The founders of the company he have a reputation to make bold claims about the potential of technology to reshape the fields, especially the sciences. But Thomas Wolf, embracing co -founder and leading science official, has a more discreet receipt.
In an essay published on Thursday, Wolf said he was afraid of becoming “men in the server” in the absence of a progress in his research. He elaborated that the current paradigms of development he will not give him capable of solving external, creative-solving problems that wins Nobel Prizes.
“The main mistake people usually make is to think (people like) Newton or Einstein were just good-scale students, for a genius to come to life when you linearly extrapolate a top-10%student,” Wolf wrote. “To create an Einstein in a data center, we do not only need a system that knows all the answers, but rather one that can ask questions that no one else has thought or dared to do.”
Wolf’s claims are contrasting with those from CEO of Openai Sam Altman, who in an essay earlier this year said that he “superintelligent” could “massively accelerate scientific discovery”. Similarly, anthropic CEO Dario Amodei has predicted that it can help formulate cures for most types of cancer.
Wolf’s problem with him today – and where he thinks technology is going – is that he does not generate any new knowledge by linking facts previously unrelated. Even with most of the internet available, he as we currently realize that it mainly fills the gaps between what people already know, Wolf said.
Some experts, including former Google engineer François Chollet, have expressed similar views, arguing that while he may be able to memorize reasoning patterns, he is unlikely to generate “new reasoning” based on new situations.
Wolf thinks that laboratories and he is building those who are essentially “very obedient students” – not scientific revolutionary in any sense of the phrase. He is not stimulated today to question and propose ideas that potentially go against his training data, he said, limiting him to answer the known questions.
“To create an Einstein in a data center, we do not only need a system that knows all the answers, but rather one that can ask questions that no one else has thought or dared to ask,” Wolf said. “One who writes” What if everyone is wrong about it? “When all textbooks, experts and ordinary knowledge suggest differently.”
Wolf thinks that the “evaluation crisis” in it is partly to blame for this disappointing state. It tells of the standards commonly used to measure system improvements, most of which consist of questions that have clear, visible and “closed” answers.
As a solution, Wolf proposes that the industry of it “moves to an extent of knowledge and reasoning” which is able to clarify whether he can take “bold counter -factors”, make general proposals based on “small suggestions”, and make “unable questions” leading to “new research routes”.
The trick will be to understand what this measure looks like, Wolf admits. But he thinks he could be worth the effort well.
“(T) That most important aspect of science (is) the ability to ask the right questions and challenge what he has learned,” Wolf said. “We do not need a student A+ (he) who can answer any questions with general knowledge. We need a B student who sees and questions what everyone else has lost. “
