Nearly two years ago, the generative variant of artificial intelligence (GenIA), which most of us identify with ChatGPT-like “chatbots,” sparked a furor over the seemingly limitless potential of a new generation of algorithms capable of performing almost any type of task. . task and revolutionize the labor market. But today, the data represents a bucket of cold water: the percentage of offers linked to GenIA remains well below 1% of the total. Can we then speak of a “puncture”? A recent study published by Indeed describes this supposed “failure” and reveals that companies Yes, they are relying more and more on this technology, but not at the speed we expected. Nor in the planned jobs. And there we find a surprise: Spain is one of the countries with the highest vacancy rate of this typealmost double that of large economies like the United States or Germany.
To understand this data, we need to move away from the exaggerated specter of “hype” (whether positive or negative) about the impact on the use of algorithms based on large language models or, in English, “large language models” (LLM). Since Open AI introduced Chat GPT, media (and stock market) noise has accompanied both the companies developing them and those making the hardware needed to meet their demanding computing demands.
Avoiding this high technical and energy cost from slowing down investments explains, in large part, the interest in “selling” opportunities to use a technology which was not as mature as we thought, from less for its implementation in the business world. could replace human labor. Almost two years after the presentation of the LLM, the noise has been moderated and the analyzes are much more pragmatic.
This is facilitated by regulations which, for a change, have moved ahead of technology itself in drawing its red lines. For reference, remember that the previous generation of algorithms with a clear impact on employment, those used by digital platforms (key in activities like drivers and food delivery) took almost a decade to be the subject of clear labor legislation, which has excluded some companies from the “gig economy”.
Indeed’s analysis, carried out by Australian economist Callam Pickering, based on data collected in nine countries, is based on a very simple postulate: “generative AI cannot yet take your job, but someone who knows how to use it can.” Even though the demand for these professionals is very low, it is increasing intensely. And this reveals a clear direction in this advance, which is also particularly intense in Spain..
Although GenIA-related published job postings represent just 0.7% of the total, this percentage far exceeds that of the United States, Germany and France. In addition, it has multiplied by 4.3 over the last year, more than the first two (even if French companies, starting from a residual rate, have increased their demand proportionally more). Thus, our country is only surpassed by Singapore and Ireland..
Very uneven weight
Does this mean that our country is a powerhouse in the use of generative artificial intelligence? The indicator produced by Indeed presents a double bias which can move it away from the “real” demand for jobs: on the one hand, the volume of job offers published on the internet, and on the other, their sectoral distribution . Even though candidates are increasingly turning to digital job searching, The most common route remains that of recommendations from acquaintances and friends.particularly in activities that require lower qualifications, such as many positions in hospitality and construction.
Furthermore, the impact of artificial intelligence, whether generative or not, is very unequal depending on the type of job. and in many of them it is practically non-existent, while in others it is much more relevant. This is why it is interesting to analyze the percentage of offers per job, to discern the intensity of demand. However, the breakdown of statistical data from this perspective not only does not reduce Spain’s position, but on the contrary places it in the lead.
The sector with the highest percentage of GenIA-related job postings across all countries is data analytics. In Spain they record a rate of 12.7%, which is not only exponentially higher than the average of all offers, but it exceeds that of the eight other countries analyzed.
The differences with the rest of the high-demand sectors are notable. The second position is occupied by marketing (2.4%) and the third by scientific research (2.2%), while software development, which in other countries is becoming more relevant, is relegated to fifth place.
This heterogeneity between labor markets is one of the main conclusions of the study. “Data analysis is the sector with the highest proportion of positions in the nine countries analyzed. Scientific research and software development also take top positions in eight of these countries. However, beyond these three sectors, the use of GenIA varies considerably between countries,” the report said.
Beyond ChatGPT
What is happening is that scientific research, to the extent that it involves many general, “creative” and poorly automatable skills, is not considered a cluster of jobs linked to generative artificial intelligence. Something similar happens with architecture, industrial engineering, and even arthritic activities. At the same time, the expectations generated in jobs considered more automatable, such as accounting, insurance or medicine, have been defraudedpartly because of ethical doubts and uncertainties about how to apply new regulatory frameworks (which requires us to rethink the security and privacy of sensitive data and the biases they can generate).
Pickering compares the percentage of potential use of GenIA in each sector with the actual weight in offerings and offers an explanation for the surprising mismatch: this technology specializes beyond the path shown by ChatGPT. “There are tools specifically designed for tasks like image creation or video processing.which may explain why GenAI positions are higher than expected in sectors like art and entertainment, or even architecture,” he says.
This leads to a double conclusion: on the one hand, the “rapid rise” of generative artificial intelligence has been driven mainly “by sectors in which skills have a relatively high probability of substitution”, although there were “notable exceptions” which could be explained by the advancement of the tools themselves.
However, job openings “are highly concentrated in a relatively small number of sectors”, suggesting that even as GenIA technology advances, its practical value “remains unproven in many sectors, even those with high exposure” . These conclusions, at least in the Spanish case, are fully consistent with public statistics such as the Survey on the Use of ICT in Businesseswhich revealed a very low weight of generative AI (even compared to other varieties of artificial intelligence) but it is practically the only modality that has increased over the last year.
