The latest IBM Japan survey found that 90% of companies do not have a sufficient understanding of their dependencies on AI vendors and models. It also highlighted the risk of operations being disrupted by changes in external factors and the difficulty of switching vendors.
While incorporating AI into the core of their operations, 89% of companies are unable to understand the internal structure of their systems and their dependencies. 75% are concerned that an outage of AI services lasting about one week would cause dysfunction across their operations, underscoring the pronounced lack of awareness of current risks.
In response to changes in the external environment, 69% of domestic companies find it difficult to switch vendors. Although many companies use multiple vendors, it has been pointed out that independent decision-making by department and regional requirements have not led to strategic risk diversification. The fact that companies experienced an average of 6 system outages over the past 2 years was also highlighted.
Organizations that have avoided black-box systems and established a foundation for controlling AI in-house are achieving results. Advanced companies with strong control capabilities account for just 7%, but they have reportedly been able to mitigate the impact on operating profit from AI-related disruption by 55% more than other companies.
Among Japanese executives, 68% are willing to accept a 20% increase in costs to secure freedom of choice and flexibility in selecting vendors. Ultimately, the view presented is that establishing “AI sovereignty,” by breaking excessive dependence on specific vendors and governing AI autonomously, will be key to protecting businesses.