The Future of AI: A Vision for Customer Service in 2024 and Beyond

With the rapid advancement of both open and closed-source AI models, unparalleled investments and immediate adoption by major tech companies, AI in all its forms is unavoidable from both an operational and strategic perspective. But the dizzying speed of development has left business leaders with glimpses of amazing benefits, shrouded by a thick fog of questions and uncertainty.  

Instead of peddling predictions, let’s look at five key areas to keep in mind when developing an AI strategy in 2024 for customer service and the overall enterprise. 

Proliferation of AI Agents Across the Enterprise 

AI Agents are evolving to become a ubiquitous part of the enterprise landscape. These agents or assistants, powered by advanced Conversational and Generative AI, will expand from their initial roles in support and helpdesk to revenue-generating areas of the enterprise such as sales and marketing.  

Over time, AI Agents will move from a primarily reactive role to a proactive one, reminiscent of current AI Copilot (agent assist) technology. This could be AI reaching out as soon as your flight is delayed or canceled with personalized suggestions, offering a business class upgrade. It could be a personal shopper sending you a WhatsApp message when it detects something you wanted to order months ago, is suddenly in stock at your favorite shop, and is answering any follow-up questions you might have. They will be able to do this in any language and in any communication channel around the clock, jacked into all your knowledge.  

The contact center will turn out to be the proving ground for AI Agents and, ultimately, the launchpad for their widespread adoption within companies. 

Emergence of Equal or Better than GPT-4 Open Source LLMs 

While proprietary LLMs from the likes of OpenAI, Anthropic, Google, and others currently maintain the lead, open source is catching up and will inevitably reach parity in terms of performance. Mistral’s LLMs, Meta’s Llama 2, and Microsoft’s Phi2 have already gone open-source and Apple has recently developed a smaller-scale LLM able to run on an iPhone. Nevertheless, it remains to be seen what type of market share equally powerful open-source models take, especially considering the still high cost to not just train, but also run the models on powerful GPUs. 

While today’s focus is on the never-before-seen abilities of GPT-4, we will ultimately see a proliferation of diverse and specialized LLMs. This may include open-source models, AI as a Service service companies, proprietary LLMs, and ones fine-tuned for specific tasks. 

Specialized LLMs will lead to better results and lower costs, not to mention a reduced energy footprint compared to today’s energy-intensive training and inference requirements. With the incredible language mastery we have seen from LLMs, let’s not forget that they need not be able to compose Shakespearean sonnets to perform call summarization. 

Easy access to LLM technology combined with affordability may guarantee its adoption and set consumer expectations, but it ensures neither quality nor success. Mobile phone technology is ubiquitous and yet Apple still has 25% of the global market share. While there are several reasons, the largest ones are user experience and quality. There is no reason to believe it will be different for LLMs, whether proprietary or open source.  

That means the true differentiators in the market when choosing AI technology will not be the base technology itself but rather the “secret sauce” driving their results. This will include abilities such as the orchestration of multiple LLM based on the use case or even within use cases, enterprise readiness, security and privacy, speed, accuracy in their domain, and the overall user experience they deliver. That is the reason most readers are viewing this article on Windows or Mac and not Linux which is free. 

Release of GPT-5 

In a recent interview, Sam Altman, CEO of OpenAI, told Bill Gates: “At least for the next 5 or 10 years we will be on a steep improvement curve, this is the stupidest these models will ever be” and will largely focus on improving its reasoning capabilities and reducing hallucinations.  

With its enhanced capabilities, businesses can expect even more sophisticated AI interactions. GPT-5's advanced understanding and response generation means AI Agents will be able to handle more complex, nuanced conversations, bridging the gap between human and AI interactions. 

The potential of GPT-5 in the enterprise sector, especially in customer service and sales, is immense. With Altman's vision of a steep improvement curve, we are looking at AI that continually evolves, becoming more intuitive and reliable. That means businesses deploying AI Agents not just as first-line responders in self-service scenarios, but as sophisticated tools capable of handling intricate customer needs. For instance, in a customer service scenario, GPT-5's enhanced reasoning abilities could enable these agents to understand and solve complex customer issues in a more human-like manner, significantly improving the overall customer experience.  

The Next Generation: Interactive AI 

The first generation of AI focused on data classification, being able to understand things such as text or images. That meant for example, that AI could process a sentence, understand the parts of speech like nouns and verbs and figure out what it meant. The second generation, which we are currently experiencing goes a step further, training AI models to not just understand, but generate multimedia content, hence the name “Generative AI”. 

The next major generation of AI, which may be kicked off by GPT5 or Google’s next iteration of Gemini may be interactive AI which is ultimately fulfilling the long-standing promise of Star Trek-like conversational interfaces. This means a real assistant right out of Sci-Fi that can understand you, generate answers but critically, break down more complex requests that require research, reasoning and intuition into logical steps, action them and deliver the results required.  

Moreover, as AI Agents become more adept at processing and understanding large volumes of data, they can provide insights that were previously unattainable. In sectors like healthcare, GPT-5 could empower AI Agents to sift through vast medical data, assisting healthcare professionals in diagnosing and offering tailored healthcare advice. This capability extends the role of AI from mere assistants to critical decision-support tools, reflecting a significant leap in how enterprises can leverage AI for growth, innovation, and enhanced customer satisfaction. 

Generative AI Everywhere – Voices, Images, Videos, 3D Worlds, Bots of All Kinds 

As mentioned, Generative AI has moved from its initial focus on text to sound, images and video. This opens up entirely new use cases and opportunities for enterprises, both in and outside of the contact center making interactions not just more human, but also more visually rich. AI Agents would be capable of generating not just text-based responses, but also visual aids or even video responses, making interactions more engaging and informative. This leap from text to on-demand multimedia responses represents a significant advancement in how businesses can connect with and support their customers. 

Consider the implications in complex support scenarios, such as technical troubleshooting or product demonstrations. AI Agents, equipped with Generative AI capabilities, can analyze a customer's situation in real-time and provide a tailored, visual step-by-step guide. This is not only more helpful but adds a layer of personalization that was previously unattainable with traditional text-based AI responses. 

Imagine the all-too-common scenario of a frustrated consumer puzzling over how to finish putting together furniture. They could simply turn on their camera, show the AI assistant what they have done so far and the remaining pieces, and the AI assistant could generate a video on the fly with the remaining steps.  

In the realm of training and education, AI-generated content could provide immersive learning experiences. Employees in various industries could benefit from AI-generated simulations, interactive 3D models, and personalized training modules, enhancing their learning curve and efficiency. 

Conclusion 

In summary, the future of AI in customer service and the overall enterprise is not just promising; it is already here. With the current pace of exploration and innovation, the only limits are our imagination and creativity, which is also evidenced by the speed of adoption of Generative AI technology across a wide range of applications. The key will be to harness these advancements in a way that enhances human capabilities, fosters better customer relationships, and ultimately drives significant business value.