What trends for generative AI in 2024?
The year 2024 will make generative AI sustainable.
GPS routes and favorite series recommendation algorithms are all examples of AI that has become so commonplace that it is almost imperceptible. Then comes the year 2023, and the democratization of generative AI with ChatGPT, adopted in just two months by 100 million users, mainly individuals curious to test the capabilities of this robot.
What a shock ! It must be said that this new tool is not only a question of technology, but of economic, social and ethical issues. Decision-makers must lay down the fundamental principles of an AI that respects human-machine relationships. However, few organizations have established such principles. A Gartner study from June 2023[1] highlights that only 9% of organizations had a vision around AI, and more than a third of respondents had no plans to create one. Thus, the deployment of generative artificial intelligence cannot be an end in itself. It is necessary to associate a vision and objectives, in order to resolve the operational frictions of your organization.
Trend 1: place for specialized models!
Therefore, it becomes essential that generative AI models specialize. ChatGPT impresses with its ability to summarize ancient Greek history, but is this skill really useful to a financial institution?
First example: generative AI at the service of banking
For example, banks strive to maximize the equipment rate of their customers, that is to say the number of services subscribed to by each customer. Artificial intelligence plays a crucial role in this process by evaluating the profile of a potential customer, and comparing it to existing similar profiles.
On the one hand, generative AI helps classify the customer profile according to segmentation triggered by specific signals. This approach makes manual segmentation unnecessary, since generative AI takes care of it. It is able to detect, analyze and classify customer behavior quickly and efficiently.
On the other, a predictive AI algorithm, often referred to as “look alike”, evaluates the client's eligibility based on various criteria, such as assets and risk rate.
The implementation of AI in this context allows a more strategic reallocation of resources, embodied by the principle “do more with less ».
The result is increased efficiency, which promotes the creation of added value. The exploitation of these models therefore opens the way to more precise and targeted applications, intended to maximize their impact in various industrial sectors.
Second example: generative AI at the service of industrial brands
As a second example, let's take the optimization of commercial promotions in the consumer goods sector.
According to a McKinsey study from October 2019[2], companies invest almost 20% of their turnover in these promotions, but 41% of them result in lossess, amounting to billions annually.
To address this trend, consumer goods experts must evaluate the profitability of their promotions effectively. This involves setting up a supported process for key account managers (KAMs) to create new promotions.
Generative AI supports these managers by, for example, performing preliminary data analysis and dynamically adjusting to market developments. After the promotion is deployed, the manager can engage generative AI again to automatically generate an email summarizing the promotional data for field sales teams.
The field representative leverages the assistance of generative AI to develop the ideas to include in their sales presentation. Alternatively, he can also take help from the practical guide generated by the AI, bringing together the best trading strategies encountered by the company.
These examples highlight the considerable potential of generative artificial intelligence in the commercial domain. They demonstrate that the impact of this technology extends well beyond simply increasing productivity, by significantly influencing value creation.
It is also important to emphasize that the effectiveness of generative artificial intelligence systems relies on their ability to optimally exploit data specific to each company.
Trend 2: effective use of data, key to successful implementation
The implementation of such scenarios essentially relies on efficient use of data.
It is crucial that systems do not operate in silos, but are integrated in a unified approach. With access to relevant data, their processing must be done using appropriate algorithms, with emphasis on the specialization strategy.
Companies can choose between specialized solution models or their own models (BYOLLM : Bring Your Own Large Language Model). To integrate data from operational systems in real time without the need for prior learning, the method RAG* (Retrieval-Augmented Generation) is recommended.
In order to fully optimize the use of generative AI as a whole, it is essential to explore different approaches to presenting results. Given that natural language is the main mode of interaction with these tools, interfaces will tend to reinvent themselves.
Trend 3: towards new user interfaces
Naturally, interaction with generative AI can take the form of a simple, linear conversational exchange with a virtual agent. Users are no longer forced to navigate through traditional clicks, but can engage in a natural conversation with their interface. This revolutionary approach transforms interaction with technology, making it more intuitive, accessible and aligned with natural human behaviors.
These more sophisticated, conversational interfaces will not only enable the execution of tasks or presentation of information on demand, but will also be able to learn and adapt over time. Conversational agents, thanks to machine learning, will be able to understand the preferences, habits and specific needs of each user. This will enable personalized support over the long term, rather than one-off, isolated interactions.
In the near future, these agents will become true personal assistants, capable of anticipating user needs, suggesting proactive actions and offering tailor-made solutions. This development marks a turning point in human-machine interaction, where the interface becomes an interactive and evolving partner, continuously adapting to best meet the user's requirements and preferences.
Trend 4: a nuanced impact on employment
Finally, among the debates surrounding generative AI is its impact on employment. Recent research sheds interesting light on this question.
In the field of customer service, for example, organizations are not planning job reductions or layoffs. On the contrary, they plan to hire more agents in 2024... in varying proportions.
Contact centers using AI plan to hire an average of 39 agents, compared to an average of 91 agents for those not using it. This difference is even more pronounced in larger contact centers. On average, those who don't use AI will need 2,3 times more agents than those who do.
Combining the average annual savings across all contact centers yields a savings of $4,3 million[3] in employee compensation at companies planning to use AI. This saving does not come from job cuts, at least in the short term, but rather from not hiring new agents.
These results indicate that, rather than eliminating jobs, implementing generative AI in contact centers can lead to a reduction in the need for new hires, while maintaining or increasing the number of existing jobs.
To learn more: discover why some believe that it AI should be bannedhow to keep a strengthened mind in the face of AI and how AI is locking in youth employment.