Gartner urges data trust over AI hype for business success

At the Gartner Data & Analytics Summit in Sydney, analysts warned businesses against rushing into AI without ensuring data trust. Carlie Idoine, Gartner VP analyst, cautioned that "'Garbage in, garbage out' isn’t the worst thing in the world, but putting trust in that garbage is."

Key Takeaways:

1. Data Trust is Critical for AI Success

• Gareth Herschel, another Gartner VP analyst, cited a 2024 survey revealing that data availability and quality are the top obstacles to AI implementation.
• "If you cannot trust the data, you cannot trust the AI that uses it," Herschel emphasized.
• Solution: Implement trust models that rate data reliability based on lineage and curation.

2. AI Agents vs. Generative AI

• Erick Brethenoux, Gartner’s chief of AI research, clarified the distinction between AI agents (used for decades in tasks like predictive maintenance) and generative AI (like LLMs).
• Agentic AI is largely a marketing term, but combining AI agents with LLMs can yield innovative results.
• Challenge: Generative AI’s non-deterministic nature requires guardrails and simulations to ensure reliability.

3. Autonomous Business Processes

• AI agents can execute multistep processes in parallel, saving time and costs. For example, loan applications can be processed faster by terminating failed checks early.
• Autonomy levels: Low-risk decisions (e.g., travel arrangements) can be fully automated, while high-stakes scenarios may require human oversight.

4. ROI and Business Impact

• Luke Ellery, Gartner VP analyst, noted that tools like Microsoft Copilot save only 14 minutes daily per employee, with a net cost of ~$1,150/year.
• Upside: Employee satisfaction (NPS of 59 vs. 21 average) and perceived productivity gains.
• Recommendation: Categorize AI use cases by value type and build a diversified project portfolio.

5. Future of Analytics: Perceptive Systems

• Georgia O’Callaghan, Gartner director-analyst, predicted that by 2027, autonomous analytics platforms will manage 20% of business processes.
• Trends: