At Salesforce Connections 2026, MIT's Human First AI group presented insights into the widening trust gap between marketers who largely trust agentic AI and customers who remain skeptical, urging marketers to adopt an ECSIT strategy to rebuild trust.
- Marketers trust agentic AI 81%, customers only 36%
- Trust increases when AI logic mirrors customer thinking
- Human-AI interaction timing critical to calibrated trust
What happened
During Salesforce Connections 2026 in Chicago, research findings were shared that highlight a significant discrepancy in trust towards agentic AI between marketers and their customers. While the majority of marketing professionals express confidence in these AI agents handling customer data, this confidence is not equally reciprocated by the customers themselves, with trust levels noticeably lower among them.
Dr. Renee Gosline of MIT’s Sloan School of Management, leading the Human First AI group, has been exploring behavioral science approaches to understand decision-making processes in humans interacting with AI. Her team conducted experiments demonstrating that customers are more likely to trust AI agents when they can see that the AI’s reasoning closely matches their own logical processes.
Why it matters
The trust gap poses a challenge for marketers relying on agentic AI to deliver personalized and data-driven customer experiences. Customers’ reluctance undermines the effectiveness of these technologies and suggests potential reputational risks if AI agents are perceived as untrustworthy or incomprehensible.
Gosline’s research suggests a practical solution by showing AI logic flows that reflect a customer’s own thinking patterns can create a sense of empathy and similarity, fostering greater trust. This aligns with marketers’ ability to use AI as digital twins tailored to individual customers, improving acceptance and adoption of AI-driven services.
What to watch next
Further exploration will focus on optimizing how AI agents interact with humans, particularly the timing of AI suggestions relative to human input. Experiments show that when AI recommendations come first, human review and justification of the AI’s reasoning help calibrate trust to a balanced level, avoiding both over-trust and skepticism.
Enterprises should monitor developments in ECSIT (Empathic, Calibrated, Synchronous, Interactive Trust) strategies that integrate transparent AI decision-making processes with behavioral insights. This approach could become a core best practice for marketers aiming to close the trust gap and maximize AI’s potential in customer engagement.