How prepared are businesses to take full advantage of the insights that artificial intelligence affords? The tools may be ready, and talented people may have come onboard, but it’s likely there’s a gap in the data. Yes, there is plenty of data flowing through enterprises, but harnessing it in a productive and unbiased way is another story.
At this point, only 24% of organizations consider themselves to be data-driven, and only 21% have what can be considered “data cultures,” a new survey of senior data and analytics executives out of Wavestone NewVantage Partners finds. In addition, only 24% of companies report they are doing enough to ensure responsible and ethical use of data within their organizations and the industry. “Becoming data-driven is a long and difficult journey that organizations increasingly recognize playing out over years or decades,” the study’s authors, Tom Davenport and Randy Bean, point out. “Companies continue to fall short in attention and commitment to data ethics policies and practices.”
The data gap is likely the most pressing issue affecting AI success, agrees Mona Chadha, director of category management at Amazon Web Services. “There are issues that companies need to be aware of, such as poor data quality, unfair bias, and lax security to name a few,” she states. “Quality of predictions of AI models depends strongly on the data used to train the models. Poor data quality can result in inaccurate results and inconsistent model behavior, leading to lack of trust from customers and internal stakeholders.”
Data bias and security also are issues that need to be tackled in AI, Chadha continues. “It’s easy to fall prey to the assumption that AI can make decisions more impartially than people can. Unfair biases, present in the data used to train AI models, can result in discriminatory behavior that can put businesses at risk. Attackers are constantly trying to exploit AI vulnerabilities. Businesses must ensure that AI systems are protected against adversarial attacks across their data and algorithms.”
When it comes to data quality, organizations need to focus on the processes employed to oversee their data assets. “Existing data often resides in multiple databases and data warehouses, which often contains duplicates, outliers, and irrelevant data points,” Chadha states. “There are also gaps in the existing datasets. Organizations need better tools to clean and label the data. Poor data quality can result in inaccurate results and inconsistent model behavior, leading to lack of trust from customers and internal stakeholders.”
Once the data gap is closed, businesses can start to build their business cases for moving AI forward. “As AI gains traction, a number of business use cases, across industries, are seeing the results,” Chadha relates. “Examples include driving product innovation by speeding up drug discovery and the training of autonomous vehicles to navigate complex traffic scenarios. AI is supporting risk mitigation by helping to combat financial fraud and reduce unplanned downtime for industrial equipment. Consumers are also seeing an improved user experience with AI driving content engagement through recommendation engines or improving customer service by using AI to assist human agents. Lastly, AI is making great strides in overall efficiency and safety improvements by helping the manufacturing sector through computer vision.”
(Disclosure: Over the past year I have conducted project work for AWS, mentioned in this article, in my capacity as an independent analyst.)