Will the shift towards generative AI happen without data scientists?

Generative AI sparks discussions and discussions at any trade show or event today. Regularly during these exchanges, Data scientists are singled out because they haven't seen – perhaps because they're focused on their daily tasks – the effects that large language models can generate (LLM or Large Linguistic Model). They seem to completely miss the fundamental revolution brought by IAGen…unless that revolution simply doesn't need them.

Yann LeCun, head of artificial intelligence at Facebook, pointed out this idea by confirming this The real revolution lies not in operational improvement, but in the next step, a conversational interface capable of replacing the smartphone. So what is the place for data scientists?

From the Personal Digital Assistant (PDA) to IAGen, customer relationships have been reinvented

The concept of a personal digital assistant was born in the mid-1990s with PDAs, the direct descendants of which can be found today in our smartphones. For 20 years, these tools have played an intermediary role in customer relationships.

Since the advent of the first conversational agents, the world has witnessed significant technological advances. However, there is still a long way to go before we can offer an omniscient and omnipotent digital assistant, one that could be the equivalent of a smartphone-equipped user. clearly, What all these agents lack is their inability to learn on their own. So, Enriching their capabilities or knowledge base depends entirely on the good faith of the publisher in this regard.

In this particular case, the revolution brought about by IAGen will reshuffle the cards. Ironically, it is by listening to “non-experts” in AI that the full extent of the revolution brought about by generative AI can be gauged, i.e. the birth of a new “operating system” for customer relationships via personal assistants.

Towards a new “operating system” based on generative artificial intelligence

As everyone knows, An operating system is a software layer that allows humans to control a device, through a more or less advanced language. Historically, it was based on a text interface (UNIX and its router), and would evolve by incorporating a more graphical aspect that made it possible to control the new peripherals provided by computers (mouse, sound card, graphics, network).

Even if today the word “mentor” regains real vitality in the news of this new world of Masters and Management LLM, thanks to generative artificial intelligence, a new highly sophisticated software layer is emerging, which makes it possible to create virtual assistants that occupy every moment of our lives. Spirits. However, data scientists who are not accustomed to the practice of acculturation with other teams take over the machine. It does not show the direction we should be aiming for, as those who have brought the value of innovation and transformation have, until now, had the ability to blend many disciplines. For their part, senior architects do not have the authority to “sell” this kind of rhetoric… There is a risk, for the company, of missing the mark.

IAGen-based virtual assistants: the new servants of the digital age

These new virtual assistants, which we were able to taste at CES thanks to the presentation of the Rabbit-R1, are digital tools that, in our pockets, take care of the management of our daily lives, like the famous butlers in popular culture (Higgins in Magnum or Alfred in Batman). This new operating system will be able to offer, like its “ancestors”, Windows, OSX or Android, the ability to add interfaces to services or extensions. Like the “drivers” that control extensions to the Windows operating system, digital services in the form of APIs will allow businesses to expand the capabilities of these new virtual assistants.

To illustrate, let's imagine that we ask our virtual assistant to prepare a little romantic weekend getaway for us. He will be able, if “connectors” are available, to verify our financing capacity, the availability of transportation and hotels, and book all of this for us. Booking, Air France and our banker still need to provide the necessary services.

While it's still too early to know whether this virtual assistant will be specialized (like Rabbit) or built into a smartphone (like SIRI), one thing is certain: Information systems that are not open will miss the turn.

Navigating the Digital Future: Tips for a Successful Transformation

The implementation of generative AI will rely on the use of existing LLMs – not custom-created – without resorting to the help of data scientists. Development skills will be used to design a generative AI platform. The latter will then depend on the orchestrator and APIs.

The question that will then arise is to provide the appropriate drivers for the most important operating system for the assistants. To date, this targeted model has been hampered by delays in the availability and exposure of these interfaces. Hence the emergence of the Rabbit-R1 model capable of complementing the LLM with LAM, a “large business model”.

However, the cost of using these tools is more expensive and complex than a simple API call. We know that the way generative AI works is to prioritize shorter and faster arriving services. Therefore, API services will be more successful than any other specially developed technology. This is why it is important to understand that IS API programming should be considered a priority.

For the company, the challenge is to understand data-related issues, rethink its exposure and business, so that it is integrated as a distinct extension of these AI operating systems. For this reason, there is a greater need for developers than for data scientists.

So, it's time for the company to turn the corner and open its own IS, at the risk of ending up like the bistro on the N7 when the A6 motorway opened.