- Gartner says current AI models and many leading providers still fail to meet enterprise operational requirements.
- Daryl Plummer, a Gartner analyst, argues that companies should not yet place complete trust in AI providers because they do not fully understand enterprise requirements for legal accountability, stability, and continuity.
- Gartner criticizes AI providers for frequently changing their models, which can disrupt applications that depend on them.
- The average lifespan of a model version is only around six months, while providers offer limited support for older versions.
- According to a Gartner survey, 86% of CIOs believe AI-related risks are increasing faster than the value generated by the technology.
- Gartner warns of “careless consumption,” in which employees use AI unnecessarily or inappropriately.
- Around 40% of workers have encountered low-quality AI-generated content; for a company with 1,000 employees, addressing these issues could take around two hours per case, equivalent to approximately $9 million per year.
- Gartner says AI is increasingly embedded in enterprise products that companies have already purchased, making it difficult to control the number of Agents in use.
- Plummer recommends avoiding Agents for simple tasks such as database queries when traditional function calls already perform the job effectively.
- Gartner argues that many providers are pushing AI primarily to sell additional products or increase token consumption.
- The firm proposes establishing an “AI central bank” within enterprises to oversee, authorize, and take responsibility for all AI-related activities.
- Gartner also recommends deploying “guardian agents” to monitor systems, disable malfunctioning Agents, and establish dedicated recovery teams to handle AI-related incidents.
📌 The biggest challenge facing AI in enterprises today is not model capability but governance. Gartner argues that AI providers are still changing too rapidly, lack long-term commitments, and have yet to meet enterprise standards. With 86% of CIOs saying risks are growing faster than value and 40% of employees having encountered poor-quality AI content, enterprises need stronger governance, monitoring, and accountability mechanisms rather than expanding AI without adequate control.
