After launching ChatGPT 2.0 in July this year, the world’s premier artificial intelligence research organization, Open AI is likely to release its most powerful AI model between September and November. The new version will be integrated into ChatGPT-5 into its existing chatbot and virtual assistant that the company launched in late 2022.
The project that was earlier known as Project Q* (Q-star) is now codenamed Project Strawberry. The new chatbot is expected to feature autonomous Internet research and improve AI reasoning capabilities. The project has been proposed as OpenAI’s ambition to create Artificial General Intelligence i.e. AI with capabilities of the human brain.
Taking to his X (formerly Twitter) account last month, Open AI CEO Sam Altman posted an image of strawberries growing in two pots. Many speculated this post was a confirmation that OpenAI is working on the new and powerful large language model (LLM).
i love summer in the garden pic.twitter.com/Ter5Z5nFMc
— Sam Altman (@sama) August 7, 2024
Reportedly, OpenAI demonstrated a primary version of the new model to national security officials. This was to make a statement of its commitment to transparency amidst the growing security concerns among people and national governments.
Fitting in with math
As per reports published in California by a tech industry publication last month, Project Strawberry would be better at math and programming than any existing chatbot. The publication involved quotes from two people involved in the project. This somewhat confirms that OpenAI is working on a project that’s bigger than the ones it has launched in the past.
The project will be integrated with ChatGPT which makes it the most powerful AI chatbot. The existing ChatGPT has sometimes struggled with math as experts have pointed out that the bot lacks adequate mathematical information in the training data.
The same report mentioned that a demo by Project Strawberry staff had shown that the new AI model is capable of advanced levels of thinking. The new settings allow it to solve puzzles including one of the most difficult word puzzles by The New York Times ‘Connections’.
Better training plans
The information published by the Californian tech firm also highlighted that Project Strawberry aims to raise more capital, which OpenAI needs for its next-frontier application, codenamed Orion. Generating high-quality data for Orion is believed to be one of Project Strawberry’s key applications.
Generating data is particularly significant now as most of the training data on the Internet has already been consumed by AI, and no additional source of information is available for free to train AI models.
Thus, Project Orion, which is being designed to perform better than GPT-4 may use both Project Strawberry and high-quality synthetic data that would reduce errors compared to the older versions of AI bots.
No room for error
In his briefings, Altman has said that OpenAI has been trying out different ways to train AI models to generate large amounts of synthetic data which the generative AI models create on real-world data samples. After it is trained on algorithms, correlations, and statistical properties of the sample data, the AI model can produce statistically identical data.
As of now the large data set that AI models rely on could be prone to biases and errors. This could also have incomplete and inaccurate information, which is likely to be balanced by high-quality synthetic data produced by Project Strawberry. Synthetic data is also believed to fill the gaps in real-world data sets and provide balanced, wholesome, and inclusive training.
Experts believe that the use of synthetic data can help make future AI models more neutral and fair, and reduce the clutter of irrelevant information, improving both the efficiency of training and the accuracy of the model.
As per the available information so far, Project Strawberry will have improved reasoning, logic and the ability to plan and carry out research. It could also allow the model to autonomously conduct experiments, analyse data, and curate new hypotheses. Experts note that this could lead to a huge scientific breakthrough, which could also help discover new drugs. For learners, it could offer personalised education by generating educational and informational content and interactive lessons for students.
It is yet to see how much OpenAI's Project Strawberry stands on the grounds of security and safety of the users.
Also Read
