Despite widespread enthusiasm for AI, with 95% of companies already using the technology and nearly all expecting to do so in the future, most organizations have struggled to scale beyond pilot projects. According to a report produced by Boomi in collaboration with MIT Technology Review Insights, 76% of companies have implemented AI in only one to three use cases, with just half expecting to fully deploy AI across all business functions within the next two years. The report found a critical factor for successful AI deployment is data liquidity, which enables organizations to seamlessly access, integrate and analyze data from diverse sources.
However, data quality remains a significant barrier, particularly for larger companies with extensive legacy IT infrastructure.
Half of the respondents identified data quality as the most limiting factor in AI deployment, with larger enterprises more likely to face these challenges.
“Inadequate data quality is a major limitation for AI deployment,” explained Boomi CTO Matt McLarty.
He said despite its unprecedented boom, generative AI is a relatively new technology, and organizations are still learning to manage the risks that come with it, including producing inaccurate information and “hallucinations”.
“These risks can harm customers and expose companies to legal liability,” he cautioned. “For example, Air Canada was recently held liable for damages caused by misleading information provided by an AI chatbot.”
Survey results also indicated companies are cautious about rushing into AI, with 98% indicating a preference to delay adoption if it ensures safer and more secure implementation.
Factors Slowing AI Deployment
Governance, security and privacy concerns are the primary factors slowing AI deployment, cited by 45% of respondents, and even more so by 65% of those from the largest organizations.
Matt McLarty, Boomi CTO explained because GenAI technology has such broad applicability, it’s taking companies a while to figure out the best place to apply it.
He noted some of the easiest applications – like language summarization and generation – don’t end up driving much business value at scale.
Areas that have the most potential business impact – like higher levels of business automation and more personalized customer experiences – require more complex solutions that combine Gen AI technologies with existing data and systems.
“The organizations best equipped to deal with this complexity are the ones who have highly accessible data and connectable applications that allow them to ground the new AI technologies and more easily execute automated tasks,” he said.
McLarty added inaccurate information and hallucinations can also contribute to heightened bias in AI models.
In financial services, for instance, AI-powered credit risk assessments could help mitigate systemic discrimination in credit applications — but only if the data used to train these AI systems is itself reliable, accurate and free of distortions and biases.
“Similar biases can crop up in health care, policing and advertising, where algorithms can accentuate gender biases or overlook underrepresented groups,” he said.
Despite the challenges, spending on AI readiness is poised for a significant increase in 2024, after modest or flat growth in the previous two years.
Investing in Data Readiness
Nine out of ten companies said they plan to boost investments in data readiness — covering areas such as platform modernization, cloud migration and data quality — as well as in adjacent fields like strategy and business model innovation.
Forty percent of survey respondents said they anticipate increasing their AI-related spending by 10% to 24%, while one-third expect a rise of 25% to 49%.
McLarty noted the data indicated medium-sized companies – those in the $1 billion to $10 billion range – were more aggressive with their spending on AI last year.
“Perhaps this shouldn’t be too surprising, given the agility advantage of smaller companies, but it’s always interesting to see how large, successful companies that have the most resources struggle to respond to disruptive technologies,” he said. “GenAI is truly a disruptive technology on the scale of the web or smartphones.”
