The agency, commissioned by the State Department, warned of the need to follow a series of recommendations to limit the risks of artificial intelligence.
Artificial intelligence. Photo: Canva.
He Department of State US appointed GladstoneA technology company processes reports on potential threats to national security Artificial Intelligence (AI). In this statement, the company issued a withering verdict: If we don't act “swiftly and decisively,” humanity risks extinction. on its sides The threat to global stability of this technology is compared to that of nuclear weapons.
Time magazine exclusively published the report's findings ahead of publication. “Action Plan to Increase Safety and Security of Advanced AI”, Commissioned by Washington in November 2022 under a $250,000 federal contract. To carry out this study, Gladstone AI spoke to over 200 government officials and leading AI companies Meta, OpenAI and Google DeepMindas well as with other experts.
In this sense, many AI security workers in labs have raised concerns about the wrong goals of administrators when developing this technology. “The rise of advanced AI and artificial general AI has the potential to disrupt global security in ways reminiscent of the introduction of nuclear weapons” The report states.
World governments are beginning to debate how to regulate I.A, technology companies are already laying the groundwork for the next generation of more powerful systems, increasing their capacity exponentially. This situation increases the fear of using this technology, which is increasing in the United States 80% of people believe AI will lead to a disaster, and 77% of Americans believe the government should do more to control it. According to the AI Policy Institute.
Artificial intelligence. Photo: Unsplash.
The State Department did not respond to Time magazine's request for comment. However, the first page of the report makes it clear The advice given “does not necessarily reflect the views of the US State Department or the government.”
This may interest you:
From home theaters to virtual museums: 5 best rainy day activities, according to ChatGPT
Report recommendations to the United States government for regulating the use of AI
However White House The company is yet to take a public stance on the proposal Gladstone AI, these are some of the things included in the report for world governments:
- Set limits to the computing power that can be used to train AI models.
- Apply legal sanctions if this limit is exceeded.
- If companies want to deal with it, they have to apply for special permission.
- Prohibit publishing the “weights” (internal structures) of powerful AI models. Violation of this rule is punishable by imprisonment.
- Greater export and production control of AI chips.
- Increased federal funding to research how AI can be safer.
