Editorial
The Gist
- Most AI fails — but CX is the exception. MIT finds 95% of generative AI pilots flop, yet CX programs repeatedly show measurable ROI because they already run on structured processes and omni-channel data.
- Execution gaps — not technology — drive the GenAI Divide. Hallucinations, shadow AI, and misapplied automation cripple implementations long before they reach production.
- A CX playbook shows where AI actually works. Validated knowledge, governed tools, and targeted automation turn AI from hype into productivity, cost savings, and improved human-centered service.
Like all emerging technologies, generative AI is following the trajectory of the Gartner Hype Cycle. Over three years on from OpenAI’s launch of ChatGPT, generative AI has climbed rapidly past the "peak of inflated expectations," followed quickly by a plunge into the "trough of disillusionment."
With the global AI market expected to reach $4.8 trillion by 2033, leaders urgently need to understand what separates successful pilots from costly dead ends, while generative AI looks to grow from the "slope of enlightenment" and into the "plateau of productivity."
One of the clearest roadmaps lies in customer experience (CX), where AI is already delivering measurable ROI.
Table of Contents
Why Are AI Projects Failing?
A recent report by MIT found that 95% of generative AI pilots at companies are failing, creating a stark divide between the 5% of high-value projects that deliver return on investment (ROI) and the rest, which the researchers are terming the “GenAI Divide.” The same MIT study found that while 60% of organizations explored enterprise-grade AI tools, only 20% ever reached pilot stage, and just 5% went into production. The report aligns with Gartner’s earlier prediction that 30% of generative AI initiatives will be abandoned after proof-of-concept by the end of this year.
According to IMB, only 25% of AI projects have delivered their expected ROI. Yet, CX stands out as a sector where results are tangible and repeatable. Contact centers are uniquely positioned to generate demonstrable ROI because they already operate with optimized people, processes and omni-data. This makes CX one of the first areas in which organizations can apply AI in a meaningful and measurable way.
Related Article: Is This the Year of the Artificial Intelligence Call Center?
Where AI Delivers Real ROI
A recent study found that 77% of organizations saw either “significant” or “some” cost savings from AI implementations in CX. These benefits go beyond finance, including employee productivity, well-being and stronger customer loyalty.
So why are AI projects failing, and what are three things organizations can learn from the application of generative AI in CX to demonstrate ROI?
1. Tackling Hallucinations
Generative AI hallucinations produce incorrect or fabricated outputs and can have catastrophic consequences in critical situations; they are a fast way to erode customer trust. Deloitte reports that 77% of businesses were concerned about hallucinations, particularly when it came to cybersecurity. Current hallucination rates range between 17%-45%, which could create serious risks for highly regulated industries such as legal, finance or healthcare.
To combat this, organizations should use Retrieval Augmented Generation (RAG), which validates generative outputs against approved data sources. In contact centers, providing links to original sources allows employees to check context instantly, cutting the risk of errors while maintaining customer trust.
2. Shutting Down Shadow AI
The MIT report also highlights the spread of “Shadow AI,” where employees use unsanctioned AI tools without IT oversight. Workers gravitate toward these tools because they integrate into workflows and are easily customizable to their specific needs, even if they bypass governance, which makes it harder to measure ROI. Between 2023 and 2024, the adoption of generative AI among employees jumped from 74% to 96%, with the post popular applications named as ChatGPT, Grammarly and Microsoft Copilot. Use of Shadow AI demonstrates strong demand, but also the governance challenge: businesses need visibility to measure ROI and manage risk.
In CX, shadow AI is less of a concern when enterprise-grade tools are embedded directly into the systems employees already use. For example, AI can act as a "second pair of ears" during interactions, automatically surfacing appropriate knowledge articles for contact center workers from pre-approved sources to enable smoother, more informed dialogue.
Here, the strengths of human and machine complement each other: instead of forcing people to operate like machines, AI can take over the “machine work” such as repetitive tasks, allowing employees to excel at decision-making and trust-building.
Related Article: The Rise of AI as a Real-Time Coach in the Modern Contact Center
3. Applying AI Where It Matters
The MIT study found that the strongest ROI from generative AI comes from back-office automation, an area closely tied to CX. AI can take on much of the routine admin work that often consumes over 50% of a contact center worker’s time, from capturing customer intent in the queue to transcribing conversations and surfacing relevant information in real time. For straightforward issues, intelligent self-service frees up human agents to focus on complex or sensitive interactions, boosting productivity while enabling more meaningful, empathetic support. By removing the burden of note-taking or data entry on contact center workers, AI also allows contact center workers to focus fully on customer conversations.
Critically, CX leaders know when not to automate. When an interaction is highly complex, emotionally charged or time-critical, it is better to be given to a human who can solve problems and can offer real empathy and understanding, leaving AI to deal with straightforward, low-sensitivity queries. This balance maximizes both ROI and customer satisfaction, as customers don’t feel like they are being fobbed off to an AI chatbot in moments of need.
Webinar
Dec
10
Beyond Composability: How Modern Marketers Build Connected Experiences
Ready to launch campaigns faster, personalize smarter and prove your marketing ROI? Discover the power of a modern DXP.
Register
Webinar
Dec
10
Unlock Connected Service: How to Forecast, Staff & Support Every Channel
Stop juggling tools. 73% of CX leaders say silos damage CX. Build a seamless service operation instead.
Register
Webinar
Dec
11
Empowering Non-Profits: Smarter Crisis Communication and Community Engagement
The cost of miscommunication is measured in more than words. Learn to deliver outreach that's fast, clear and trusted.
Register
Webinar
Dec
16
Roundtable: Turning Real-Time CX Signals into Business Results
Four big brands. One live, unscripted discussion on how modern CX teams move from dashboards to real impact.
Register
Webinar
On demand
From Manual to Magical: How AI Transforms CX Teams
Learn how to replace manual support processes with automation that actually delivers.
Watch Now
Webinar
On demand
How to Build a Solid Knowledge Foundation for AI Success
See how leading brands keep their AI honest, compliant and actually helpful.
Watch Now
AI Implementation Failures and CX-Proven Solutions
A breakdown of the biggest reasons AI projects stall — and how customer experience leaders overcome them.
| Problem | Why It Derails AI | Proven CX Solution |
|---|---|---|
| High hallucination rates | Incorrect or fabricated outputs erode trust, create compliance risk, and prevent production deployment. | Use Retrieval-Augmented Generation (RAG) to validate answers against approved knowledge and surface source links for agents. |
| Shadow AI adoption | Employees use unsanctioned tools, creating governance blind spots and making ROI unmeasurable. | Embed enterprise-grade AI directly inside existing workflows (e.g., contact center platforms) so staff adopt approved tools naturally. |
| Automating the wrong tasks | Organizations aim AI at high-sensitivity interactions or poorly suited use cases, leading to customer frustration and low ROI. | Automate back-office and repetitive tasks; route complex, emotional, or time-critical cases to humans for empathy and problem-solving. |
| Lack of data foundations | AI models underperform when fed inconsistent, siloed or unstructured data. | Leverage CX environments where optimized processes and omni-channel data already exist, enabling cleaner, higher-quality AI outputs. |
| No enterprise governance or visibility | Organizations can't measure ROI, manage security risk or maintain compliance across disparate tools. | Partner with vendor-agnostic AI experts who provide governance frameworks, cross-platform oversight and horizon scanning. |
| Poor change management | Teams don’t know when to rely on humans vs. automation, leading to customer dissatisfaction or operational breakdowns. | Use CX’s “human-first” deployment model: AI handles machine work; humans handle empathy, judgment and complex problem-solving. |
ROI for Generative AI
The arrival of AI creates unheralded opportunities for end-user businesses to build competitive advantage through operational efficiency and better service, at the same time dooming to failure those organizations that fail to invest. To fully embrace AI, organizations need a vendor-agnostic partner that understands the aims of implementation and has a holistic view of the latest technology and vendors, including horizon scanning for emerging players. An AI partner will be able to offer fully developed solutions that can make an instant impact, based on their ROI objectives. The right AI partner can deliver AI technology that fits an organization’s requirements while keeping up to date with the latest innovations.
Closing the GenAI Divide
Ultimately, the “GenAI divide” is not about technology, but about execution, and CX offers a clear roadmap for organizations looking to get the most ROI from their AI investments. This principle of AI "helping humans to help humans’ should guide deployments across all industries, not just CX. By focusing on collaborating with trusted AI vendors, businesses can balance the risks of AI with enterprise-grade tools that automate tasks to maximize their investments.
Learn how you can join our contributor community.
