Artificial intelligence, according to Bill Gates, could cause the death of a billion people
Up to a billion people could fall victim to the rise of artificial intelligence, warns Microsoft co-founder Bill Gates. Given the possible risks, he therefore considers it absolutely essential to adopt appropriate legislation. Leading developers have also recently spoken out in favor of slowing down the development of this phenomenon. They are reacting, among other things, to independent hacker attacks carried out by software with artificial intelligence.
Bill Gates said in September on the American network NBC that artificial intelligence could cause the deaths of a billion people. He therefore advocates for legal regulation of this technology.
"Artificial intelligence is undoubtedly powerful enough that it could trigger events that would claim a billion victims,"
explained the 70-year-old Gates when asked whether this technology could wipe out humanity.
A lot of attention was recently drawn to Anthropic researcher Jacob Coxon, who decided to leave both the company and the entire AI field. According to Coxon, people working in artificial intelligence development "sincerely believe that it could kill us all by the end of this decade".
Technology isn't to blame. The risk is in people
According to Gates, the biggest problem is not artificial intelligence that would decide to act on its own. More attention should be paid to people who can already use powerful AI tools for attacks, as he claims. If someone uses artificial intelligence to create malicious code, fraud, or prepare a cyber attack, the problem may not arise directly in the system itself. The model may "just" fulfill the user's request.
"There has never been such a powerful weapon as people with malicious intent using the latest artificial intelligence tools. No one believes that self-regulation is sufficient,"
Gates added. However, he clarified that it would be relatively difficult to reach 100 percent of people on the planet.
He therefore proposes that creators of advanced AI systems implement monitoring of their use and keep records of what is happening in the systems. Such oversight could help, for example, in detecting attempts to misuse AI for cyberattacks or bioterrorism. It is therefore not only about controlling the model itself, but also about monitoring the way people use it.
According to Gates, the rules should be set by legislators. At the same time, he stated that the United States should ensure the implementation of such measures and pointed out that China will want a similar approach.
"It is necessary for the police and politicians to engage in a discussion about what security measures and surveillance should look like, and it must be a mandatory part of the system,"
Gates said.
Interstate Meeting of the Most Powerful
This is also why China and the United States agreed at their recent heads of state meeting to launch a dialogue on artificial intelligence about the benefits and risks of this technology. Both countries will also create a communication channel to address potential AI-related incidents. According to Beijing, as AI continues to develop, both countries should strengthen mutual communication and strive for common positions that correspond to new technological developments.
For example, the July incident drew great attention when OpenAI's artificial intelligence penetrated the systems of Hugging Face. This was followed by the disclosure of several similar incidents at OpenAI and competitors such as Anthropic and Meta.
Nvidia offers a solution
In this context, Nvidia has also introduced a new software platform designed to help artificial intelligence developers ensure that their AI models don't get out of control.
"According to our information, this new security platform could prevent a breakthrough if it were already being used in leading development laboratories at an early stage to evaluate models,"
stated according to ČTK the company's vice president Justin Boitano. The new software called OpenShell, according to him, allows developers to "formally verify that an AI agent has exactly the permissions it needs for its work, and no more".
Part of this new platform is also a separate security layer called Sentry, which continuously monitors the activity of AI agents and can immediately intervene if an agent begins to attempt to exceed the set boundaries.
Nvidia's CEO Jensen Huang recently claimed that nothing dramatic is happening. At the same time, however, he pointed out that it is in his company's best interest to ensure that AI products are developed responsibly.
"The success of our business is directly linked to the safe deployment of products and services. If we don't continue in this, our value will decline,"
wrote ČTK.
Sources: original article, ČTK, NBC
