Meta Announces Research To Create Human-Level AI

Facebook parent Meta announced that they are starting a long-term research project to build the next generation of AI that can learn and process speech and text, just like the human brain. Meta described an effort to create human-scale AI.

Meta is partnering with Neurospin, a neuroimaging company, to create images of the human brain, and Inria, a software company, to study how the human brain processes speech and text and then use it with AI language models. Compares.

Neurospin is a research center specifically focused on brain imaging. Researchers include physicists, mathematicians, neuroscientists and doctors who work together to create tools to learn about the human brain in a variety of ways.

Neurospin explains what it does,

“Focused on neuroimaging, research ranges from technical and methodological developments (data acquisition and processing) to preclinical and clinical neuroscience, including cognitive neuroscience.”

Meta published:

“Today, we are announcing a long-term AI research initiative to better understand how the human brain processes speech and text. In collaboration with the neuroimaging center Neurospin (CEA) and Inria, we are comparing that AI language models and how the brain reacts to the same spoken or written sentences.

We will use the insights from this work to guide the development of AI that efficiently processes speech and text as people.”

The problem with AI language models is that they require a lot of examples to learn. The human brain only needs a few examples to learn.

Current research into AI language models such as the brain explores:

“Language models that most closely resemble brain activity are those that best predict the next word from context (e.g. once… time).

While the brain predicts words and ideas far ahead of time, most language models are only trained to predict the next word. Unlocking this long-range prediction capability could help improve modern AI language models.”

The announcement cited current research into AI modeling on human brain activity that used MRI and other imaging tools to visualize human brain activity when humans were completing various language-related tasks.

The cited research paper is from 2021 and is titled, Language processing in the brain and deep neural networks: computational convergence and its limitations (PDF).

The summary of the findings is discussed in the opening paragraph of the research paper:

“The results suggest that (1) the position of the layer in the network and (2) the ability of the network to accurately predict words from context are the main factors responsible for the emergence of brain-like representations in artificial neural networks.

Together, these results show how perceptual, interpretive and creative representations are accurately manifested in each cortical region and contribute to uncovering the governing principles of language processing in the brain and algorithms.

The importance of the above research is to show that research into how the brain processes data can provide insight into algorithms creating similar processes.

META research teams are using thousands of scans of human brain activity to see which areas of the brain were activated during tasks.

The research was said to show “the computational organization of the human brain,” which yielded useful insights for META’s goal of developing “human-scale AI.”

The benefits aren’t limited to generating human-scale AI, the research also helps neuroscientists better understand the human brain.


Read the official meta announcement

Building an AI that processes language as people

Read a more in-depth description of Meta’s human-level AI research

Studying the brain to build an AI that processes language like people

Source link

Leave a Comment