This IDC Survey looks at key Canadian barriers to putting artificial intelligence (AI) pilots into operational production. The negative impacts of the directly identified barriers are as follows:
- Getting the data for use in operational deployment was much harder than getting it for the pilot.
- Privacy and/or regulatory issues emerged for operational deployment that we didn't anticipate during the pilot.
- The AI model was accurate during the pilot but inaccurate when we tried to deploy it in production.
- Targeted users didn't find the output of the AI model relevant/beneficial.
- Ensuring the accuracy of the data the AI model uses was too difficult/required too much time when we tried to deploy it in production.
This was magnified by Canadian organizations' failure to implement formalized data governance and management frameworks and practices. The survey results also show that outmoded, manual data-handling practices are another key Canadian barrier against putting AI pilots into operational production.