In 2025, almost all customer service will involve AI, but less than half of customers trust companies to use AI the right way. People, especially Gen Z, want to know when they are talking to a robot and expect their data to be used in ways that help them, like getting special offers or better service. The best companies mix AI with human help and are open about how they use AI. Trust, clear communication, and showing real benefits are the keys to keeping customers happy and loyal.
What are the key trends shaping customer trust and AI-powered experiences in 2025?
In 2025, AI drives 95% of customer interactions, but only 42% trust companies to use AI ethically. Key trends include declining trust – especially among Gen Z – rising demand for transparency, effective hyper-personalization, hybrid human-AI service, and ROI measured by business outcomes, not ticket closures.
Salesforce’s 2025 “State of the AI Connected Customer” report arrives at a moment when the phrase “AI-powered service” no longer turns heads – yet the numbers still surprise.
Based on a global survey of 16,585 consumers and business buyers, the study shows that while 95 % of customer interactions are expected to involve AI by year-end, only 42 % of customers trust companies to use AI ethically, down from 58 % just two years earlier.
Below are the five shifts that matter most for every CX and marketing team right now.
1. Trust is becoming a product feature, not a policy footnote
| Metric | 2023 | 2025 |
|---|---|---|
| Customers who trust ethical AI use | 58 % | 42 % |
| Customers who want to know when they are speaking with AI | n/a | 72 % |
The decline in trust is fastest among Gen Z (-19 pp year-on-year).
Take-away**: expose the agent’s identity early, offer a one-tap “human please” button, and publish plain-language model cards – customers now read them.
2. Hyper-personalisation at scale is working – but only when data value is obvious
- 73 % say brands now treat them “as individuals” (up from 39 % in 2023)
- 49 % believe their data is used *beneficially * – a nine-point gap that fuels churn
Best-in-class firms ask for consent *and * show the payoff in real time (e.g., “Your size is back in stock because you shared your fit preference last month.”).
3. AI agents are shifting work, not eliminating it
Salesforce’s own support organisation reports:
– 30-50 % of routine tickets are fully deflected by the Agentforce suite
– Escalations that reach humans carry 2.3× higher sentiment scores, thanks to contextual hand-offs and pre-drafted responses
Outside Salesforce, an Accenture study finds companies treating service “as a value center” see 3.5× revenue growth versus peers who keep it a cost center.
4. Transparency expectations are segmenting by generation
| Segment | Comfortable with AI agent shopping on their behalf |
|---|---|
| Gen Z | 32 % |
| Millennials | 21 % |
| Gen X | 11 % |
| Boomers | 4 % |
The spread shows why hybrid journeys (browse with AI, checkout with human advice) now outperform purely automated flows.
5. ROI is real – when measured on business outcomes, not ticket closure
Independent 2025 data illuminate where dollars actually move:
| Outcome | Median uplift |
|---|---|
| Average order value | +47 % |
| Cart abandonment reduction | -20 % to -30 % |
| Lead generation (real estate example) | +138 % |
Sources: Sprinklr customer benchmark, Tidio AI chatbot study
Quick checklist for CX leaders in 2025
- Label every AI touchpoint within 5 seconds of first contact
- Co-create privacy policies with customers – in-product voting on data uses lifts opt-in 23 %
- Design escalation paths before agent deployment – the costliest error is hiding the escape hatch
- Measure revenue per conversation, not cost per ticket, to align AI metrics with growth targets
Why is customer trust in AI ethics declining in 2025?
Salesforce’s 2025 State of the AI Connected Customer shows that only 42 % of customers now trust businesses to use AI ethically, a sharp fall from 58 % in 2023[7][3]. Independent 2025 data echo this concern: over 75 % of consumers worry about misinformation and data misuse when AI is involved[1]. The takeaway is clear – transparency about how data is collected and decisions are made is no longer optional; it is the fastest lever to close the trust gap.
How much of customer service is already AI-driven?
By the end of 2025, 95 % of customer interactions are expected to be AI-powered[4]. Salesforce teams alone report that AI agents now handle 30 %–50 % of coding and customer-service workloads, with accuracy in customer-facing replies reaching 93 %[2][6]. For readers managing service operations, this means three out of every four tickets may soon be resolved without human agents, making the quality of AI guardrails a direct driver of CSAT scores.
Do customers still want human agents at all?
Yes. Complex or emotionally sensitive issues remain firmly human territory. Surveys show that consumers and business buyers prefer human expertise for high-stakes decisions, and easy escalation to a person is a top trust signal[7][4]. The most successful companies are therefore running human-AI hybrid teams, where AI triages routine questions in seconds and humans step in for empathy, negotiation, or edge cases.
What measurable ROI are companies seeing from AI agents?
- Revenue uplift: Firms using AI as a growth engine report up to 3.5× more revenue growth with only modest increases in service spend[1].
- Cart recovery: Retail chatbots reduce cart abandonment by 20 – 30 % and lift average order value by as much as 47 %[4][5].
- Lead generation: A real-estate data platform using Tidio’s AI chatbots recorded a 138 % jump in qualified leads[5].
These numbers prove that AI is no longer a cost-cutting play alone; it is a profit center when deployed with clear KPIs.
What happens next for human agent roles?
In 2025, the role of the human agent is shifting from ticket-closer to CX orchestrator. AI supplies real-time sentiment analysis and knowledge suggestions, while agents focus on relationship-building and complex problem-solving. Training programs now emphasize prompt engineering, AI oversight, and escalation design, ensuring that the agent’s “soft skills” become the premium differentiator in an AI-first service world.
