#联合国安理会将开会讨论AI风险问题#
Still didn't run away! The dangers of artificial intelligence, entering the UN agenda
On July 17, it was reported that the UN Security Council would formally discuss the risks of AI this week.
This is the first time that a discussion has taken place, and the first time that a new technology has attracted such international attention so quickly (in the nine months since the outbreak).
Big model AI has swept the world in only 9 months, but it has been frequently pushed to the forefront because of data security, privacy, copyright and other issues.
For example, several employees of Samsung in South Korea leaked chip design information because of the use of AI; Many people filed a lawsuit, saying that many large models used personal privacy information in violation of the law when training; Artists and designers sue large models for infringement; There are also more scams of AI changing faces and simulating sounds; There are even fake precedents made up by AI to deceive judges...
Therefore, a few months ago, including Musk, the inventor of the big model and other technology experts, began to call for the supervision of the big model, and countries also changed the past tolerant thinking of the development of emerging technologies first, and then supervised, and quickly supervised AI from multiple levels such as administration.
There are even AI leaders initiatives that propose the creation of an AI regulator like the International Atomic Energy Agency.
Now, that statement is being echoed by the United Nations.
There are several points worth noting about this meeting:
1. The logic of global AI supervision is basically the same, on the one hand, to prevent and control the new harm derived from AI, on the other hand, it is also necessary to support the development of AI, because it does represent advanced productivity, greatly optimize the efficiency of economic development, and change the social and global pattern.
However, different countries have different levels of AI development and different industry needs, can we agree on the leniency and details of supervision?
2. From the current existing frameworks in China, the European Union and other countries, the basic regulatory direction: to eliminate the uncertainty of the development of AI technology, let it eliminate the "black box" attribute, and continuously improve transparency;
Means: Determine AI risk levels, safety assessments, algorithm filings, pre-set new regulations for data annotation, coordinate license licensing, content responsibility system to promote compliance, and strengthen technical transparency.
But how much can commercial platforms afford the regulatory costs involved? How will it be played? Will there be a gap in the mechanism of various countries, forming regulatory arbitrage? Ultraman, the boss of Open AI, has withdrawn from the European Union due to regulatory threats...
3. Be wary of some countries using influence other than technology to dominate the international AI regulatory framework and form unfavorable results for our AI development.
In short, AI is exactly what scientist Hawking said: it's either the best thing that ever happened to humanity, or the worst. Will it really "lead to the destruction of civilization" as Musk said? I believe that soon the United Nations will give an answer.
After all, according to the plan, in September, the advisory board of artificial intelligence may be established at the United Nations, and many things may accelerate and become clear.
What do you say?
