Successful Early Adopters of AI Reveal Best Practices for Design, Deployment and Use

Developing and deploying artificial intelligence (AI) solutions efficiently and successfully in businesses requires a new set of skills, for both individuals and organizations.  In a recent study, over half of companies that have successfully deployed AI applications have embraced an enterprise-wide strategy that is inclusive, open, and pragmatic, using homegrown AI models 90% of the time. They have spent time understanding and documenting consistent and effective ways of rolling out projects and processes to drive efficiency. 

AI is Booming. Wanted: More People & Best Practices.

AI in business is advancing at a brisk pace. The market is forecast to grow at a Compound Annual Growth Rate (CAGR) of 36.2% between 2022 and 2027, when it will reach $407 billion, according to a recent study by MarketsandMarkets. But the report cautioned that: “The limited number of AI technology experts is the key restraint to the market.” The same lack of enough skilled personnel, along with established processes for deploying AI, was also cited in a recent global study of 2000 businesses by IDC.

Thirty-one percent of companies surveyed were actively using AI while the others were still in prototyping, experimentation, or evaluation stages. Significantly, companies using AI – considered early adopters – have integrated their AI platforms with the rest of their data center and cloud environments instead of running AI in silos used by separate groups. They have defined holistic, organization-wide AI strategies or visions along with clearly defined policies, guidelines, and processes. 

Another characteristic of these early AI adopters is that they use internal staff instead of external vendors to deploy AI applications. They also prioritize training line of business managers to use outcomes from algorithms and to tap these stakeholders to help guide new projects. This connection between IT and business leaders results in a high degree of support from C-level executives on down. 

AI Environments are Complex

To provide the massive compute power and data storage resources required for AI applications, businesses typically use systems with graphical processing units (GPUs) that accelerate applications running on the CPU by offloading some of the compute-intensive and time-consuming portions of the code. High-speed storage, parallel processing, in-memory computing, and containerized applications running in clusters are other techniques that are part of AI solution environments. 

Working with such complex technology requires the right training and experience. According to Datamation, there are 55,000 jobs currently listed under “artificial intelligence” on LinkedIn. Many if not most of these jobs (e.g., AI engineer, data scientist, AI/ML architect, AIOps/MLOps engineer) require years of education and advanced degrees. Yet the IDC study makes clear how much more effective AI projects are with these personnel designing models and collaborating with stakeholders in-house.

Scaling an AI Environment for Critical Healthcare Diagnoses

A leading pathology diagnostics firm in the U.S., that works with top biopharmaceutical and medical organizations around the world, has developed its own best practices for designing and deploying AI applications. Project teams at the firm include IT professionals, machine learning engineers, and data scientists who specialize in the biomedical industry.  Line of business managers also help guide the development of algorithms, 90% of which are developed based on the use of inhouse models. 

Many team members work primarily alone, then collaborate to deliver complex projects. With fluid, continually evolving project requirements, the company uses the Agile software development process that anticipates the need for flexibility in a finished product. To ensure that the technology they use (including GPU-based compute with high-speed and object-based storage and file-based access to Kubernetes clusters) is kept up-to-date and future proofed, the firm relies on close partnerships with vendors to review product roadmaps and anticipate and incorporate new features.

Agile development requires a pragmatic approach. IT managers at the firm insist that developers evaluate their work critically in the design phase and be willing to start from scratch if an approach isn’t working. In IDC’s survey, the companies actively using AI take an average of three months to build machine learning and deep learning models where AI laggards commit a fraction of that time. Deployment in AI early adopter companies like the pathology diagnostics firm, however, is accelerated because developers have already done their homework and obtained buy-in on models and validation from data scientists on technology purchases. 

Summary of Best Practices for Effective Use of AI 

As more C-level and line of business executives recognize and prioritize the use of AI as an effective tool to enhance competitiveness and drive efficiencies, the barriers to adoption have also become clear. Companies achieving success with AI have invested in people with skills and expertise. They have established vendor partnerships to future-proof solutions by staying up-to-date on evolving product roadmaps. They have fostered collaborative and highly flexible development environments that can alter course based on changing business dynamics. Using mostly homegrown models, they are committed to taking the time required to get the design of algorithms right before moving to well-defined established deployment processes.  Finally, AI development teams mentor business stakeholders, working with them to uncover and apply actionable insights from data analytics. 

Download the new IDC report to learn more about what is separating AI leaders and laggards. 

***

Intel® Technologies Move Analytics Forward

Data analytics is the key to unlocking the most value you can extract from data across your organization. To create a productive, cost-effective analytics strategy that gets results, you need high performance hardware that’s optimized to work with the software you use.

Modern data analytics spans a range of technologies, from dedicated analytics platforms and databases to deep learning and artificial intelligence (AI). Just starting out with analytics? Ready to evolve your analytics strategy or improve your data quality? There’s always room to grow, and Intel is ready to help. With a deep ecosystem of analytics technologies and partners, Intel accelerates the efforts of data scientists, analysts, and developers in every industry. Find out more about Intel advanced analytics.

Artificial Intelligence