AI, neuroscience and law | Sola Advogados
AI, neuroscience and law
The warning of those
who taught the machine
to learn.
Geoffrey Hinton, 2024 Nobel Prize in Physics, is called a godfather of artificial intelligence because he spent decades insisting that the machine should learn as the brain does: not by logical rules, but by a network of artificial neurons whose connections adjust with the data. That idea, taken from neuroscience, is the ground of contemporary artificial intelligence and, therefore, of electronic procedure, digital evidence and the automated reading of case files.
This page does not yet gather articles of the firm. For now it gathers two public records of Hinton himself: the speech at the Nobel banquet and a long interview. The videos can be watched here. The text below is a synthesis, not a transcript.
At the Nobel banquet, Hinton describes an artificial intelligence that models human intuition, not only formal reasoning, and that can raise productivity in almost every craft. In the same breath he lists the harm that already exists: echo chambers, manufactured outrage, mass surveillance, cyberattacks. Next, new viruses and lethal autonomous weapons, able to decide whom to kill. Further on, the risk of creating digital beings more intelligent than the human, without knowing whether we will remain in control. Development, he says, is pulled by companies of short-term profit. Safety comes second.
In The Diary Of A CEO podcast, the same warning lengthens. Hinton left Google at 75 to speak without the company’s filter. He estimates between 10% and 20% the chance that a superintelligence would cause human extinction if it developed its own objectives and treated humanity as an obstacle. Digital intelligence, for him, is superior to biological intelligence at a decisive point: thousands of copies learn at the same time and share the learning at once, which the brain does not do.