Video games and brain activity; a triple black hole system; neutralizing Skynet

We are in August, which means that Hot Science Summer is two thirds. This week, NASA published an exceptionally pretty photo of Mars, a modified net panorama color to make the sky blue (???). The California health authorities warn the hunters and contaminated game’s trappers after a trapper caught a wild pig with bright blue muscle tissue. Porks and other wild animals may have been exposed to Diphacinone anticoagulant rodenticide. And an international team of astronomers identified the oldest known black hole ever confirmed – the object was present 500 million years after the Big Bang.

In addition, the digitization of the functional brain suggests that video games are more engaging (in the right direction) than visualization of television or the use of social media; Chinese astronomers have teased a triple black holes system from Ligo data; And the company Ai Anthropic reports a new method to prevent LLM from becoming bad.

The pair of black holes may have had a larger ride

Researchers from the Shanghai Astronomical Observatory report that a fusion of recently observed binary hole can have occurred in the gravitational field of a third object, probably a supermassive black hole. This suggests an entirely new model for binary black holes and very compact objects interactions.

The team analyzed an event with gravitational waves called GW190814, the merger of two black holes. The pair was an extreme example of binary fusion; The size ratio between objects was 10: 1. This suggests that they may have been part of a triple system with a supermassive black hole and gradually turned through gravitational interactions.

The researchers noted that if a merger of binary black hole occurs in the gravity of a third object, the orbital movement around this third object would create an acceleration of the vision line: to observers on earth, this period of relative acceleration would create a Doppler alteration of the frequency of gravitational waves. An analysis of the event has shown that the acceleration of the vision line has significantly outperformed a traditional “isolated” traditional model to explain the gravitational wave signal.

The study contradicts all dads everywhere

While governments around the world are working on internet age verification plans and social media prohibitions for young people, researchers from the University of Technology in Swinburne compared the brain activity of young adults while watching television, played video games or used telephone media applications. Using close infrared functional spectrography, they found that different forms of screen use had different models of brain activity.

The levels of oxygen hemoglobin increased more after the use of social media and games compared to the visualization of television, and deoxygenated hemoglobin increased more after the games. This suggests that interactive types of entertainment are more engaging than passive forms.

Dr. Alexandra Gaillard, one of the researchers, said: “Interesting, however, as far as social media is concerned, people said they felt less concentrated – and those who felt less concentrated have also shown lower cerebral activity. On the other hand, the game, which actually means an increase and has shown an increase in deoxygenated hemoglobin, which means that the brain oxygen.

Researchers identify and neutralize evil

The company Ai Anthropic, creators of the LLM Claude, reports a new method to prevent artificial intelligence from becoming evil. The method identifies activity models called persona vectors in neural networks. These models control the character traits of the model. The researchers used two open source llms to test the deletion and manipulation of three personality vectors: evil, sycophance and hallucination, describing the trend of an LLM to invent information.

They call their “direction” method; When they have manipulated the “evil” vector, the model would produce responses contrary to ethics; The direction of the vector “Sycophance” led to answers that flattered the user. The management of “hallucination” has resulted in responses now familiar and not factual with confidence in the guests.

The authors write: “Our method for [steering] is somewhat counter-intuitive: we actually direct the model to unwanted character vectors during the training. The method is vaguely similar to give the model a vaccine – by giving the model a dose of “evil”, for example, we make it more resilient to meet “evil” training data. It works because the model no longer needs to adjust your personality in a harmful way to adapt the training data – we provide them with these adjustments ourselves, relieving it of the pressure to do it. “”

© 2025 Science X Network