Four AI and CX Predictions That Didn’t Come True This Year – and What We Learned
Many experts predicted an immediate AI revolution in customer service. Instead, 2025 has shown that real transformation takes longer than expected. Here are four predictions that didn’t materialize, and why we need to rethink our approach.
1. “Autonomous AI agents will dominate contact centers this year.” Reality: Many projects have already been canceled.
In 2025, industry leaders expected “autonomous” AI agents (agentic AI) to handle customer interactions with minimal human supervision. Reality challenged that optimism. According to a 2025 analysis, more than 40% of agentic AI projects will be canceled by 2027 due to rising costs, uncertain ROI, and operational complications[1].
While interest remains genuine, mass adoption hasn’t happened. Many pilots were paused or shut down entirely. The most successful organizations are combining AI with human oversight rather than relying solely on automation.
2. “Sam Altman: AGI will be achieved in 2025.[2] New AI models will make a radical leap in capabilities this year.” Reality: In practice, progress is incremental.
A cornerstone of 2025’s optimistic predictions was the arrival of AGI (Artificial General Intelligence), with deeper comprehension, context awareness, autonomy, and the ability to perform a wide range of tasks at human-level skill. Those launches have been delayed due to technical challenges in meeting such high expectations.
The reason: today’s advances do not yet guarantee deep understanding, creativity, or contextual judgment, all essential components of AGI. For now, progress will remain incremental, focused on specific improvements rather than radical breakthroughs.
3. “Most companies will integrate AI into CX this year.” Reality: Only a minority have succeeded.
Despite the hype, widespread integration of AI into customer experience remains limited. According to a recent McKinsey study[3], only a portion of organizations have moved from pilot to full implementation: nearly two-thirds of respondents have not begun scaling AI.
The main blockers: fragmented data, lack of governance, internal resistance, and difficulty managing organizational change. Very few companies have all the elements of clean data, clear processes, and trained teams required for effective AI deployment.
4. “AI will manage most customer interactions by 2025.” Reality: actual adoption remains limited.
At the beginning of this cycle, many predictions suggested that by 2025, most customer interactions — chat, voice, support, and post-sales — would be handled predominantly by AI. The expectation was that automation would deliver efficiency, scalability, and 24/7 availability without depending so heavily on human resources.
A recent structural report warns that even in companies that adopted generative AI or automated agents, customer satisfaction and operational outcomes such as resolution time or first-contact success have not improved as quickly as promised [4].
In practice, many companies have had to maintain a hybrid model: AI supporting repetitive or low-value tasks, while human agents remain responsible for complex or sensitive interactions. That mix appears to be, in 2025, the most realistic formula, not a full replacement.
What did we learn? Reflections for building realistic AI strategies in CX
- Full automation is not the rule: AI should be seen as support, not a replacement. Successful projects combine technology with human oversight, robust processes, and a pragmatic approach.
- Investing in data, governance, and organizational culture is essential. Without a solid foundation, even the most advanced AI struggles to deliver results.
- Adopt gradually and measure carefully. Run pilots, test rigorously, and evaluate results. Jumping directly into “total AI” often leads to overspending or project cancellations.
- Maintain realistic expectations. AI is advancing, but not as quickly as some predictions suggest. Deep transformation takes time.
[1] Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027
[2] Sam Altman’s Bold Claim: OpenAI is on the Verge of AGI by 2025
[3] McKinsey: The state of AI in 2025: Agents, innovation, and transformation
[4] AI Agents For Customer Support: Trends, Predictions & Providers
