AI Talk: Scaling AI tools into your key processes

Views on news 

AI takes many forms: machine learning, deep learning, predictive analytics, natural language processing, computer vision and automation. Companies must start with a solid foundation and a realistic view to determine the competitive advantages an AI implementation can bring to their business strategy and planning. The article outlines the benefits and drawbacks businesses might experience when adding AI to their environments, discusses the prerequisites needed to integrate AI in a business setting and provides the incremental steps to follow for a successful implementation. Many failures happen because businesses don’t realise how much discipline it requires to take AI from the experimental to the operational phase.  

 

How to achieve data readiness 

To bridge the gap that exists between business problems and technology, the first step to a digital transformation should be gathering input from all levels of the organisation regarding what the pain points are that AI deployment should address. Then it should be mapped out where the organisation’s foundation readiness stands and which areas for development should be prioritised.  

 

Data quality is the number one barrier to AI adoption., but other factors, such as the overall infrastructure of the company, also have an impact. Data used by AI models should also be protected with guardrails and via data governance. Building a model seems to be the easiest piece now, it’s making a model that users can trust that is the biggest challenge, which can be achieved by building in guardrails for what should be done when the model is wrong. Platforms and shared services play an increasingly important role in AI deployments and have the potential to accelerate the journey from pilots to scaled outcomes.  

 

The panel’s advice 

• Here is the sequence to follow for AI projects: value – feasibility – risk – time to implement. 
• Data quality and quantity are equally important – for data analytics to work, you need large sets of data, although, with a reasonable amount of data you can always synthetise or buy more.    
• Not all business problems require AI for a solution.  
• Prioritise low effort, high RoI AI projects and make sure they align with the company strategy.  
• Don’t scale your running AI projects at the same time.