OpenAI is reportedly working on significantly improved AI models with advanced research and planning capabilities, potentially revolutionizing the field.
Currently, Large Language Models (LLMs), the foundation of AI chatbots, can summarize dense texts and instantly compose prose. However, they struggle with common-sense problems and multi-step logic tasks.
The AI race is intensifying, with tech giants investing billions into developing new models. US-based OpenAI became a major player after launching ChatGPT, an AI chatbot capable of answering questions based on prompts and processing images. However, this landscape may change soon.
OpenAI is reportedly developing a new AI model that could dramatically improve AI reasoning capabilities, allowing it to undertake ‘autonomous’ internet research—a capability existing models lack. Here, we delve into this project and its potential disruptive impact.
What Is Project Strawberry?
Nearly six months ago, a secretive OpenAI project named Q* (Q-Star) made headlines. It aimed to develop AI capable of training itself in a novel way, enabling planning, logical reasoning, and capabilities akin to the human brain.
On July 15, Reuters revealed more about the project, stating that OpenAI is working on new reasoning technology under the code name “Strawberry,” believed to be the new name for Project Q*. Internal documentation from the company indicated OpenAI’s intention to use Strawberry for deep research. The project’s exact features and release date remain closely guarded.
However, it is clear that OpenAI aims for Strawberry models to plan ahead, autonomously scour the internet, and conduct deep research. An OpenAI spokesperson said, “We want our AI models to see and understand the world more like we do. Continuous research into new AI capabilities is common in the industry, with a shared belief that these systems will improve in reasoning over time.”
How Will It Differ From Existing AI Models?
Currently, LLMs can summarize dense texts and compose prose instantly but struggle with common-sense problems and multi-step logic tasks. Strawberry models, with enhanced reasoning, could act as catalysts for landmark scientific discoveries and complex problem-solving.
Reasoning involves enabling AI to plan, understand the physical world, and handle multi-step problems. Presently, LLMs cannot plan ahead effectively without external support.
With Strawberry, AI would perform tasks requiring planning and a series of actions over an extended period. In the past, OpenAI CEO Sam Altman discussed why AI models should take time, even days, to think through problems and provide the best answers. Achieving this could revolutionize AI’s ability to accomplish complex tasks autonomously.
How Will Such Powerful AI Models Be Used?
Strawberry models aim to significantly enhance AI capabilities, potentially leading to multiple scientific breakthroughs. In medical research, they could assist in drug discovery, genetic research, and personalized medicine development by analyzing vast datasets.
With enhanced problem-solving abilities, AI could tackle complex mathematical problems, aid in engineering calculations, and participate in theoretical research. It could also handle logical deductions, legal analysis, and strategic planning. In education, the models could offer personalized tutoring, create educational content, and design interactive lessons.
In business, these models could analyze market trends, predict economic changes, assess risks, and assist with investment decisions. Similarly, in creative fields, they could aid in writing, creating art and music, generating videos, and designing video games.
With human-like reasoning skills, these models could transform various industries. However, improved AI capabilities also raise concerns about job impacts, the power required to run these systems, and ethical questions surrounding the reproduction of existing human works.
English