Three Areas to Prioritize for AI
Beyond an artificial intelligence (AI) primer, this article highlights three key areas of your organization where AI technologies can be timely and beneficial today…not five years from now.
Whether your industry focus is healthcare, retail, finance or a variety of customer types, now is the time to adopt a strategic view of AI and how it will converge with your business.
While it has been a staple of computer science for decades, AI has gained significant interest and activity due to the confluence of a number of developments:
- Improved Algorithms: Better techniques, such as Neural Networks, has allowed for more advanced outcomes from AI systems.
- Specialized Hardware: Advancements in Graphics Processing Units (GPUs) has made them ideally suited for AI applications; massively parallelize complex mathematical processes with continuing scalability.
- Extensive Data: The explosion of data and data sources through web technologies and utilization of the Internet of Things has greatly improved AI’s ability to learn and identify patterns.
- Cloud Services: Companies such as Amazon, Google, and Microsoft have extended their cloud offerings to include AI services readily available to developers.
All of these developments have made AI easier, cheaper, and more accessible to a wider scope of developers and applications.
In fact, in the healthcare space alone, the AI market is expected to grow to $6.6 billion by 2021. That’s 10x growth since 2014. [i]
What’s so great about AI?
Artificial Intelligence (AI) includes hardware or software that exhibits behavior which appears intelligent. There are a few ways we can assess “intelligence,” but one of the primary ones is that AI systems are self-learning. This means that these systems can determine actions based not on a predefined set of business rules but based on success or failure from incoming data.
There are three main capabilities associated with AI today:
- Capturing Information or Supervised Learning: Vision Recognition, Audio Recognition, Search
- Doing Useful Work: Natural Language Processing, Reasoning, Prediction
- Understanding: Why is something happening?
Of these three, the first two are the most relevant to today’s business environment.
To illustrate AI techniques versus capabilities. (see Figure 1 ).
- Techniques: (top row ) Lists approaches that drive capabilities
- Approaches on how to mimic intelligence
- Examples: Machine Learning, Robotic Process Automation
- Capabilities: (bottom row ) Lists outcomes that can be achieved
- Tools that provide an outcome
- Examples: Facial Recognition, Biometrics, Neural Networks
Opportunities for AI Now
Machine Learning (ML) is an approach to achieve AI. ML gives computers the ability to learn without being explicitly programmed. The process involves machines learning through training with large amounts of data and algorithms (“gaining experience”). There are many approaches to ML, each which have advantages and disadvantages.
ML has gained traction lately, as data is now available on a massive scale. Coupled with steady progress in computing power, this scale of data makes teaching a machine much more feasible than previously, and it has shown to be superior for many operations over other approaches (such as decision tree learning and Bayesian networks).
Deep Learning & Neural Networks
One of the most common approaches to ML today is Deep Learning (DL). Where older AI models were focused on modeling the world, Deep Learning is focused on modeling the brain by creating neural networks. A neural network is a collection of software-based calculators (“neurons”), which are connected to each other. Each neuron receives an input, analyzes it, and delivers an output. The inputs are weighted based on how correct they are relative to the task and then determines the output. The network is formed when the output of one neuron is the input of another neuron. Initially, neural networks will be wrong more often or not, but given enough data to input, they will vastly improve. Given a large number of neurons, machines can be trained for many different types of tasks. For example, Google’s AlphaGo learned the Japanese game of Go and is now able to beat human masters regularly.
Robotic Process Automation
Robotic Process Automation (RPA) is a form of automation. Where traditional automation was done through IT services (developing applications and integrations, RPA leverages software robots to automate complex, rule-based work. These robots mimic human interactions, interpreting signals, trigger responses, and communicate with other systems. One of the main goals is to automate complex, rule-based work that is not value-add, allowing people to focus on the interesting work[i], with a potential ROI of 30-200%[ii]. In many cases, RPA is disruptive, as RPA can be cheaper and easier to implement from a business perspective than an IT solution. This makes it important to think strategically on the RPA approach and get strong champions from both IT and the business engaged.
It is important to realize that RPA is not by itself AI, despite the marketing hype. Software robots need to be provided explicit instructions and don’t “learn” in the AI sense. However, RPA is a natural fit to work with AI. For example, RPA could help prepare disparate data formats to feed into neural networks. Additionally, AI could provide specific inputs to RPA to drive pre-determined behaviors (“Intelligent Automation”).
To learn more about SDLC Partners’ approach to RPA, read our article: Rise of the Robots — Robotics Process Automation.
How will this impact you?
AI techniques and capabilities, with the growth of processing power and availability of cloud-based offerings, are not “future state” capabilities. They are available now, and they can make immediate impact on your bottom line. Some examples include:
- Healthcare — Capture EMR information and data via voice interfaces.
- Improve the experience for both patients and doctors by offloading.
- 30% of hospital personnel are dissatisfied with their EMR systems. [i]
- Insurance — Streamlining claims by using AI to identify patterns and spot fraudulent claims more quickly. [ii]
- Handles large amounts of claims data to speed up the process.
- Learning capabilities mean that as new fraud patterns emerge, they can be flagged with little to no human intervention.
- Banking — Real-Time recommendations based on up-to-the-minute data on any channel
- AI systems can react and act more quickly than humans to provide financial and product advice.
- Coupled with AI-driven experiences such as chatbots and improved voice systems, overall customer engagement can be improved.
- General Customer Experience — Humana uses an AI bot to identify challenging member experiences with their agents, given them instant feedback on how to improve the interaction. [iii]
How can we help now?
As you think about how AI may be a strategic asset for your organization, SDLC can be your partner in navigating the AI journey. We look at particular areas first, including those listed below.
We’re here to help you prioritize AI initiatives based on benefits desired — maximized efficiency, streamlined operations and improving big data use for consumer engagement and delivery.
- Education and Strategy
- With our Services team, we can help you understand the impact of AI and be your partner in transformation into an AI-driven company
- We utilize an engagement model to reveal how AI can meet your goals, what AI tools are useful, and where is best to invest in this technology
- AI Data Architecture and Transformation
- AI works more effectively with a flexible and resilient architecture and a clean, well-prepared data set
- We utilize anti-fragile, large-scale architectures to ensure that clients are maximizing their AI investment
- Next-Generation Automation
- View our latest Point of View: Robotic Process Automation and the Age of the Digital Workforce
- AI can have a major impact on optimizing business processes. For example, Administrative Workflow Assistance is expected to become a $18 billion value segment [ii]
- Real-Time Interactions
- We can determine which approaches will help members follow their care plans more closely and dynamically adjust based on variables like the “next best action”
- Leverage the AI Cloud
- We can maximize the work of key AI capabilities, such as Amazon, Google, and Microsoft, capitalizing on what these companies have done with the “data science” work and exposing these capabilities for our use
- We built an internal product for managing our meetings using voice-based interfaces with an AI capability. It use AI capabilities to remove rote “meeting minutes” work and automates processing so that attendees can focus on high-value tasks
[i] Artificial Intelligence: Healthcare’s New Nervous System – Accenture
[ii] Artificial Intelligence May Change the Face of Business – Techonomy
[iii] Why Artificial Intelligence is the Future of Growth – Accenture
[iv] The next acronym you need to know about: RPA – McKinsey
[v] How Artificial Intelligence Will Cure America’s Sick Health Care System – Newsweek
[vi] How Artificial Intelligence is change the Insurance Business – Insurers.AI
[vii] How Health Plans Harness Artificial Intelligence – AHIP