Artificial intelligence (AI) and machine learning (ML) technologies are the next set of transformative technologies in the tech sector. And while much as been prophesized about their impact, data points are few and far between. 451 Research's Voice of the Enterprise (VotE): AI & Machine Learning survey is a new biannual offering that addresses the need for quantitative metrics around these emerging technologies. It provides insight into the adoption patterns, as well as benefits, barriers and applications of these critical technologies. (Similar to a recent report on industry-specific use cases), this Spotlight leverages the results of the most recent survey to expose the primary strategic approaches that enterprises are employing to adopt AI and ML technology. The data comes from responses from over 1,600 line-of-business and IT professionals in North America, EMEA and Asia-Pacific.
The 451 Take
AI and machine learning adoption is not monolithic. In fact, according to new data from 451 Research's VotE: AI and Machine Learning survey, enterprise adoption strategies segment into four categories. These distinct approaches have different levels of popularity, which vary depending on factors such as the size of the company, its willingness to invest in new technologies, and its reason for using ML in the first place. Understanding which customers are deploying which strategy and why will be integral to the success of the many vendors with machine learning products and services.
Just as AI and machine learning are multi-use technologies that can automate and improve a variety of processes, the route to adoption of the technology can itself take many forms. At 451 Research, we segment these approaches based on two conditions:
1. Build or buy: Is the machine learning application assembled from component pieces by in-house experts or is it a product or service procured from outside vendors?
2. Bespoke or broad: Are the machine learning tools and solutions tailored to an enterprise's use case or do they consist of general, off-the-shelf offerings?
The figure below depicts the strategic approaches to ML adoption using these two axes to segment four quadrants.
1. Build or buy: Is the machine learning application assembled from component pieces by in-house experts or is it a product or service procured from outside vendors?
2. Bespoke or broad: Are the machine learning tools and solutions tailored to an enterprise's use case or do they consist of general, off-the-shelf offerings?
The figure below depicts the strategic approaches to ML adoption using these two axes to segment four quadrants.
1. Service providers: Respondents in this category are primarily working with third-party service providers such as Accenture, Cognizant and Wipro that in turn build custom machine learning applications for the customer. The survey data shows that only 11% or respondents are pursuing this strategy, making it the least popular of the four strategies. We suspect that many companies think they can build ML systems on their own that are better and cheaper, or maybe they worry about sharing sensitive data with outside organizations. Therefore, it's not surprising to find that the service-provider approach is the most popular among respondents who describe their overall strategy toward new technology as 'conservative.'
2. Machine-learning-enhanced software: The adoption strategy of these organizations involves purchasing applications with built-in ML capabilities from vendors such as Adobe, SAP, SAS Institute and Salesforce. These applications cover a breadth of different use cases, from intelligent CRM offerings to vertical-specific products. The minimal technical expertise required to adopt these systems perhaps explains why a plurality of survey respondents – 38% – said this was their primary approach to adopting the technology. This approach is particularly popular among respondents whose primary use case is related to customer service, probably due to the relatively maturity of this segment of the ML software market.
3. Open source machine learning tools: This set of respondents is employing a much more DIY approach to bringing ML into their organizations. Overall, 27% of respondents cite the use of open source ML tools such as Keras, Spark ML and TensorFlow. This approach is the most preferred among larger organizations (10,000+ employees), probably because they have the financial and human resources necessary to build an ML application from scratch.
4. Vendor-provided machine-learning-specific tools: These respondents are taking advantage of an emerging class of component products and services provided by vendors such as AWS, Google, H20.ai and DataRobot that can be leveraged as part of an end-to-end ML application. The survey data shows that only 24% of respondents currently cite this approach as their primary adoption strategy, although other data leads us to think that this segment will expand in the future. Currently, deploying vendor-provided ML tools is the most popular approach among respondents whose primary use case is related to business analytics, most likely reflecting the rise of citizen data scientists and self-service analytics – trends that analytics providers with ML capabilities have also been focusing on.
2. Machine-learning-enhanced software: The adoption strategy of these organizations involves purchasing applications with built-in ML capabilities from vendors such as Adobe, SAP, SAS Institute and Salesforce. These applications cover a breadth of different use cases, from intelligent CRM offerings to vertical-specific products. The minimal technical expertise required to adopt these systems perhaps explains why a plurality of survey respondents – 38% – said this was their primary approach to adopting the technology. This approach is particularly popular among respondents whose primary use case is related to customer service, probably due to the relatively maturity of this segment of the ML software market.
3. Open source machine learning tools: This set of respondents is employing a much more DIY approach to bringing ML into their organizations. Overall, 27% of respondents cite the use of open source ML tools such as Keras, Spark ML and TensorFlow. This approach is the most preferred among larger organizations (10,000+ employees), probably because they have the financial and human resources necessary to build an ML application from scratch.
4. Vendor-provided machine-learning-specific tools: These respondents are taking advantage of an emerging class of component products and services provided by vendors such as AWS, Google, H20.ai and DataRobot that can be leveraged as part of an end-to-end ML application. The survey data shows that only 24% of respondents currently cite this approach as their primary adoption strategy, although other data leads us to think that this segment will expand in the future. Currently, deploying vendor-provided ML tools is the most popular approach among respondents whose primary use case is related to business analytics, most likely reflecting the rise of citizen data scientists and self-service analytics – trends that analytics providers with ML capabilities have also been focusing on.
Toward a Future of Multifaceted Machine Learning Strategies
The segmentation above demonstrates that there is clearly more than one way to implement AI and machine learning in an enterprise context. In fact, other data from the survey suggests that many respondents are currently using more than one of the four strategic approaches outlined above, and most anticipate that their future ML initiatives will involve some combination of these strategies. The future of ML adoption will be fundamentally multifaceted.
This conclusion has important implications for both customers and vendors. For adoptees looking to accelerate their AI initiatives, they should not fear trying new strategies, and new adopters should weigh the variety of techniques available to them. Vendors, on the other hand, need to keep their eyes peeled on customer needs – they must be aware that customers are willing to try various approaches to machine learning, and adjust their product and service portfolios accordingly.
This conclusion has important implications for both customers and vendors. For adoptees looking to accelerate their AI initiatives, they should not fear trying new strategies, and new adopters should weigh the variety of techniques available to them. Vendors, on the other hand, need to keep their eyes peeled on customer needs – they must be aware that customers are willing to try various approaches to machine learning, and adjust their product and service portfolios accordingly.
Nick Patience
Founder & Research Vice President
Nick Patience is 451 Research’s lead analyst for AI and machine learning, an area he has been researching since 2001. He is part of the company’s Data, AI & Analytics research channel but also works across the entire research team to uncover and understand use cases for machine learning. Nick is also a member of 451 Research’s Center of Excellence for Quantum Technologies.
Jeremy Korn
Senior Research Associate
Jeremy Korn is a Senior Research Associate at 451 Research. He graduated from Brown University with a BA in Biology and East Asian Studies and received
Keith Dawson
Principal Analyst
Keith Dawson is a principal analyst in 451 Research's Customer Experience & Commerce practice, primarily covering marketing technology. Keith has been covering the intersection of communications and enterprise software for 25 years, mainly looking at how to influence and optimize the customer experience.