancer patients.
“It is very difficult to determine if chemotherapy was effective, says Kazemie. And it is also very time-consuming. We have seen that medical image analytics on SAS Viya provides a more objective, accurate and automated way to determine the effect of chemotherapy.”
Kazemie and colleagues can visualize chemotherapy response rates based on an increase or decrease of the total tumor volume. Understanding this volumetric analysis, which is hard to see with a naked eye, can significantly improve overall survival rates.
“That is really powerful,” says Harris. The ability to start with statistical analysis and visualizations and then drill down to 3D renderings of the patient scans is extremely insightful.”
Kazemie agrees, saying, “Analytics enables us to be more efficient, reduce time-consuming repetitive tasks, and eventually help us make better decisions for our patients. We believe that together with SAS, we should not only develop innovative analytics but also use analytics in our daily clinical process. Together, we are working toward that goal.”
Composite AI in banking
Jason Grasso from the Global Security Intelligence Practice at SAS joined Harris to present a final demonstration of composite AI. This talk introduced the trade risk analytics compliance kit (TRACK), which transitions manual compliance reviews to an automated environment using AI.
Automating this risk process reduces compliance checks from 40 minutes to one minute. Celent Model Bank recognized Citi Trade and Treasury Services with an industry award for using this composite AI technique.
The process uses natural language processing to extract crucial data from trade documents, looking for certain terms and language. Next, 200 compliance checks are executed in real-time. Finally, the process uses a gradient boosting model to automate processes and reduce false positives.
Some clients have seen 25% improvements in operational efficiencies using this technique, explains Grasso. That can equal millions of dollars in savings when you consider the process can involve 9 million transactions, 25 million documents and 200 risk typologies.
Composite AI benefits all
To conclude the demo portion of Technology Connection, Harris said, “You just saw three unique demonstrations of Viya across different industries, all delivering improved time to value.”
In addition to these strong business benefits, teams across the company see value in composite AI too. Data scientists can incorporate different analytic capabilities in a repeatable workflow. And IT can deliver digital agility with composite AI, adding significant value to existing investments in data and applications.
"Composite AI offers more sophisticated decisioning by combining multiple analytic techniques, such as descriptive statistics, natural language processing, deep learning, audio processing, computer vision and more. Companies that can quickly harness these analytic techniques ultimately have a competitive advantage in their digital transformation."Log in to SAS Global Forum to learn more about composite AI