Yes, artificial intelligences have arrived to “turn everything upside down”. That’s a fact. In a very short time, AIs have developed so much that they are already capable of writing movie scripts, political speeches or even your term papers. And, while they still have their “little problems” and limitations, the truth is that artificial intelligences will only get better and better over time, whether we want them to or not. But how “better”?
As we reported a few months ago, one of the main concerns in the world of work today is that AIs will reach a point in their development where they will be as effective or even more effective and efficient than flesh-and-blood workers. A disturbing scenario that seems to be at least partially confirmed by a new study conducted by the University of Pennsylvania and OpenAI, the parents of the most widely used text chatbot in the world: ChatGPT.
After conducting joint research, the institution and the software company claim that 80% of the U.S. workforce could lose 10% of their workload due to the introduction of ChatGPT or chatbots with a similar purpose in their companies. In addition, they found that at least 50% of the tasks of about 19% of workers would be jeopardized by text-generating technologies.
They also found that the advance of AIs would eventually affect almost all sectors, with higher income jobs standing out, which would be much more exposed to chatbots due to the savings from (at least partial) replacement of their functions by an AI.
The researchers clarify that they are not claiming that the tasks of workers with the greatest “exposure” (whose workload could be reduced by at least 50% with an AI) would be fully automated by AIs, but that it would serve (in theory) to save workers “a significant amount of time in performing a large part of their tasks”.

According to the study, while some of the jobs with the highest exposure are writers, authors, mathematicians, interpreters, accountants, and legal secretaries, those with no exposure at all are tilers, meat packers, miners, railroad maintenance technicians, cooks, and mechanics, among others.
In a Twitter thread, one of OpenAI’s researchers, Pamela Mishkin, highlighted the following about the study, “Current GPTs can do a lot. In recent years we’ve seen them get better and better at solving increasingly complex tasks with fewer and fewer examples of increasingly unrelated tasks. The paper examines this trend, not any particular model currently available.”
What will end up happening? Although researchers have qualified that this “can save time for workers”, the harsh reality is that many employers will use this reduction in time to their benefit and to the detriment of workers, either with a greater workload or a reduction in their salary. And all this if they do not directly reduce their workforces to put a chatbot in their place, even if the quality of the work would be lower.
Some of the links added in the article are part of affiliate campaigns and may represent benefits for Softonic.