Intelligent CIO APAC Issue 22 - Page 76

t cht lk

t cht lk

Many firms will increasingly rely on external consultants to meet their business goals .
respondents ( 75 %) are still exploring or are struggling to operationalize AI / ML models .
More than one-third ( 32 %) of respondents report Artificial Intelligence R & D initiatives that have been tested and abandoned or failed . The leading causes for these failures included poorly conceived strategy ( 43 %), lack of data quality ( 36 %), lack of productionready data ( 36 %), and lack of expertise within the organization ( 34 %).
The failures underscore the complexities of building and running a productive AI and Machine Learning program . Upon closer scrutiny , businesses are struggling with their AI / ML efforts for several reasons , which include :
• Lack of organizational collaboration – Designing the right Machine Learning training and AI algorithms requires a holistic understanding of the data and processes being automated across organizational boundaries . Lack of collaboration often yields a poor implementation , lower-quality data and rejection of the applications / automation projects by key parts of the organization .
• IT and business process immaturity – IT and business processes should be well-formed to ensure the quality of data and seamless AI / ML execution . Also , AI / ML is best served with rapid iterations and improvements in the data and algorithms – something that happens most effectively in a DevOps culture .
• Lack of expertise in mathematics , algorithm design or data science and engineering – Since AI and Machine Learning are built on highquality , timely data and well-formed algorithms – representing the best in processes and models of the real-world – skills are critical . Finding the talent is tough in today ’ s market .
• Failure to get the right data to the right app or point-of-analysis in real-time – A company ’ s Machine Learning training is only as good as the data that is fed into the AI / ML frameworks and intelligent applications . If the data is bad , old or incomplete , the training will be poor and the answers and results generated will be equal to the quality of the data .
To overcome these challenges , organizations can take the following steps :
Step 1 : Build the foundation
Start by preparing data and applications to migrate to the right multi-cloud and data architecture
76 INTELLIGENTCIO APAC www . intelligentcio . com