Industry executives and experts share their predictions for 2022. Read them in this 14th annual VMblog.com series exclusive.
Three Foundational AI Developments in 2022
By Wilson Pang, CTO, Appen
AI has the potential to transform societies in ways we cannot yet imagine, dwarfing the Internet and cloud and perhaps rivaling the printing press and electricity. However, AI isn’t easy, and reaching its full promise will take many years of incremental advances that allow more organizations and agencies to mature their AI capabilities and launch more AI projects, from the most rudimentary process acceleration to life-saving research.
Let’s look at the top three development predictions that we see being the central focus for 2022. Each provides an important foundational step forward in the ability of organizations to understand how to do more with AI and do AI better.
Rise of synthetic data
In 2022, we expect to see many more companies experiment with and adopt generative AI as a method of ML model data collection. Generative AI is using synthetic data – a type of data created using machine learning models instead of, or combined with real-world data – to train ML models. Generative AI is already being used to address key AI challenges, such as generating 3D worlds for AR/VR, as well as for training autonomous vehicles and for pharmaceutical research. Early implementations of generative AI technology also let companies do things like identify marketing content more accurately and leverage highly nuanced NLP capabilities to diagnose health issues through text and image data.
By 2025, Gartner expects generative AI to account for 10% of all data produced, up from less than 1% today. Gartner has also forecasted that by 2024, the use of synthetic data and transfer learning will halve the volume of human-generated data needed for machine learning – reducing costs and accelerating ROI for many AI use cases, especially when initially training a model. The combination of human-generated data and synthetic data will expand the possibilities of use cases and lead to successful models. Exploring synthetic data use cases will become critical in 2022, so that organizations learn and identify their needs ahead of the predicted adoption.
Acceleration of internal efficiency use cases
According to the Appen 2021 State of AI report, AI budgets are on the rise, with 74% of respondents reporting AI budgets over $500K. Budgets from $500K to $5 million have increased by 55% year over year, with only 26% reporting budgets under $500K. The good news is that most business leaders (67%) said AI projects have “shown meaningful ROI”, and the No. 1 AI use case was supporting internal operations (62% of respondents).
In 2022, this move toward internal-facing use cases will lead to a much deeper understanding of AI and set the foundation for more successful projects by compelling the organization to clearly identify how data moves through the business and what happens to it as it does. This will play out in two important ways. First, to fully enable these use cases, enterprises will focus more attention on deploying platforms that enable them to eliminate data silos and centrally manage all data. Second, to ensure the success of these initiatives, they will work internally or with partners to develop strategies focused on being able to manage the entire “data for AI” lifecycle. This hard work internally will set the stage for the strategically important external-facing AI use cases as the enterprise reaches new stages of AI maturity.
Model evaluation and tuning goes mainstream
In 2022, the need for regular model evaluation and tuning becomes AI program table stakes. Machine learning models are dynamic – they can’t be deployed and forgotten. ML models in production need to be updated and retrained based on a variety of factors, including the ongoing results, as well as changes in infrastructure, data sources and business models. According to the Appen State of AI report, 87% of organizations update their models at least quarterly, up from 80% last year, with 57% updating their models at least monthly; 91% of large organizations update their models at least quarterly, and organizations that use external data providers are most likely to update their models at least monthly.
In 2022, given the overall maturing of the AI industry – Gartner found that at least 40% of businesses already had some form of AI program last year- enterprises will shift focus from implementation to optimization, resulting in increased reliance on model evaluation tuning and solutions and the vendors that can assist in this process. Like the first two predictions above, this optimizing of AI is critical to the more advanced use cases we will see in the coming years.
AI cannot predict the future (yet), but I can predict that 2022 will be a very exciting year in the AI industry.
##
ABOUT THE AUTHOR
Wilson joined Appen in November 2018 as CTO and is responsible for the company’s products and technology. Wilson has over 17 years’ experience in software engineering and data science. Prior to joining Appen, Wilson was Chief Data Officer of CTrip in China, the second largest online travel agency company in the world where he led data engineers, analysts, data product managers, and scientists to improve user experience and increase operational efficiency that grew the business. Before that, he was senior director of engineering in eBay in California and provided leadership to various domains including data service and solutions, search science, marketing technology, and billing systems. He worked as an architect at IBM prior to eBay, building technology solutions for various clients.
