Is Conversational AI Dead in the Age of GPT and Other LLMs?


Share on LinkedIn

The rapid advancements in Large Language Models (LLMs) like GPT have sparked a debate about the future of conversational AI. While it’s tempting to view these developments as a sign of conversational AI becoming obsolete, the reality is more nuanced. As we witness the rapid evolution of Large Language Models (LLMs) like GPT, a pressing question emerges in the tech community: Are these advancements the end of conversational AI? On the surface, the robust capabilities of LLMs seem to outshine traditional conversational AI systems. But is that true in a Customer Service setting?

The LLM Revolution

The advent of GPT and similar models has undeniably revolutionized the field of artificial intelligence. These models excel in processing and generating human-like text, reshaping expectations in AI-powered communication. Now everyone wants GPT-like experiences everywhere. This technological leap forward has sparked discussions about the future role of more specialized conversational AI systems.

LLM vs Conversational AI

With the advancements of LLMs, does conversational AI have a future value in the AI ecosystem?
LLMs like GPT are enriching the customer service landscape but there are pitfalls. Contact Centers risk unprecedented costs, lack of control and regulatory issues that need to be addressed before venturing on this journey:

Broad Knowledge vs. Specialized Expertise:

LLMs have wide-ranging understanding, making them suitable for general use cases where a variety of topics might come up, but can lack specialized knowledge crucial to successful customer interaction. For an organization, this means that while LLMs can provide a robust foundation for building AI-driven applications due to their extensive knowledge base, they need orchestration to be controlled for customer service.

LLMs Cost Implications:

The use of LLMs has significant implications for cost and training within an enterprise.
Deploying LLMs like GPT will be expensive due to the computational resources required to run them, especially for models that are kept active and are constantly learning from new data. Additionally, there may be costs associated with API calls if the enterprise is using a cloud based LLM service. When handled via a NLU, LLMs will require fewer API calls to operate than a general LLM and can be fine-tuned to use resources more effectively for their specific tasks.

Training Considerations:

Training an LLM like GPT involves vast datasets and considerable machine learning expertise. This can be costly and time-consuming, and it requires ongoing retraining to keep the model up to date with the latest information and language use trends. LLMs might provide a quick setup and a wide range of functionality but at a potentially higher operational cost. Specialized conversational AI will lead to cost savings through efficiency and the ability to provide precise assistance in specialized fields, which translates to higher customer satisfaction and retention. When it comes to cost and training, enterprises need to carefully evaluate their specific needs. If the need is for a versatile system capable of handling a wide range of topics, then an LLM might be more appropriate despite higher running costs. If the enterprise operates within a industry where accuracy in customer interactions and compliance are paramount, then orchestrating the LLM will be more cost-effective and beneficial in the long term.

Controlled Interaction:

Conversational AI allows for greater control over dialogue flows, ensuring conversations adhere to specific scripts or guidelines. This control is vital in environments where maintaining a focused and objective-oriented conversation is key, such as in customer service scenarios.

Ethical and Privacy Safeguards:

Tailored conversational AI systems can be designed with built-in ethical standards and privacy protections, addressing concerns that broad-spectrum LLMs may not inherently consider.

Resource Optimization:

Given their intensive computational demands, LLMs may not be feasible or cost-effective for all scenarios. In contrast, conversational AI can often provide more resource-efficient solutions.

Regulatory Compliance:

In heavily regulated industries, conversational AI can be custom-built to comply with specific legal and regulatory requirements, a challenge that general-purpose LLMs like GPT might not readily meet.

Seamless Integration:

Conversational AI can be more seamlessly integrated into existing communication channels and customer service platforms, providing a consistent user experience across various touchpoints.

Navigating the Future

As we look ahead, it’s clear that LLMs like GPT is amazing to enhance productivity in individual use cases. But for an enterprise there are challenges in control and costs, especially in customer facing scenarios. The rise of LLMs like GPT marks a transforming milestone in AI development, but it doesn’t spell the end for conversational AI. Instead, these technologies form a collaborative relationship, each playing a unique role in the broader AI landscape. By understanding and utilizing their respective strengths, customer service strategists will ensure that Contact Center continues to evolve in a way that maximizes benefits while addressing the challenges and limitations inherent in each AI approach.

Marie Angselius
Marie Angselius-Schönbeck is Chief Impact Officer and Chief Corporate Communications at Artificial Solutions, a company in Conversational Customer Experience. In 2019, she founded Women in AI by Amelia, an IPsoft-company, a global initiative to help close the gender gap in STEM. She has worked in th Conversational AI-industry for 5 years.


Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here