Co-founder of Ethereum, Vitalik Buterin, has criticized the current strategy for artificial intelligence development, stating that excessive autonomy in agent-based AI models is not the optimal direction for the industry. In his view, the future of artificial intelligence lies in creating systems that allow humans to have a more active influence on the processes and outcomes of AI work.
This is reported by Finway
Buterin on Human-AI Interaction
Vitalik Buterin emphasized that modern artificial intelligence technologies are overly focused on creating highly autonomous agents; however, a more effective and safer approach is to integrate mechanisms for human input. He believes this leads to better results and ensures a higher level of safety in interactions with AI.
“I am frustrated that many developments in the AI field strive to be as ‘agent-like’ as possible, while in reality, creating more pathways for human input both now and in the future yields better results and is safer,” Buterin stated.
Buterin also paid special attention to AI models with open weights and advanced editing capabilities, which, in his opinion, open new prospects for users and provide greater flexibility in configuring the system’s operation.
Prospects for Brain-Computer Interface Development
In the medium term, the Ethereum co-founder sees significant potential in the development of brain-computer interfaces (BCI). Such technologies will allow for real-time reading of a person’s emotional response to specific elements of generated content, enabling prompt adjustments to the results according to user needs.
The idea of harmonious interaction between humans and artificial intelligence was also supported by Tim, founder and CEO of OpenServAI. He emphasized the importance of new architectures capable of transforming user intentions into valuable actions by the agent:
“New architectures that convert user intentions into valuable actions are the only possible path for development, in my opinion. This is what we are working on at OpenServAI, applying it directly to agents for coordinating complex tasks.”
Experts believe that further progress in the development of open, flexible, and interactive AI models will enhance their safety and efficiency, as well as provide broader opportunities for users across various fields.