In its nearly seventy years of history, artificial intelligence has never had this much time attentionAnd finance resource And concrete application prospects.
In these seven decades, its pioneers have had to go through very dark lean periods, alternating with an explosion of expectations, often with disappointment.
Yet they persisted, believing that there was something incredibly important to humanity in artificial intelligence: the possibility of automation Any process is tedious and cumbersome at the time Building tools for predicting and evaluating information higher than the existing one.
But for now, and the efforts of these pioneers seem to be succeeding, many of the great AI scientists are walking away from their work. The last one to occupy the news is Jeffrey Hintona senior scientist in the field of neural networks has been working on their development for a decade Google. A few days ago, Joshua Bengio and Gary Marcus signed an important letter with over a thousand other industry leaders and, in a separate initiative, many followers ofAssociation for the Advancement of Artificial Intelligence Including Eric Horvitz, chief scientist at Microsoft.
To understand the phenomenon, it is necessary to clarify what is the point. The interests of these scholars vary: The risk of a massive spread of fake newsAnd Automate many functionsAnd Attacks on computer resources multiplyAnd Military use of artificial intelligence, and so on. But these guys don’t make it an ethical issue: In fact, these scientists haven’t had too many problems in recent months, when big digital companies have been impressed. Twitter (X.Corp) e Facebook (half), Google (alphabet) h Microsoftfired AI ethicists.
In fact, the scientists we are talking about are making a deep critique of the AI development paradigm. The eruption of their criticism rests on a negative assessment of the effects of competition resulting mainly from Open AIcontrols it MicrosoftAnd Googlerecently merged Deep mind.
In particular, the ChatGPT version by Open AIin the course of the success of many generative AI solutions, has changed the course of development.
The 100 million people who chose to use ChatGPT in the first two months of its release had an unprecedented impact on competition in the industry.
Googlewho was treading water in exposing her artificial intelligence to the general public, had to abandon the wisdom strategy and decided to put all of her power into play. The speed of launching new developments does not allow us to assess their consequences.
The first and most obvious concern is Proliferation of false texts and imageswith the loss of the reliability of search engines: “We will no longer be so sure of knowing what is right and what is wrong,” Hinton said.
At this point it is necessary to acknowledge its existence Two development models for artificial intelligence. The first is what Focuses on competition between firms: In this context, artificial intelligence is aThe technology that controls it is of its own making Thus it developed in the context of inter-firm competition. Competitors must move quickly, gain market share, convince with the credibility of their capabilities, attract investment, and be adopted by users. In this context, everything related to the side effects of using these techniques is considered less of a priority.
The second paradigm of development considers AI not so much as a technology, but in a sense See it from the point of view of scientific research. In this context, scholars work in the direction of collaboration, exchange of knowledge, dissemination of experimental results, and increasing society’s awareness of the knowledge structure emerging from their studies.
If there is a flaw in this approach, it is probably inHyper specialization that have been carried over historically. But recently Instead, it is the interdisciplinarity that drives the research. Because the relationship between artificial intelligence and society has become a priority field of study.
The competitive model is likely to be more risky than the cooperative model. And if we are to follow through on the concerns expressed by scientists who have recently distanced themselves from the current development paradigm, we need to relaunch the scientific approach to AI. once again: In addition to being an ethical problem, it is also and above all an epistemological problem.
2 minutes to read
4 minutes to read
“Infuriatingly humble social media buff. Twitter advocate. Writer. Internet nerd.”
In Rome, on June 10, the General Council of the Luca Coscioni Society. Documents calling for individual liberties and sciences, even in the European dimension.
Sciences. The eclipse that proved Einstein right
Science shows that children who are exposed to violence at an early age