There is a chatgpt fever among public officials. What we do not know is how it will affect us as users

In the City of Bétera (Valencia) they have turned Chatgpt into one more employee. He told so In the country Marcos Gallart, the Deputy Secretary of the Urban Planning Area of said town. According to him, AI saves “20% of the time in the writing of reports.” The OpenAI chatbot, like its rivals, allow it, of course to gain time to time, but there is a problem with that adoption. Or several.

Be careful to save time. Although chatgpt can of course Help perform all kinds of administrative tasksGallart himself explained how formation, accompaniment and how far this type of tools can be used. And there is the problem, because the dimension and complexity of public administration makes this type of processes of adaptation and use of new technologies suppose a colossal challenge.

There are no standards. Despite the EU regulatory obsession and Spain in the field of AI, there is no clear regulation that guides officials on how to use AI and how to manage data that are handled with it. Here teachers, health personnel or judges are included that are part of a huge group (1.6 million workers) who can of course use these tools, but very carefully.

To tell the police. In recent months we have proven how the indiscriminate use of AI and confidence in these systems can be a real disaster. The National Police, for example, had been using ia for six years to detect false complaints, but The real reliability of the system was very debatable. In the recent ‘Ábalos’ an AI to transcribe the statements of witnesses and accused in the interrogations, but There were paragraphs that were a gallimatisms. Even more serious was what happened with the IA Viogén system, which was theoretically destined to solve cases of gender violence and It has ended up causing mortal tragedies.

Spain wants in administration. Meanwhile, the Ministry of Public Function advertisement These days his intention to incorporate AI to the public administration. To do this, he raised a “sovereign platform of AI” with an investment of 14 million euros. His mission, among other things: to expedite procedures in the administration to provoke the one according to Minister Óscar López will be “the biggest revolution of the general administration from the Internet.”

A nightmare for privacy (and security). Someone asked ChatgPT about personal issues is already delicate for both the answer – which may not be accurate or even correct – and for the fact that the chatbot keeps that data. The thing is especially serious If an official introduces documents of all kinds in this or other chatbots to summarize or analyze them: if those documents contain sensitive or private data, they are under the control of these chatbots, which in fact They can filter them By mistake to other users.

Citizens, possible victims. That makes AI become a double row weapon for public administrations and citizens. On the one hand they can help expedite efforts and even solve problems much more efficiently. On the other, a Incorrect use Chatgpt and its alternatives can make private and personal data They end where they should not, or even something worse: That the result of a management is wrong because an official used it and considered that it was correct without adequate supervision.

Zero Data Retention. In this sense There are many services offered by plans without data retention. (ZDR, Zero Data Retention) This is: The data you enter will not be stored on the supplier’s servers. OpenAI It has it In its Chatgpt Enterprise service, a business version precisely designed so that professionals can use Chatgpt’s capacity without fear of data leaks. Microsoft It is another example.

Public administration is more “released”. In March It was approved he Draft law for ethical, inclusive and beneficial use of AI. That document was an adaptation to our legislation of the European Regulation of artificial intelligence Approved in March 2024but there we found a contradiction. It was criticized that the law of AI It was too restrictive “The EU had to back down,” but the funny thing is that it was not with the public administration: there the regulation is warm, does not specify bad clear uses and only considers minor offenses those referred to the deployment and use of the systems (articles 25, 26 and 27).

Image | Pickpik

In Xataka | The EU regulatory obsession raises a world in which AI will have two speeds. And Europe will lose