DATA & ANALYTICS
Royles thinks that getting research and prototypes out of the lab is the biggest barrier to success . He says that it ’ s important to make this easy by having clear access to the raw materials , as well as the build processes for an AI service . “ And engaging with stakeholders on what this really means in respect to business impact ,” he adds . “ For example , at Cloudera we package our research and patterns into Applied Machine Learning Prototypes ( AMPS ), which organisations can deploy against their own systems with just a few clicks .”
He says that being able to scale inference , fine-tuning and training across the parallel compute and GPU resources required , must be inherently available to your practitioners , and organisations should have the choice as to where and how this is provisioned - on both public and private cloud .
He says : “ Ensuring trust in your data through strong governance and high durability is vital , necessitating investment in data governance practices and robust data collection .
“ Through enhanced interpretability , businesses can employ explainable AI techniques that provide transparency and insights into model decisions . Also comprehensive auditing of how humans interact with models can help understand and fine tune the models themselves and drive continuous improvement . This is often referred to as building a data flywheel .
“ Finally , driving adoption requires effective change management efforts , including comprehensive employee training and fostering a data-driven culture . By proactively addressing these challenges , businesses can successfully implement AI-driven data analytics projects , gaining valuable insights and a competitive edge in their decision-making processes .”
How AI and analytics can help organisations in making data-driven decisions : Best practices for integrating AI into the decision-making processes According to Royles , AI and data analytics provide organisations with the ability to make data-driven decisions . He says : “ To effectively integrate AI into the decisionmaking processes , several best practices should be followed .” He says the first is essential and that ’ s to define clear objectives for the decision-making processes and identify specific questions or problems that AI and analytics can address . Organisations must be clear on what they want the outcome to be .
“ If an AI service is too slow , inaccurate , or responds in ways that don ’ t add value , users will quickly lose trust , and it will be very hard to recover ”
CHRIS ROYLES EMEA FIELD CTO | CLOUDERA
96 August 2023