AI for Discovery: Uncovering Insights for System Optimization
Pat Stanard, chief mainframe architect for Kyndryl U.S., explains how AI can be used to identify dependencies, analyze code, address security threats and more
I’m one of those guys that has tried to stump GenAI from the beginning, but in that process, I’ve found it to be an invaluable tool that makes my job easier.
Large language models have improved immensely in a very short period. I can ask detailed questions about the mainframe and get a good response 90-95% of the time. This is a time saver for me (although one caveat is you always need to review the responses for accuracy).
AI’s relationship to the mainframe doesn’t stop there, and I wanted to dig a bit deeper into one important form of AI that interests me, an area known as AI for discovery: the use of AI to uncover new insights, patterns and knowledge. This can include pursuits like scientific research and drug discovery, but in IT, AI also relates to the discovery of system infrastructure, dependencies, code analysis, and security and data structures that are in place.
This article will focus on those factors, and how organizations can ultimately find business value by optimizing them with AI.
System Infrastructure
AI for discovery in the context of system infrastructure is defined as the use of AI to improve the discovery and utilization of data, services and resources within a system. Doing so can set up a system for various AI uses, including the potentially transformative agentic AI.
Enabling Agentic AI With a Dedicated Service Layer
Agentic AI involves self-directed AI systems capable of autonomously performing tasks and interacting with other AI systems without human intervention. This is mind-blowing for me, but these systems require a strong infrastructure to support AI-to-AI communication and data exchange.
To facilitate efficient AI interactions, a dedicated service layer is essential. This layer acts as a communication protocol, enabling AI agents to discover and utilize services and data across the network, ensuring that AI systems can operate without the overhead of human-friendly protocols.
AI for discovery relies on high-performance computing infrastructure, such as NVIDIA’s AI Data Platform, which integrates enterprise storage with accelerated computing to power AI agents. This infrastructure supports near-real-time data processing and insights generation, optimizing AI workflows.
Data Orchestration
Effective AI for discovery also requires sophisticated data orchestration mechanisms to manage and distribute data across various layers of the AI infrastructure. This includes tools for intelligent routing, load balancing and advanced caching to improve inference speed and accuracy. Think about Rocket software’s DataEdge platform.
Identifying, Analyzing and Managing Dependencies
AI for Discovery in the context of dependencies means using AI to identify, analyze and manage the relationships and interdependencies between various components within a system.
AI tools can automatically detect dependencies between software components, services, and infrastructure elements, significantly reducing the time and effort required compared to manual methods. Today, in my opinion, manual methods would be impossible to implement.
Additionally, AI can predict potential issues and risks associated with dependencies by analyzing historical data and patterns, forecasting how changes in one component might impact others and helping to prevent system failures and outages. AI tools can identify successor/predecessor relationships of the components and provide real-time visualizations of dependencies, making it easier for teams to understand complex relationships within their systems. These visualizations can highlight hidden dependencies and provide insights into how different components interact.
By providing a clear and accurate map of dependencies, AI tools facilitate better collaboration among development, operations and IT teams, helping in planning and executing changes more effectively. With AI-driven dependency management, organizations can improve the reliability and performance of their systems by proactively identifying and addressing dependency-related issues, thereby reducing downtime and enhancing overall system stability. Think about the Amazon Q Developer.
Analyzing Code
AI for discovery in the context of code analysis is defined as using AI to enhance the process of examining and understanding code.
AI tools can automatically review code to identify potential issues such as bugs and security vulnerabilities, thus speeding up the review process and ensuring thorough analysis. Additionally, AI can recognize patterns in code that might indicate common errors or inefficiencies.
By learning from vast amounts of code data, AI systems can pinpoint areas that need improvement. AI-driven tools provide suggestions for improving code quality, such as refactoring recommendations, which can optimize performance and enhance readability to help developers write cleaner, more efficient code. Integrated into DevOps, AI-powered code analysis tools offer real-time feedback as developers write code, providing immediate guidance to prevent errors and promote best practices.
Furthermore, AI can detect security vulnerabilities in code, such as unsafe coding practices. This helps developers create more secure applications by identifying these issues early. It also improves the DevOps process. Think about solutions like IBM’s Watson X Code Assistant.
Addressing Security Threats
When it comes to security, AI for discovery can help to enhance the identification, analysis and management of security threats and vulnerabilities.
AI can automatically detect security threats by analyzing patterns and anomalies in data, including identifying unusual activity such as abnormal IP addresses, user agents and API usage. AI systems can continuously scan and analyze code, infrastructure and applications to identify vulnerabilities, taking a proactive approach to discover potential security risks before they can be exploited.
AI can assess the security posture of an organization by analyzing data sensitivity, compliance with regulatory frameworks and potential exposure to threats, including identifying badly permissioned or misconfigured access controls. AI-powered tools provide real-time monitoring of systems and networks, enabling immediate detection and response to security incidents, which helps minimize the impact of breaches and maintain system integrity.
Additionally, AI can suggest and implement remediation actions to address identified vulnerabilities and threats, such as patching software, adjusting access controls and removing malicious code. Think about BMC’s Automated Mainframe Intelligence solution.
Improving Data Structures
AI for discovery can also be used to analyze, optimize and manage data structures more effectively. In this process,it can identify inefficiencies and suggest improvements, including detecting redundant or suboptimal structures and recommending more efficient alternatives.
it can identify inefficiencies and suggest improvements, including detecting redundant or suboptimal structures and recommending more efficient alternatives.
AI systems can recognize patterns in data usage and access, helping to optimize data structures for better performance by identifying frequently accessed data and suggesting caching strategies or more efficient indexing. Additionally, AI can optimize data structures by dynamically adjusting them based on usage patterns, leading to improved performance and reduced resource consumption, such as balancing tree structures or optimizing hash tables.
AI can also predict potential issues with data structures before they cause problems by analyzing historical data and usage patterns, allowing for proactive maintenance. Furthermore, AI can identify vulnerabilities in data structures that could be exploited by malicious actors, continuously monitoring and analyzing data structures to help secure them against potential threats. Think about the NVIDIA AI Data Platform.
The Ultimate Goal: Business Value
For enterprises, the purpose of AI for discovery is not merely to optimize software systems, but to ultimately find business value. So, let’s look at the potential bottom-line benefits that extend beyond the IT department.
Innovation and Competitive Advantage
Companies that effectively use AI for discovery can innovate more rapidly and stay ahead of competitors. AI can uncover new business models, products and services that were previously unknown or not possible.
Reduction of Operational Costs
To understand the efficiencies that can be found with AI for discovery, look at drug discovery as an example. In this field, AI can identify potential compounds more quickly and accurately, reducing the need for expensive and time-consuming laboratory experiments.
Informed Financial Decisions
In the financial sector, AI helps in discovering patterns and trends in large data sets, which can be used for fraud detection, risk management and investment strategies. AI is actively being used as a method on loan approvals and denials with automated credit scoring and lending. AI analyzes a wide range of data points to assess creditworthiness, enabling more accurate and fair lending decisions.
New Revenue Streams
AI for discovery can lead to the development of new streams of revenue. For instance, AI can help create personalized products and services, enhancing customer satisfaction and loyalty. Key customer data can be analyzed to provide personalized recommendations and offers targeting specific customers.
Implementing AI for Discovery
To take advantages of the capabilities of AI for discovery, digital platforms are available to help organizations manage and transform their complex technology environments. These tools can provide integration, observation and orchestration across various technology systems to provide actionable insights, optimize spending and drive innovation.
AI is not new—the term “artificial intelligence” was coined by John McCarthy in 1956 during the Dartmouth Conference, which is considered the birth of AI as a field. But only recently has AI become a part of our everyday lives, poised to transform the way we discover information and make information actionable. AI is here to stay, and we will need to embrace this technology to take advantage of all the efficiencies it promises.
Happy 68th birthday, AI!!!