AI and its main promoters are not enterprise-ready, says Gartner

Gartner analyst Daryl Plummer and Kristin Moyer warned that AI is not yet enterprise-ready, stating that trust in AI vendors cannot be relied upon at this time. Plummer highlighted the lack of understanding of enterprise terms and conditions from AI vendors, and the rapidly changing nature of their models. He criticized the pace of innovation as "out-of-control" and noted that AI companies have not developed a willingness to support legacy technology, as their model lifespans are only six months long.

Moyer pointed out that 86% of CIOs view risks from AI growing faster than its value, due to the demand for AI created by early successes that IT departments cannot safely satisfy. She also mentioned the "careless consumption" of AI by workers, which takes about two hours to decipher and costs organizations $9 million annually. Plummer suggested that enterprises should have experience in controlling this careless consumption, as AI has infiltrated products that users already own.

Plummer proposed the idea of an "AI central bank" to oversee AI use and ensure its ongoing health by considering the technology's systemic impact. He also recommended the creation of "guardian agents" that could monitor AI agents and "kill" them if they deviate from intended bounds. Additionally, Plummer stressed the importance of establishing disaster recovery teams responsible for unravelling AI-related problems, as CIOs will be held accountable for AI failures.