Published by Jessica Weisman-Pitts
Posted on February 8, 2023
By Mohit Joshi, President, Infosys
We do expect AI adoption to escalate over the next few years as awareness continues to grow around the significant benefits that can be achieved, especially that of incremental profit which is a notable benefit of AI adoption. To reach this point, it is critical that companies look to improve their data practices, foster trust in advanced AI and integrate AI with their business operations.
Low data sharing preference amongst U.K. organizations does point to underutilised AI value and it’s important for business leaders to start to think differently about data and AI overall. If companies transform data management to data sharing and embrace the data-sharing economy, we know that it generates greater value from their data, and that data increases in value when treated like currency and circulated through hub-and-spoke data management models. Companies that continue to refresh data with low latency also generate more profit, revenue, and subjective measures of value.
From our research we found that talent was not a significant concern or challenge regarding deploying or scaling AI. Our report showed that in order to handle the forthcoming wave of AI, other factors played a more crucial role, for instance the need to focus on the customer experience, core operations and even the process. Where process is concerned for example, our report showed that even with clean data, AI practitioners face a number of challenges starting with missing data. When we asked how the practitioners handle that missing data, we found that 34% of survey respondents said they handle missing data in ways that can allow bias to creep into AI models – an issue that obviously needs minimizing.
When we look at how companies put AI to work on their data, we see real value in bringing a multidisciplinary team together to frame the problem and prepare the data. This is a natural opportunity for companies to challenge experienced employees to work together differently and learn new skills.
Our research showed that a number of factors contribute to building a resistant AI team, with one being trust across all dimensions. We know that the best data and the most elegant AI system can be developed but unused, unless there is a belief that the system is effective, fair and adaptable by those using it within an organisation. Following through on this theme, it’s also critical that companies develop strong ethics and bias management systems as part of their resilience strategy. Our survey categorically shows that by doing so companies have reported higher satisfaction and trust for their data and AI use cases. Another component for a resilient AI team is to ensure there is a robust AI risk management practice in place, one that will ultimately be a contributing component of any overall enterprise risk management process. Last but not least, we know from the research that high-performing AI teams should not only include the obligatory data scientists, but business leaders too such as business or problem-solving experts and senior executives.
The research found that clean data and AI infrastructure were the top challenges in fielding AI systems. Furthermore, three out of four respondents of the survey were found to want to scale AI across their organisation, but this was being limited mostly by high costs and data. Experts tell us that scaling requires building trust in AI, and currently this appears to be a blind spot for most enterprises.
Infosys assessed the state of data and AI in business in this report. We wanted to understand what works for businesses currently. We now use these insights to guide customers on data management strategies and AI cloud systems.
Explore more articles in the Technology category
