MIT gives (AI) the power to Reason like Humans

MIT researchers have advanced a new method to enhance artificial intelligence (AI) system’s ability to perform complex reasoning tasks in coding, strategic planning, and robotics. Large language models (LLMs) like ChatGPT and Claude 3 Opus process and generate text based on human input, known as “prompts”.

The researchers noted that while these technologies have improved in the last 18 months, they still struggle with understanding context and performing well in reasoning tasks. However, MIT scientists claim to have solved this problem by creating a “TREASURE TROVE” OF NATURAL LANGUAGE “ABSTRACTION” that could lead to more POWERFUL AI MODELS.

Abstractions simplify complex subjects, omitting non-essential details. This can help chatbots REASON, LEARN, PERCIEVE, and represent KNOWLEDGE LIKE HUMANS. However, scientists argue that LLMs struggle to abstract information in a human-like way.

The scientists organized natural language abstractions into three libraries to achieve greater contextual awareness and provide more human-like responses. They detailed their findings in papers published on the arXiv preprint server on October 30, 2023, December 13, 2023, and February 28.

The LILO library synthesizes, compresses, and documents computer code. The Ada library covers AI sequential decision-making. The LGA framework helps robots understand environments and plan movements using language for context in complex tasks.

The findings were presented on May 11 at the International Conference on Learning Representations in Vienna, Austria. According to Robert Hawkins, an assistant professor of psychology at the University of Wisconsin-Madison, library learning represents an exciting frontier in artificial intelligence, offering a path to discovering and reasoning over compositional abstractions. Hawkins, who was not involved in the research, noted that previous attempts were too computationally expensive to use at scale.

The scientists mentioned that three library frameworks employ neurosymbolic methods, an AI architecture that combines neural networks with classical logical approaches. The model improved accuracy by 59% for putting chilled wine in a kitchen cupboard and by 89% for building a bed, compared to carrying out the tasks without Ada’s influence.

The researchers aim to find more domestic uses for Ada soon. The LGA framework helps robots understand their environments better, allowing them to perform tasks more effectively based on natural language commands like “bring me my hat.”

The researchers showed that LGA is effective by using Spot, Boston Dynamics’ robot, to fetch fruits and recycle drinks, demonstrating the robot’s ability to scan the environment and create plans in chaotic settings.

The researchers believe neurosymbolic frameworks like LILO, Ada, and LGA will lead to “MORE HUMAN LIKE” AI models with better problem-solving skills and improved environmental navigation.

Revelations 13:15 And he had power to give “life” unto the”image” of the beast, that the image of the beast should both “speak,” and cause that as many as would not worship the image of the beast should be killed.

Revelations 13:16 And he causeth all, both small and great, rich and poor, free and bond, to receive a mark in their right hand, or in their foreheads:

Revelations 13:17 And that no man might buy or sell, save he that had the mark, or the name of the beast, or the number of his name.

Revelations 13:18 Here is wisdom. Let him that hath understanding count the number of the beast: for it is the number of a man; and his number is Six hundred threescore and six.

Click here to read more articles transformedbythetruth.com

Share this: