Scaling AI: a key route for recovery

However, the demand for AI-based technologies is soaring, even among the general population. Almost two thirds of consumers expect to increase the use of touchless interactions through voice assistants, facial recognition, or apps to avoid human interactions and touchscreens post-COVID-191. These technologies, of course, rely on scaled AI.

Businesses need to understand the impact that AI can have on the bottom line and how it can support their recovery strategy as they bounce back from the impacts of COVID-19. Adding AI to business processes speeds up decision making and creates the essential companion for symbiotic operations. When used correctly, the true value of AI is that people can delegate not just processes, but also decision making, and this has a significant impact on efficiency and productivity. But, as many organizations have already discovered, implementing AI at scale is a challenging journey.

Data quality, skills gaps and ethics are all vital considerations to building an AI-powered enterprise. Our recent research also found that seven in ten organizations find a lack of mid to senior-level talent a major challenge for scaling AI. Less than one third of struggling AI organizations feel they have a detailed knowledge of how and why their AI systems produce the output they do. Moreover, nine out of ten organizations believe that ethical issues have resulted from the use of AI systems over the last 2-3 years.

To overcome these barriers and harness the power of AI, there are four principles essential to successful implementation which organizations must consider:

Strategize: While the promise of AI might make IT leaders excited to delve straight into implementation, strategy is key. It’s important to think beyond just the short-term goals of implementing AI and consider what the potential goals are for the next 3-5 years. Laying the necessary foundations before commencing wide-scale deployment is also essential. AI needs access to vast amounts of quality proprietary and third-party data which has to be stored and managed properly; deciding how to do so is a fundamental building block to effective AI implementation.

However, legacy IT systems can delay the collection, analysis and understanding of data. To resolve this, data needs to be managed as a strategic asset in an organization. Establishing data governance to design, set up, scale, and continuously monitor the data in a firm has clear benefits in supporting the scaling of AI use cases. IT estate modernization also addresses the challenges of fragmented and legacy IT systems and provides faster access to information within a secure environment.

Ethical considerations must also be woven into an organization’s strategy from the outset. Customer and employee privacy in particular are a prerequisite. The mindset among IT leaders must be one of transparency, accountability and fairness, building AI systems with ethics-by-design in mind. To do this, the right governance structures must be put in place, as well as building diverse teams to prevent AI bias, with the aim of empowering people in the knowledge that they are interacting with AI – and that the AI systems themselves are trustworthy.

Operationalize: By creating a tiered responsibility system, organizations can ensure that AI implementation is pushed forward at a steady pace. It’s recommended to have a central team for

policy and strategy; a center of excellence (CoE) for optimizing resources, embedding ethics and facilitation of ideas; helping to weave AI initiatives into the enterprise’s wider business goals. After all, AI initiatives are not scaled in silos. They impact multiple business units, so wider involvement makes sense if the aim is to have buy-in across the business.

Nurture: AI requires a new host of skill sets within an organization, from data architects to designers to data scientists. To keep implementation initiatives on track for the long haul, business leaders must also consider a range of business and change management roles including data strategists, AI ethicists and process and automation engineers.

While data literacy is important, so are the soft skills to communicate the importance of the new technology to an organization and make sure the whole business is on board. Training and upskilling are key here. Given the complexity of achieving scaled AI many organizations choose to work with service providers to address the challenges of this structural and cultural transformation. This can help to alleviate the workload for internal teams while also bringing fresh perspectives, knowledge, and guidance to AI initiatives.

Monitor: AI models cannot be left to run without intervention. Variations in the nature of data and new information can change outcomes, leading to mistakes and vulnerabilities in AI algorithms. In response, organizations should systematically rate AI models based on the likelihood of them making mistakes and determine appropriate action plans, such as the frequently of monitoring or updating, or which modelling technique to implement.

Once these four principles have been established, businesses must create a collaborative digital workplace that’s fit to embrace AI at scale. This means eliminating information silos and enabling self-organizing teams for better collaboration and faster decision making. Combining AI with agile cognitive systems makes business processes more automated and intelligent, driving both decision-making and execution forwards to boost performance.

COVID-19 has put the onus on organizations to embrace AI faster than before. Adopting AI is a complex journey, but the benefits are transformational both in the short- and long-term. No matter where organizations are on their journey, they must invest now to build resilience and agility – and so the future can be AI-enabled.