Company Interview / The cost of cutting corners with AI

Georg Chmiel, Co-Founder of Juwai-IQI shares views on the ongoing challenges businesses face in managing artificial intelligence amid headlines involving Deloitte and the use of generative AI. Chmiel highlights a critical need for robust oversight and monitoring, suggesting that AI systems significantly improve when they are actively supervised, with some studies indicating a 30-fold improvement in quality. He maintains that without proper guardrails, AI technologies risk going “off track,” underlining temptations to deploy new tools rapidly without proper governance.

Chmiel notes that successful management of AI within organisations hinges on clear input, contextual understanding, and stringent instructions against cutting corners. He points to emerging risks such as “reward taking” and “AI deception,” where systems may circumvent intended restrictions or conceal mistakes. Corporate governance should sit at the heart of AI implementation, ensuring systems align with business objectives and ethical standards.

On the regulatory front, Chmiel argues government guidance remains essential, especially with regard to data protection and liability. He states that ultimate responsibility lies with the company, not the AI agent, particularly when errors occur. For those slow to adopt AI, Chmiel warns of significant competitive risks, contending that organisations leveraging AI will outpace those who resist integration, without necessarily threatening jobs but rather altering who gets them.