The majority of organisations globally lack the internal resources to support critical artificial intelligence and machine learning initiatives, according to a new study from Rackspace Technology.
The survey, Are Organisations Succeeding at AI and Machine Learning? was conducted in the Americas, Asia Pacific and Japan and EMEA (Europe, the Middle East and Africa) regions of the world, and indicates that while many organisations are eager to incorporate AI and ML tactics into operations, they typically lack the expertise and existing infrastructure needed to implement mature and successful AI/ML programmes.
“This study shines a light on the struggle to balance the potential benefits of AI and ML against the ongoing challenges of getting AI/ML initiatives off the ground,” Rackspace says.
“While some early adopters are already seeing the benefits of these technologies, others are still trying to navigate common pain points such as lack of internal knowledge, outdated technology stacks, poor data quality or the inability to measure ROI.”
Participants of the survey in the APJ region rated themselves slightly higher at 18% compared to global statistics a 17% for advanced maturity in AI/ML). APJ participants were more likely to be using AI/ML in more applications and use cases, and are spending significantly more on average than global participants ($1.3 million vs $1.06 million). Respondents in the APJ region also noted seeing more benefits of their AI/ML efforts such as increased productivity and better streamlined processes.
Interestingly, the APJ region rates the inability to find the right data as a bigger challenge (26%) compared to global participants (22%). Also, one of the biggest impacts of AI/ML for businesses in APJ has been the blurring of lines between human and technology factors, which is 5% higher from what the global respondents have stated. .
Additional key findings of the report for Singapore respondents include::
Organisations are still exploring how to implement mature AI/ML capabilities. A mere 25% of respondents report mature AI and machine learning capabilities with a model factory framework in place. In addition, the majority of respondents (75%) said they are still exploring how to implement AI or struggling to operationalise AI and machine learning models.
AI/ML implementation fails often due to lack of internal resources. More than one-third (32%) of respondents report artificial intelligence R&D initiatives that have been tested and abandoned or failed. The failures underscore the complexities of building and running a productive AI and machine learning programme. The top causes for failure include poorly conceived strategy (43%), lack of expertise within the organisation (34%), lack of data quality (36%) and lack of production-ready data (36%).
Successful AI/ML implementation has clear benefits for early adopters. As organisations look to the future, IT and operations are the leading areas where they plan on adding AI and machine learning capabilities. The data reveals that organisations see AI and machine learning potential in a variety of business units, including operations (68%), IT (57%), customer service (45%), and Supply Chain Management (45%) . Further, organisations that have successfully implemented AI and machine learning programmes report increased productivity (47%) and increased understanding of your business and customers (42%) as the top benefits.
Defining KPIs is critical to measuring AI/ML return on investment. Along with the difficulty of deploying AI and machine learning projects comes the difficulty of measurement. The top key performance indicators used to measure AI/ML success include, revenue growth (69%), data analysis (66%), and Process enhancement/ improvement (66%).
Organisations turn to trusted partners. Many organisations are still determining whether they will build internal AI/ML support, or outsource it to a trusted partner. But given the high risk of implementation failure, the majority of organisations (66%) are, to some degree, working with an experienced provider to navigate the complexities of AI and machine learning development.