Industry executives and experts share their predictions for 2022.  Read them in this 14th annual VMblog.com series exclusive.

4 AI/ML Trends to Watch

By Brian Bartell, Ph.D. | VP - AI & Machine Learning at Conga

Artificial intelligence (AI) and machine learning (ML) may feel like newer technologies to some, often used as buzzwords or part of the overall conversation related to ‘digital transformation.' However, those closer to AI/ML and its development know that this technology is far from new. And it has been providing real, practical value to customers for several years.

As someone who has been "doing ML" for a couple of decades, it has been astonishing to see how these technologies have evolved and advanced over time. While there is still maturation ahead of us, AI and ML are ‘emerged' technologies rather than ‘emerging.'

Companies are also realizing the benefits of these emerged technologies, and prioritizing AI and ML in their growth strategies as a result. According to data from Accenture, 84 percent of business executives believe they need to use AI to achieve their growth objectives. However, 76 percent acknowledge that they struggle with how to scale AI across their business.

The data proves that business executives will continue to leverage AI and ML, but there's still room for improvement in terms of how they apply it and use it effectively at scale. For C-Suite executives and those looking to prioritize these emerged technologies in the New Year, here are the four predictions I have specific to AI and ML:

The democratization of AI/ML is a major trend that will have a significant impact on enterprises and the kinds of solutions being delivered to customers.

In the past, ML was practiced by a select few. For instance, business leaders would have an individual from their IT department dedicated to AI/ML and have no additional visibility into the technology. To paint a better picture of what I mean: ML was ‘magic practiced by highly trained wizards' and would just ‘appear' for an organization. But now, ML is transitioning to become a tool in every engineer's and product manager's toolbelt, as well as a tool for customer support and other allied roles.

In the future, ML will be a common way to solve computational problems, accessible to most product professionals in the company. AI and ML will continue to have an impact for those in the financial services, insurance, healthcare, telecommunications industry and more. To list a few benefits: AI and ML can personalize customer experiences, allow enhanced insight into customer behavior and improve risk analysis.

As more people understand ML and can deliver it themselves, the applications will be more successful across an entire organization.

As the democratization of ML moves this technology into many hands across an organization, there will need to be better platform support. A framework must be in place to ensure the success of the "ML everywhere" approach that companies are executing on.

Furthermore, as the variety of those contributing to an ML solution continues to grow, the platforms will not only need to support heterogenous ML technologies, but a true diversity of skill sets. However, deploying and managing that diversity is a challenge that has yet to be solved, so we can expect the call for better platform support to only increase in 2022.

3. Data availability and security will be a driving concern for CIOs

With the rise of cybersecurity and stringent data quality requirements within an organization, it's no surprise that data availability and security will continue to be a driving concern for CIOs and IT decision-makers.

As data is central to AI/ML practices, there will continue to be a focus on improving data quality to drive better supervised models and an emphasis on metadata so existing data can be reused on new problems in automated or semi-automated ways.

We can expect organizations to go to lengths to ensure data is high-quality, secure and available for use.

4. DevOps will get more complicated

DevOps is getting more complicated with the addition of Machine Learning to the stack, especially for those in IT and IT decision-maker roles. As DevOps combines software development with IT operations, the integration of ML is muddying the waters. "MLOps" is the new moniker, however, there is nowhere near a ‘best practice' for the addition of this technology in the stack.

"MLOps" can also be challenging due to the complexity of its data. While DevOps focuses primarily on software and code (and processes around that code), an ML model is created from both software and data. Therefore, different models can be created as the data changes, even if the software doesn't change. As this data can be large and dynamically changing, it can add a layer of complexity to managing all artifacts within a model.

With their different facets, DevOps and ML are inherently different. However, the synergy of the two can ensure success for an organization. IT leaders will be tasked with determining the most effective way that DevOps and ML can intersect.

For businesses already implementing AI and ML, or looking to apply these technologies, it's important to keep a pulse on the trends affecting these areas. As we look ahead to the New Year, business leaders should take note of these key themes and the impact they will have on their business operations.

##

ABOUT THE AUTHOR

Brian Bartell, PhD, is a text analytics and machine learning specialist, accomplished software professional, and entrepreneur. He thrives simultaneously as an executive leader and strategist, and also as a hands-on innovator in advanced technology. He has founded, or as CTO has been instrumental in the growth of, a variety of successful technology startups. This includes: Contract Wrangler Inc (legal analytics), Notiora Inc (text analytics in life sciences and travel), Covario Inc (search engine analytics), ContentScan Inc (academic meta-search), and GeneCraft (gene cloning analytics). Brian's doctorate is in Computer Science from UC San Diego specializing in learning algorithms for natural language processing.