Gartner: Avoid AI project money pit

During the European Gartner Symposium in Barcelona, analysts cautioned that generative artificial intelligence (GenAI) has not lived up to expectations despite being past its peak in the Gartner hype cycle.

In her keynote address, Alicia Mullery, a vice-president for research at Gartner, highlighted two AI races: one among tech providers and the other to ensure safe and secure AI outcomes. She emphasized that the latter race is crucial for IT executives.

A key takeaway from the event was the potential for wasteful spending with GenAI. Mullery and co-presenter Daryl Plummer, chief research analyst at Gartner, stressed the importance of understanding and monitoring costs associated with AI projects.

According to Plummer, most organizations are not adequately prepared for AI adoption on emotional, technological, organizational, or managerial fronts. Gartner recommended two approaches to minimize failure: focusing on productivity improvements and driving transformational change through AI.

Gartner data revealed that proof of concept projects for AI can be costly, ranging from $300,000 to over $2 million. Plummer explained that expenses related to training AI models on GPU hardware and AI inference can quickly escalate.

Plummer criticized tech providers for failing to guide customers on the journey to achieve objectives with advanced AI systems. He highlighted the need for a shift from showcasing capabilities to demonstrating practical applications.

Many organizations are spending a significant portion of their budget on IT consulting to understand the benefits of new AI technology, as they are not yet ready to fully leverage advanced AI solutions from major providers.

IT leaders were advised to focus on desired outcomes when deploying AI, with some organizations running fewer pilots for efficiency improvements while others undertake multiple pilots for industry transformation. Gartner introduced the concept of TRiSM (trust, risk, and security management) to ensure compliance and oversight of AI systems.