AI Agents in Business: An Unrealistic Futuristic Vision

Imagining AI agents taking the reins of companies and negotiating with one another seems more like a science fiction scenario than a credible digital transformation. This vision, reminiscent of the famous Planet of the Apes franchise, is far from the actual uses of AI today.

AI agents, often described as more intelligent and tireless digital clones than humans, are envisioned to replace humans in business. However, this perspective is criticized for its excessive anthropomorphism. While some elements of science fiction, such as autonomous combat drones or Asimov's laws, have become reality, the idea of an "Agent Enterprise" remains unrealistic.

An anthropomorphic model of an Agent Enterprise circulates in minds, conversations, and offerings from certain AI SaaS providers. In this vision, AI agents would embody personas within the company, interacting with one another to solve problems, make decisions, negotiate, and even transact. For example, a purchasing agent from one company could negotiate with a sales agent from another, while legal agents would validate contracts and financial agents would approve payments. This scenario would extend to logistics, with autonomous trucks and automated warehouses.

The company would be duplicated by a digital twin, a sort of video game replicating historical human roles and responsibilities, playing out the same processes in the virtual world. Humans would then be relegated to other activities, marking the end of work and the advent of universal basic income, an idea that Singularity has anticipated for decades.

The concept also echoes Nick Bostrom's article "we are almost certainly living in a computer simulation," published in 2003. This article, which resonated widely at the end of the decade, suggests that as soon as computing power allows, humans will duplicate the real world and observe the tribulations of their digital avatars on virtual planets. These simulations would multiply rapidly, almost infinitely. Bostrom concluded that if one accepts that computing power could become virtually infinite, then this scenario of multiplying digital simulations of Earth would occur.

From then on, there is no reason to believe that we are evolving in the only embodied reality rather than in one of these many simulations. All of Silicon Valley, in turmoil, was now convinced they were already living in the Matrix. But back to the topic of agents in business. While this dysmorphic vision of the Agent Enterprise captures the imagination, it is not very realistic.

On one hand, the human organization of companies, that is, the current division of tasks and roles, is far from being the optimal solution for solving a problem. Why gather agents embodying personas and let them debate to align on the marketing and commercial plan of a product launch, develop the annual budget, or establish the factory roadmap within the S&OP process, when a "super application," unique for each of the subjects taken as examples, and underpinned by various technologies, including AI, can address these issues?

This "super application by subject" can issue recommendations, or even take actions in information systems, depending on the degree of autonomy granted to it, integrating all dimensions of each problem.

On the other hand, these "super applications" (which should be reserved for the term "agents," rather than anthropomorphic doubles of human roles in a digital twin of the organizational chart) handle transactional workflows, perform complex optimization, or analyze the dimensions of a decision and act under human control. Systems serve humans, not the other way around.

AI agents, in this definition of super applications, can replace humans in tedious tasks, perform complex optimization, augment human value-added activities, or provide "insight" for delicate decision-making involving risk assessment. However, they cannot substitute for humans in critical decision-making or defining strategic directions. Humans remain in control.

The best example is undoubtedly the anecdote of Netflix's marketing teams who consulted AI about acquiring the rights to the Korean series Kpop Demon Hunters: catastrophic score, negative recommendations from the AI. At the insistence of local teams, Netflix eventually made a limited acquisition of the rights. The rest is history: a global hit! The human audience will never be predictable…

As good as AI engines are, for value-added activities, particularly "charged" interactions such as negotiations that involve building trust between two organizations, and for delicate decisions in uncertain environments that involve taking risks, human instinct and experience remain irreplaceable!