Where Is Conversational AI Headed in 2024?

If you’re feeling a little lost as you try to keep up with the latest news about artificial intelligence, don’t worry. You’re definitely not alone—especially coming off the whirlwind of advancements that we saw in 2023.

Last year, we saw conversational AI adoption come truly into the mainstream with generative AI and large language models (LLMs) such as ChatGPT embraced in a number of different settings. When it comes to how conversational AI will be used for businesses, 2024 will continue to bring exciting news, as well as some changes to how we put AI to use.

At Verint, our experts have been developing conversational AI technology for decades, and they continue to innovate in this space. Their ongoing research continues to provide better customer engagement and more streamlined operations for companies around the globe.

We asked a couple of our experts what they see happening in the conversational AI space this year. Here’s what they had to say.

Ian Beaver: Chief Scientist, Verint Da Vinci AI and Analytics

On Generative AI and Large Language Models…

The ongoing commoditization of GenAI and LLM-as-a-service price wars will continue between major model providers, bringing down costs for users but also likely creating more pricing structures for “enterprise” level service.  These price structures will require real time and enterprise use cases to pay more to support their product SLAs despite cost reductions for occasional use.

On where AI models are heading…

AI engineering will continue to increase in demand as a skill set.  Companies will also begin to struggle with model overload as model providers continue to launch updated flagship models as well as maintain a growing pool of “large enough” language models.

New models and version updates require re-testing existing prompts used in applications to verify no changes are needed. AI engineers will need to be familiar with the nuances of more and more model families, and prompt management in deployed products will increasingly add overhead to maintain. 

On AI regulation…

With the ratification of EU’s AI Act and penalties of 1.5-7 percent revenue for non-compliance, companies operating in the EU will begin to more closely evaluate what they are doing with AI and if it will bring them into regulated territory.

Here’s an overview of the impacts companies can expect.

Frank Schneider: Verint’s AI Evangelist

On the growing power of AI…

With great power comes great responsibility… and that doesn’t just go for Spiderman. With the advent of LLMs and generative AI, the legacy world of building intents and alternates has evolved to design thinking or user-centered approach to conversational AI.

The power of automated conversational experiences is now in the hands of enterprises with or without a conversational AI platform, and the key to execution will be platforms like we use at Verint—IVA Studio—that enable safe, compliant, and trustworthy AI automation that orchestrates customer experience across contact center technologies and robotic process automation (RPA).

Trust (the responsibility) goes hand in hand with this new power. The winners will be those technology companies that get what it means to be a proven contact center-grade trust partner.

On conversational AI bot building…

Conversational AI platforms must truly be “LEGO-like” kits that can not only build what is prescribed with established widgets and integrations—but can also allow the practitioner and brands to build outside the instruction manual. In other words, brands must be empowered to build applications and integrations that serve their interest, and not always be reliant upon the commercial interests of the platform provider.

In this world of extensibility, open means client-built applications and integrations that can be published in a hub or marketplace that enables cross-workplace and organizational sharing with vendor- and community-driven issue tracking and collaboration.

On transparency in data…

With LLMs, RPA and conversational AI solutions all working in conjunction to orchestrate experiences and automation across agent- and customer-facing tools and applications, having true transparency into data remains as important as ever.

From the data lineage of what was used to train a mode while helping to ensure compliancy, to a validated and reliable vision that turns into results, to consumable, yet secure client data usage, conversational AI platforms must now help enable true vision into AI data governance for brands to maximize and protect their investments in automation.