Elon Musk, together with experts in the field of artificial intelligence (AI), demanded to ban the training of powerful neural networks, since the loss of control in this area could threaten humanity.
This is reported in an open letter from AI experts dated March 22 on the website of the non-profit organization Future of Life Institute.
The experts in the letter called on all AI laboratories to suspend for at least 6 months the training of AI systems that are more powerful than GPT-4. It is noted that if such a pause is refused as soon as possible, governments should intervene and impose a moratorium.
Experts suggest using the pause to develop common security protocols that are needed for the design and development of advanced AI systems that can be subjected to checks and control by independent external experts.
The letter also called for the creation of new AI regulatory bodies and state control systems to ensure the safety of new systems.
Among the signatories of the open letter are Apple co-founder Steve Wozniak, leading AI experts Stephen Russell and Joshua Bendjio, as well as more than 1,100 specialists in the field of information technology and machine learning.
GPT-4 is the fourth version of a large language neural network model from Open AI. Compared to the previous version (GPT-3.5), the new model has received five radical improvements: it understands very complex queries, processes up to 25 thousand words, understands not only texts, but also images as input information, has a very high degree of manageability, and is also more secure and consistent.
Experts note that AI is progressing so fast that business, the community and the state do not have time to adequately assess the level of risks that such neural networks carry.