Brazil recruits OpenAI in brave bid to slash court battle costs

Brazil has apparently hired OpenAI in an attempt to help prosecutors get better at handling cases and save the administration money. The American super-lab's technology will be used to "expedite the screening and analysis of thousands of lawsuits," by "mapping trends and potential action areas for the solicitor general's …

[cyberdemon]

Finally a meaning to the name of the film

Monty Python's Brazil, loosely based on Orwell's 1984, where computer automation of government functions meant the wrong man was pursued as a terrorist, nobody understood the opaque and inscrutable government system but went along with it.. But the film had nothing to do with Brazil, until now!

And you thought you couldn't make it up

[Anonymous Coward]

Surely AI can trained to rubber stamp?

However much this shrinks the workload, the bureaucracy will find a way to fill the gap. And more, because they'll have to go back and do it all over.

[An_Old_Dog]

When You Do Not Know the Correct Answer(s), You Cannot Test

How will the humans know if their ML system goes off the rails -- as these ML systems tend to do -- and gives them bad answers, when the humans don't have (and cannot obtain) correct answers to check the computer's output against?

Answer: the humans cannot know.

Using ML systems in such a role is stupid and foolish. But, using these things sounds cool and modern, and that's what matters most to many decision-makers. Because it's one louder.

There are potentially-effective and "safe" roles for ML systems, but this isn't one of them.

[Anonymous Coward]

AI / ML firms gaining leverage over government

Great,

We are starting the age of government integrating AI / ML in order to save costs and be able to run its services. In the short term, they'll save some money. In the long term, they'll lose the people and skills to do it the old-fashioned human way. The savings made on wages will be spent on other government projects and departments or tax cuts. Even if later on they decided to "re-humanize" the justice department, they'd lack the budget to do so.

That creates a triple lock-in:

* loss of understanding how the processes and organization runs

* loss of skilled workers to run the department

* loss of financial ability to run the department without using said AI / ML

Due to all of three above points and no single bit of standardization, the only company able to allow to continue the department to run will be the initial AI / ML company / companies gaining the contract. Once your justice department becomes "locked in hostage" of a mega-company, your country in effect loses the ability to start legal cases against said company and it can ignore local laws at will. It will also gain the ability to block, slow or hamper more strict rules and legislation against such monopolistic abuse or to limit the ill effects of AI / ML on society at large.

Due to chronic lack of money in most governments, the same holds for the department of education (replace part of the teachers by AI / ML "assistants" and build down the budget), military, infrastructure (organize the smart grid needed to run longer wild old infrastructure rather then build new ones and allow for large amounts of electric vehicles to be charged)... That'll free up money that will be needed once AI / ML organizations take a large portion of gross national product away from local smaller companies and transfer it to their own pockets and those of large foreign multinationals. The failed small companies will no longer pay tax and the government will be utterly teeth-less to legislate or sue these mega companies do to their lock in and dependency to run the country. Great? Not so!

[tiago.pelicari]

I'm Brazilian. When I first saw this news on local channels, I imagined that the case would be to improve the lives of citizens, who are often wronged due to lack of attention from the judiciary. Now it's clear that they just want to save money.