AI governance and clear roadmap lacking across enterprise adoption

Abstract image of magnifying glass checking data/ binary codes

Jackie Niam/Getty Images

Businesses are scurrying to adopt artificial intelligence (AI) tools as more become available, but most have not implemented the necessary metrics to measure the returns on their investment. 

Many also lack a comprehensive AI strategy and are acquiring products primarily for their bells and whistles, according to IBM’s AI Readiness Barometer Study released this week. Just 17% of companies assessed in the report have a well-defined AI strategy, with the majority, 38%, still in the midst of developing an AI strategy. Another 30% have an AI strategy that is focused on specific use cases, while 7% admitted to having an AI strategy they eventually discarded or were unable to implement effectively. 

The report found that about 43% had adopted AI due to the growing availability of AI-powered business applications. The IBM-commissioned study, conducted by research firm Ecosystm, surveyed 372 technology and business leads across five ASEAN markets: Singapore, Indonesia, Thailand, Malaysia, and the Philippines. 

Also: IT leaders worry the rush to adopt Gen AI may have tech infrastructure repercussions

Moreover, while 85% acknowledged the power of AI, just 22% measured its value and noted the report. This means that most lack clear ROI (returns on investment) metrics to determine whether their AI investments result in internal efficiencies or fuel external revenue. 

There also are gaps between how organizations rate their AI readiness and the reality of this status as assessed in the study, said Ecosystm CEO Ullrich Loeffler at a media briefing in Singapore. He explained that the research firm gathered data to evaluate the organizations’ readiness and maturity in rolling out their AI roadmap across four criteria. These included culture and leadership, data foundation, and governance framework. Scores were aggregated and used to place the organizations in one of five stages of AI readiness, spanning “traditional,” “emerging,” “consolidating,” “transformative,” and “AI-first.” 

Although 39% of respondents put their organizations in the transformative stage, Ecosystm’s assessment placed just 4% in this category. Another 16% of companies said they were AI-first, but Ecosystm found only 1% qualified for this stage of AI readiness.

Also: Time for businesses to move past generative AI hype and find real value

AI-first organizations are rated high across four key areas, including governance, where they have dedicated roles overseeing the function and have developed ethical AI solutions. These businesses also have a data-centric strategy that offers seamless data access and a workforce empowered by AI, including a centralized data team with strong AI and machine learning capabilities. 

In explaining the dearth of companies making headway in their AI adoption, Loeffler noted that while it is easy to achieve proof of concept, it can be challenging for businesses to obtain scale in their AI deployment. 

He further underscored the need for organizations to monitor and evaluate the impact of their adoption to ensure their AI applications are delivering the benefits as intended. 

According to the study, 63% of companies use AI to power intelligent document processing, 60% leverage the technology for support and helpdesk applications, and 57% use it for payment and invoicing automation. Another 56% tap AI for technology documentation, while 55% use it for content strategy and creation, and 55% leverage it for recruitment purposes.

Some 25% of organizations pointed to identifying use cases to pilot or run proof of concepts as their top AI priority. 22% view improving data quality, interoperability, and consistency as their AI priority, while 21% cite the need to upskill and reskill employees to be data-ready. 

Some 39% said their organization had limited AI expertise, with few specialists in certain areas, and 26% used Al within their existing applications or platforms and did not have standalone Al capability. 

Also: Transparency is sorely lacking amid growing AI interest

The study further highlighted the lack of governance framework as a concern, where just 18% of organizations have a dedicated AI and data governance role. 66% spread this responsibility across departments or teams, and approximately 3% do not have clear policies or defined responsibilities around AI governance.

In addition, only 12% have the processes to track AI model performance variations or model drift, which can impact results over time, according to the report. 

“The tangible benefit for organizations lies in scaling AI to speed up innovation and productivity,” said Catherine Lian, IBM’s general manager for ASEAN. “Unfortunately, many technology and business leaders overestimate their organization’s ability to implement AI successfully. AI readiness requires a strong leadership, robust data strategy, the right talent to execute it, and a well-thought-out governance framework to ensure the responsible and ethical use of AI.”

“Without these strong foundations, organizations risk implementations that focus solely on the technology’s capabilities but fail to weigh up the longer-term impacts on the business,” Lian said. 

Also: AI risks are everywhere – and now MIT is adding them all to one database

Hans Dekkers, IBM’s general manager for Asia-Pacific, also noted the need for AI alongside automation to help organizations keep pace with the speed of change.

ZDNET asked if there was an increased risk of incidents such as the CrowdStrike outage — should organizations increasingly rely on automation to keep up with patch management and other key work processes?

Dekkers said automation is crucial in freeing employees from time-consuming and repetitive tasks and driving the pace of transactional processes.

Also: Fight for global AI lead may boil down to governance

However, automation needs to be implemented correctly to avoid missteps, he said.  

Loeffler added that this should also be part of an organization’s governance framework, including ensuring that third-party AI applications meet the company’s AI safety policies.  

READ MORE HERE