Over the past few months, we have tried to explain how it works, the Act of IT, the European Union, EU legislation on Artificial Intelligence. Now, however, the very institutions of the european're making a further step forward by publishing a draft of the first Code of Conduct for the Artificial Intelligence to the General (AGI).
The present document has been produced by the European Commission, on the 14th of November, and is the first step towards a Code of Conduct that will be adopted at the community level to the adjustment of the models of the GPAI, or the Artificial Intelligence to the General, who are currently still in the development stage. The development in the main companies of the sector. Therefore, the Brussels it seems that it intends not to be prepared: guidelines and will be ready to go by may 2025, ahead of we the coming of the models of the first AGI in the market.
According to TechCrunch, the draft had been published in the last few days is a supplement to the Act, HE and the EU, which entered into force on January 1, but it still represents some of the grey area, especially in the field of IT service. Therefore, the idea of the rule of law program is to clarify the expectations of their own on the subject of the models, the most advanced on him, HE was to give at the same time, the companies that are developing the ability to give feedback to the european institutions, for the betterment of the measure.
In terms of the theoretical, the experts determine the Intelligence of a Total Artificial (or AGI), as each model is trained, HE is a force, total calculation to be equal to or greater than 10^25 a FLOP. At this time, the companies that make and model of the HE of this kind are OpenAI, Google, Meta, Anthropic, and the Mistral. Just like the model HE developed with the Act of IT, the general will also need to make sure the levels with the minimum of transparency, respect for the rights of the copyright, risk management, and mitigation of the risk of the facility.
Among the guidelines, the code of the new behavior, for example, there is a requirement for the companies, who are working with HIM to deliver the clues at the zvarritësit their browser – software used to retrieve the data that is given to the Artificial Intelligence – and the implementation of measures to prevent cyber attacks aimed at IT-n-it, and even the possibility of a loss of control of a human on it.
Discussion about this post