Gestalt Tech Founder Predicts AI Could Develop Self-Awareness Between 2035 and 2045

Gestalt Technology founder Peng Lei stated that the present moment is undoubtedly a golden era for the brain-computer interface industry.

He expressed the view that brain-computer interfaces and artificial intelligence are two sides of the same coin, mutually inspiring and enabling each other in many scenarios, with the potential for future mutual interpretability.

In the current AI field, whether discussing companies about to go public with trillion-dollar valuations or ecosystems like those around SpaceX valued over twelve trillion, the industry's narrative for AI and AGI development is already very clear. However, a key point of focus is that nearly all leading founders and key opinion leaders in AI possess backgrounds in neuroscience.

Neuroscience research and artificial intelligence research have consistently promoted and inspired each other over the past four to five decades. Without neuroscience's study of neurons, today's neural networks would not exist; without research on visual cortex signal processing, computer vision would not be possible. The current hot topic of embodied intelligence research also heavily relies on neuroscientific studies of how the human brain acquires reward and punishment mechanisms in the physical world.

He has a clear judgment on medium to long-term industry trends for the next three to ten years: the industry's path to evolving towards AGI, relying on GPU computing power and the continuous iteration of language models, has been validated. Language models have matured, there are ample application scenarios, and trillion-dollar enterprises have already been born.

In the next ten to fifteen years, carbon-based biological models and silicon-based AI models will each iterate, mutually adapt, and become bidirectionally interpretable. This will ultimately give rise to a universal system akin to a "Rosetta Stone," enabling the mutual understanding and translation of carbon-based and silicon-based model languages.

"When this milestone is reached, it will likely herald a fundamental, breakthrough-level technological advancement. It is anticipated that between 2035 and 2045, artificial intelligence entities at the silicon-based level may develop autonomous consciousness, which will also bring about entirely new social and ethical challenges. On the carbon-based side, the modeling and simulation of multimodal activities in the human brain will enable a highly accurate simulation of its structure, function, and connection mechanisms," he said.