At the last meeting of the Davos Forum (Switzerland), where the main businessmen and politicians who govern the world gathered, it seemed as if the planet had paused its orbit to listen to them talk about generative AI with a mix of realities and assumptions that were not remembered in previous technological disruptions. “This changes everything,” was heard.
Or not? It took a century after the patent of the steam engine to feel the industrial power on the other side of the Atlantic. ChatGPT only needed a few weeks to win over 100 million users, an unprecedented speed.
Major investment banks and consulting firms were quick to announce the discovery of a new spice route. Goldman Sachs estimated that 7% of American workers are likely to be replaced by AI. Translated to all advanced economies, these estimates put the number of jobs in this risky line at around 300 million.
In exchange, however, it promises to reactivate a social stratum weakened by low productivity whose future will depend on its conversion to the new demands of the market, through training. “AI will have a catalytic effect on worker productivity and will take advantage of a technology that is not a substitute for human skills,” argues Gregorio Izquierdo, general director of the Institute of Economic Studies, IEE.
According to the McKinsey consultancy, generative diversity (GenIA) could add between 2.4 and 4.2 billion euros a year in terms of productivity worldwide. And this trend places Spain in a position where productivity is shackled. “We have a problem: it is very low and proposals such as raising the minimum wage have not helped,” explains Gloria Macías-Lizaso, a partner at McKinsey, who resorts to demographics and their fate: “We must think about ageing. In 2030, 23 million Europeans will retire. We need workers who learn these new skills.”
Even if they don’t know, they can do it
To what extent does AI worsen some problems and mitigate others? GenAI will avoid routine work and leave employees with more time to devote to more valuable, more qualitative activities. Those who don’t know how, will be able to do it: “Someone can be a graphic designer even if they lack drawing skills, AI will increase creativity and human potential,” says Daniel Castro, vice president of the Information Technology and Innovation Foundation (ITIF).
Those who lack the knowledge or skill to write can disguise their lack of talent with it. Others will develop software even if they don’t know the basics of programming. The code may not be perfect, like an aspirational text, but they can revise it to make the lines fit. “They are gulping down knowledge and skills that have taken lifetimes to master,” says programmer James Somers.
Someone can be a graphic designer even if they lack drawing skills, AI will increase creativity and human potential
Daniel Castro, Vice President of ITIF
But it is not a panacea, there has never been one in history. The effects on workers are not clear, the studies range from the apocalyptic to the brilliant. For OpenAI, the company behind ChatGPT, the jobs that are most at risk are precisely the best paid. Those who earn six figures a year are six times more exposed to dismissal than those who earn 30,000 euros. “The risk of algorithms falls on those workers we have called white collar workers,” says Enrique Dans, professor of Innovation at IE Business School.
For example, in a Harvard University study, staff at the Boston Consulting Group (BCG) randomly assigned to use GPT-4 in consulting work were much more productive than their colleagues without access to the tool. “AI has to be a complementary technology, not a job destroyer,” sums up Carlos Gutiérrez, Secretary of Trade Union Training at Comisiones Obreras.
On the other hand, in the round of questions after the presentation of a Mapfre report on the evolution of AI, its experts commented that in one way or another all workers will have to adapt and learn to use its tools. As everyday and naturally as a mobile phone and its applications.
And who, due to their studies or age, cannot adapt? “They must be protected. Whether with a universal basic income, a system of aid or unemployment benefits,” adds Gutiérrez. At Deutsche Telekom, workers managed, as a preventive measure, to ensure that algorithms could not fire any employee without human intervention. In this scenario, AI would pose a labor Darwinism. The risk is not so much losing one’s job because of this technology —argued Hadi Partovi, founder of the computing firm Code.orgin Davos—the risk “is losing the job to someone else who knows how to use AI. That will be an even bigger shift.” An android will not come to replace the employee, or not only, but someone better educated and two or three times more productive will come first.
“The greatest danger,” says Valerio De Stefano, head of the Canada Research Chair (CRC) in Innovation, Law and Society, “will be suffered by workers who perform routine tasks, especially administrative tasks and in areas such as human resources or accounting.” “I hope that unions will rise to the challenges posed by these technologies and begin to question how they will be used to replace workers,” he warns.
Adaptation, synonymous with salvation
“The most affected segments will be those with the highest qualifications. But it is also true that these employees are, in turn, the most adaptable (in the face of dismissal and the need to reinvent themselves) because they are better trained,” admits José Montalvo, professor of Economics. To a certain extent it is paradoxical that one of the main barriers to squeezing the economic benefits of AI is precisely the lack of qualified personnel to develop and manage it. As with other technologies, yes, but this one drives all the others.
At Deutsche Telekom, workers succeeded in making it impossible for algorithms to fire any employee without human intervention
One novelty of the wave is its speed of advance and therefore the short time for workers to adapt. But it also plays in favour of the human factor, which is still necessary. According to Carl Frey, associate professor of AI and Work at the University of Oxford, “AI requires human intervention for its initiation, refinement and editing of results.”
To begin with, the low quality of information on the Internet for training algorithms represents another barrier to development and also a job opportunity. Machines are needed that learn from “more concise and better selected” data, concludes Frey.
Whoever sets the rules, wins
These brave new technologies—as economist Nouriel Roubini called them—that could contribute to human growth and well-being, also have destructive potential: misinformation, discrimination, permanent technological unemployment and even greater inequality. “The digital divide is a great generator of inequality (…), a democratic movement is needed on how to regulate these systems,” warns Stéphane Klecha, managing partner of the investment bank Klecha & Co. “As we have seen in many areas, whoever sets the rules wins.” But the U.S. regulates a posteriori and China does not even regulate, while Europe is trying to define laws that protect workers and users.
There are always caveats, though: “American unions are active in many industries and have negotiated labor protections and restrictions on intensive surveillance, while also participating in joint labor-management technology committees to discuss training and work organization issues,” notes Virginia Doellgast, a professor of labor relations at Cornell University. For example, agreements that restrict the use of generative AI to write scripts or protect actors’ voice and image rights are resistances to technological abuse.
