Large American groups are replacing thousands of employees with software that manages office tasks and decisions previously made by human beings in total autonomy. Thus the algorithm risks “archiving” white collar workers too.
More harmful than a tsunami, more lethal than Covid. The pandemic was nothing compared to the “mortality” rate of Artificial Intelligence. There is no sector that is safe. From banking to automotive to retail, every day is a war bulletin. Thousands of employees fired overnight and replaced by a chip, or rather by Agents, as they are called, or platforms to manage artificial intelligences that act autonomously without human intervention. Unlike ChatGpt, they don’t just respond to passive commands but take the initiative.
How Agents work
For companies it is a boon. They are not simple assistants, but true algorithmic entities capable of perceiving, reasoning, planning and acting to achieve complex objectives with minimal human supervision. We talk about customer service features, marketing features, content creation features, and more. They can interact with the environment, collect data and use it. Let’s take the example of a contact center: the Agent will automatically ask the customer several questions, search for information in internal documents and respond with a solution. Based on users’ responses, it will then determine whether it can solve the problem itself or pass it on to a human. He also has a memory that allows him to remember the results of his actions.
Continuous learning through “machine learning” techniques allows them to adapt their behavior, correct errors and improve performance over time. Compared to previous generations, an AI Agent is proactive, that is, it does not wait for the command at every step but once a complex objective has been set, it decides, plans the strategy and executes it autonomously, adapting to unexpected events without requiring human supervision.
An assistant without limits
If you are already confused, here is a basic example: services like Siri and Google assistant are types of Agents that learn and interact with users. These software have an advantage that makes them successful: they have no working time constraints, they are not unionized, they do not ask for salary increases, they do not get sick or go on maternity leave. In short, I am the ideal assistant.
A revolution already underway which, according to the analysis company McKinsey, could be worth 4,400 billion dollars in added value per year in the world. An epochal turning point which however comes at a high cost on the job market.
Layoffs and hiring freezes
But the Agents who have joined the company are real killers with a license to kill even the highest professionals. In recent days, Amazon announced the cut of 14 thousand employees, the first tranche of a plan that involves the reduction of 40 thousand positions, to free up resources and invest aggressively in artificial intelligence.
JPMorgan Chase and Goldman Sachs are severely limiting hiring for the benefit of AI, while the CEO of the Ford automotive group, Jim Farley, has warned that it will replace half of its white-collar workers. Brian Chesky, number one at Airbnb, also stated that “many companies are proactively freezing hiring, imagining a downsized workforce.”
A machine that replaces man
Like Walmart, which intends to remain at a standstill for three years, while Marc Benioff, CEO of Salesforce, an American cloud computing company, after cutting 4,000 positions, said he needed “fewer heads” because 50 percent of the work in the company is already done by AI. In June the group introduced Agentforce 3, a platform that allows companies to delegate to Agents the generation of quotes, the onboarding of customers, the analysis of complaints, making decisions based on business rules. The list of “slaughter” grows longer every day. UPS has eliminated 48 thousand positions in the last year, 14 thousand in management and Meta has made official the cut of 600 positions even in the AI division. Goldman Sachs estimates that 6-7 percent of US workers could lose their jobs due to the adoption of artificial intelligence.
The first data and academic analyses
The Stanford Digital Economy Lab found that entry-level hiring in “AI-exposed jobs” fell by 13 percent. The report states that software development, customer service and clerical tasks are the most vulnerable. “We are at the beginning of a development that will have a significant impact on the labor market,” said Bad Levanon, chief economist at the Burning Glass Institute, a research firm that focuses on changes in the economy and employment.
The sectors that are already changing
Among the sectors in which the integration of these software is being studied or tested is the manufacturing industry where intelligent systems can monitor the functioning of machines, predict maintenance needs, and optimize production processes. All of this can translate into increases in productivity and reduction in production downtime. Another sector is that of consumer goods. Here the Agent can manage warehouse stocks and support marketing campaigns, generating, for example, promotional content for new products. In the automotive industry, however, software can monitor the performance of vehicles, facilitating the management of warning systems. In healthcare, AI can give the first answers to patients, help them identify and book the best specialists in time, manage healthcare documentation and define a prevention or treatment plan.
The different families of Agents
Agents are not all the same. There are those with “simple reflex”, the most rudimentary, which act based on the current perception of the environment and have no memory. This is the case of an anti-spam filter that blocks emails with a keyword in the subject. Then there are the “model-based reflex agents” which instead have memory. An example: a robot vacuum cleaner that moves around a room must remember which areas it has already cleaned. “Objective-based agents” do not simply respond, but seek the best path to reach the result, as in GPS navigation systems.
The most sophisticated versions are “Utility-based Agents,” which evaluate different paths and potential outcomes and choose the action that maximizes long-term gain; and “Learning Agents,” who learn from experience and continuously improve their performance. The new frontier, however, is that of “agentic Agents”, that is, groups of two or more Agents who collaborate and communicate with each other to achieve an objective. Among the first applications of this type, in the world of business financial platforms, are Vivid’s digital collaborators, designed as real colleagues with their own email address and chat to interact with them. They also have human names – Anna, Lucas and Emma – and help SMEs and professionals improve productivity by managing appointment calendars, analyzing contracts and making financial reports.
The final question
In short, two things seem to emerge clearly. The first is that we are entering an unknown phase. The introduction of robots in production lines which began in the 1980s aimed at eliminating wearisome, tiring and repetitive jobs. Today, artificial intelligence directly threatens the employment of lower-skilled white-collar workers and highly skilled workers. Many workers will end up integrated into this process and adequate training will be necessary. Will the machine really replace man? The risk (which for some is an opportunity) exists.
