Organizations must determine precisely which types of processes will benefit most from an injection of machine and deep learning algorithms.
As organizations head into 2020 it’s already clear various forms of artificial intelligence (AI) are transforming everything from how IT is managed to how business processes are constructed and optimized. However, while more organizations than ever are starting to invest in AI it’s also becoming apparent that early AI adopters are starting to temper their expectations.
See also: AI Job Skills In Big Demand
Multiple surveys of IT leaders indicate that overall investments in AI will accelerate in 2020. For example, an IBM survey of 4,514 businesses in the U.S., European Union, and China finds 75% of respondents either have deployed some form of AI (34%) or are exploring it (39%).
A survey of 700 business and IT leaders conducted by Microsoft also finds organizations are going way beyond simply using AI made available by various IT vendors. Just under a third (31%) are using AI to create their own intellectual property, the survey finds.
However, early adopters are learning that AI projects require a more comprehensive approach to implement that is causing many of them to re-evaluate their approach. A survey of 1,062 business and IT executives conducted by PwC finds only 4% of respondents say they plan to deploy AI enterprise-wide in 2020, a significant drop from the 20% who said they planned to do so in 2019.
Interest in AI isn’t so much declining as much as there is simply a deeper appreciation for the level of process and cultural change required to succeed, says Anand Rao, global AI lead for PwC.
Organizations are finding they need to not only develop best machine learning operations (MLOps) to make sure they are using the right data to operationalize AI models, they also to need to better understand how AI models will ultimately impact the business, says Rao.
“This is both a data and culture issue,” says Rao.
The goal should be to make certain that AI models are not only explainable but also that decisions made about what data to employ are not inadvertently introducing biases to the AI model being employed, notes Rao.
Regardless of the rate at which AI is rolled out across the enterprise, the adoption of AI is part of a larger watershed moment in the history of IT, says Deb Cupp, corporate vice president for worldwide enterprise and commercial industries at Microsoft.
Rather than solely relying on IT vendors to create solutions using enabling technologies such as AI and augmented and virtual reality technologies on their own.
“We’re going to be seeing a lot more breakthrough innovations coming from customers,” says Cupp.
The challenge organizations now face is trying to foster the level of technology intensity required insider their organizations to drive those innovations. There may be no way to precisely measure the level of intensity required, but Cupp says the need for organizations to be more IT-savvy is one reason more organizations are doing everything from setting up dedicated innovation teams to sponsoring hackathons that encourage personnel to experiment with potential new offerings and processes.
Overall, organizations would be well-advised to spend some time determining where best to apply AI to a repetitive process in a way that drives the most amount of business value over a defined period of time, says Rob Thomas, general manager for IBM Data and AI.
Instead of attempting to boil the proverbial ocean, Thomas says most organizations are going to better off focusing on discrete initiatives that will enable them to gain AI expertise. In fact, 37% of respondents in the IBM survey cite limited AI expertise or knowledge as a hindrance from successful AI adoption at their business, followed by increasing data complexities and silos (31%) and lack of tools for developing AI models (26%).
The most important thing is to get started gaining that expertise now or risk falling behind, says Thomas. Otherwise, rivals will gain a level of AI expertise that will prove insurmountable, adds Thomas.
“You can’t take a wait-and-see approach to AI,” says Thomas.
Autonomous vehicles may not be pervasive any time soon, but AI is various forms is becoming ubiquitous. It’s really more a question of in what form AI will manifest itself across a wide range of increasingly automated business processes. The challenge and opportunity organizations face now are to determine precisely which types of processes will benefit most from an injection of machine and deep learning algorithms that hopefully won’t be more trouble to manage than they’re ultimately worth.