Why Talking To A Human In India Now Feels Impossible

India loves a good tech success story. Over the years, the country has gone from a cash-based economy to a virtually cashless one, revolutionised by UPI. It has also raised eyebrows worldwide through the massive success of quick-commerce.

According to the Economic Survey 2024–25, AI is all set to transform services through “hyper-personalisation” and round-the-clock availability. On paper, this sounds like some good progress.

However, in reality, the surge of AI has made things almost impossible for an individual to try and fix a blocked bank account, report fraud or get a refund. AI has built a wall when it comes to customer service.

Why is AI taking over customer service? The reason is simple: Volume. With a population of 1.4 billion, India has enormous scale but thin margins.

Banks, telecom companies and e-commerce platforms deal with millions of queries every day, often from customers who bring in little revenue individually. Serving such a huge volume would require a considerable amount of human workforce. But human call centres are expensive. AI is not.

Thanks to the advent of AI, many large companies, including banks and telecom operators, have been able to report double-digit reductions in operating costs by shifting to chatbots, IVRs and automated workflows. This was a win-win situation for the companies. But for customers, it quietly rewrote the rules of engagement.

The central mechanism that can be attributed to this erosion of customer service efficacy is the industry's obsession with something known as the 'containment rate'.

This measures the percentage of customer interactions that begin and end within an AI layer without triggering a transfer to a human agent. It must be noted that the word 'contained' does not mean 'resolved'. The key here is minimal human interaction, not necessarily problem-solving.

If a frustrated user hangs up in navigating a recursive IVR loop, the system logs it as a success, thus incentivising the creation of what can be termed a doom loop - an interface designed to trap users in automated menus.

The failure and disservice of this model have been most evident in the banking sector. The assumption that private sector banks, with their superior tech offerings, provide better than public banks is a thing of the past.

In FY25, the RBI Ombudsman witnessed a 13.55% surge in complaints, driven largely by tech-enabled private sector banks, who have pioneered AI adoption. Historically, public banks were often criticised for lagging in tech. But in FY25, private banks accounted for 37.53% of all complaints compared to PSB's 34.8%.

This suggests that while PSBs may be slower in adopting tech, the aggressive automation strategies of private banks are generating more friction and dissatisfaction.

The challenges the digital wall poses in the banking sector are enormous. Victims of potential fraud are often forced to skim through IVR trees to find and report using the fraud option. The most insidious impact of the Digital Wall is the exclusion of vulnerable demographics.

Senior citizens and non-digital natives, in particular, face the risk of a service lockout due to the rapid AI exposure to the customer service sector, as they are often unable to navigate complex and multi-layered AI menus.

During September, a survey conducted by the Consumer Guidance Society of India in Mumbai revealed that 53% of senior citizens avoid digital banking due to fear and lack of skill. Instead, they rely on physical branches or phone support. But when phone support becomes an IVR maze, they are effectively locked out of the financial system.

This ineffective loop often lead towards unverified channels, such as fake customer care numbers found on Google, while also allowing fraudsters posing as human agents to siphon all the money.

This dark pattern is evident even more so in the Indian telecom sector, which is effectively an oligopoly of Jio, Airtel, and Vodafone Idea. With little-to-no competition, the incentive to improve service is low, and AI is used primarily as a cost shield.

While TRAI regulations mandate that IVR systems must facilitate a "Talk to Agent" option, most telecom companies bypass this by placing this option at the end of an IVR maze, meaning it may take an individual at least seven to eight minutes to end up talking to a human agent.

The sheer ineffectiveness of the IVRs has led to a 'cheat code' culture to bypass these algorithms - these are essentially means that hack the system such as selecting 'lost card', a high priority queue to report a billing error.

The consumer is now forced to 'hack' the service provider in order to be heard. At the same time, users have begun to weaponise Generative AI against these systems, such as the 'cracked egg' incident, where a user utilise AI to generate an image of a cracked egg to trick AI into processing a refund.

What's next for the technology? It appears we are moving towards a model where there might be human access or 'priority support' for a monthly fee while free users may continue to receive 100% AI services.

This has already been evident in some of the modern day app. Swiggy and Zomato both provide 'priority support' to paying customers, as human intervention becomes a luxury good. As companies move toward Agentic AI, which can take autonomous actions, the Digital Wall will only grow higher.