To measure the subjective probability that artificial intelligence condeaches humanity there is an index based on percentages, known as P(doom): It is a subjective variable who use scientists and experts to measure the possibility of catastrophes.
In general, there are 2 major models to measure the risks of AI in relation to human existence.
The first, very similar to what can be seen in science fiction films as Terminator, is known as decisive risk – imminent catastrophe–with a super intelligence that escapes human control and is capable of rapidly improving autonomously.
The second and no less worrying, the cumulative risk – slow -lived–a scenario in which the AI destabilizes institutions in a gradual way; Although it may seem little harmful, it could end at a systemic risk.
From these 2 large models of study on risks, many more edges that have already been studied by AI experts, such as those of a biological war, automated cyber attacks or, directly, the fall in economic and political systems, are broken down.
With them, each expert or scientist specialized in AI has a greater or lesser percentage of the P (Doom) values, some even touching 100%; or what is the same, that the human being is condemned to extinction.
The sponsors of AI, among the most pessimistic
In one of the latest research on this topic, published in the International Journal of Forecastingthe concern of scientists and philosophers about the claim that The risk of extinction of humanity is higher than at any other point in historythe so -called «vulnerable century hypothesis.»
Some of the most important personalities that are in this hypothesis are Geoffrey Hinton and Yoshua Bengio, both considered as sponsors of AI, thanks to their work in automatic learning and artificial neural networks.
Hinton himself, in many interviews, has significantly changed his concerns in relation to AI: in some, he has assured that There is a 50/50 probabilities of a catastrophic eventalthough during the latter it has reduced this to 10%, considering the P (DOOM).
For its part, Bengio places this percentage in 20%, given the possible decisions of an autonomous AI: «We are creating a new form of intelligence on our planet without knowing if your decisions will be compatible with our needs.»
Precisely, in the same study of the magazine focused on forecasts for the future, it is concluded that IA specialists are usually more catastrophic in longer periods of time, although not with the technology that currently exists.
Therefore, the progress towards a general artificial intelligence –AGI) is the scenario most feared by these, including the sponsors of the IA already mentioned.
A pause in the AGI, the claim of some organizations
There are some organizations that demand an immediate pause in the evolution towards AGI, for basically ignoring what the consequences for humanity would be. At this end, the Pauseai organization appears.
It is made up of volunteers and local communities based on a non -profit organization, with the aim of mitigating the risks of AI – including human extinction – and convince governments to pause the development of super intelligence.
«The AGI is not inevitable. It requires engineers with millionaire salaries,» they explain on their page. «It requires a fully functional supply chain and no more complex hardware restrictions. It requires that we all allow these companies to commit for our future.»
Among the most catastrophists, With a P (Doom) they are looking for 100%names such as Roman Yampolskiy, AI security expert and director of the cybersecurity laboratory of the University of Louisville.
A step below, with a percentage of more than 80%, appears Dan Hendrycks, director of the Center for Ai Safety and advisor to Xai, the division in this matter of Elon Musk, as shared in his X profile.
Interestingly, the founder of Tesla and Spacex, one of the most important figures in this industry, does not believe that the P (DOOM) should be placed above 20%, consistent with what has expressed by Geoffrey Hinton, as shared in an interview with Peter H. Diamandis.
In this way, opinions are very found among experts and scientists who drive this new modality of AI. In any case, everyone agrees that there are real dangers if it escapes human control, something that nobody could affirm sharply today.
Know How we work in NoticiasVE.
Tags: Artificial Intelligence, Elon Musk