Edgar Cervantes / Android Authority
TL;DR
- OpenAI is aiming to launch a new AI as part of a chatbot — possibly ChatGPT — this fall.
- Codenamed Strawberry, this AI has advanced math-solving and programming skills you won’t find in other LLMs.
- A larger version of Strawberry is being used to improve the successor to GPT-4, codenamed Orion.
Although OpenAI has long been in the driver’s seat of the AI race, competitors have caught up to, and in some cases surpassed, GPT-4, leaving all eyes on the company’s next-generation large language model (LLM). The successor of GPT-4 is reportedly getting beefed up with the help of a new AI and that new AI could get a spinoff that eventually becomes part of ChatGPT later this year.
A new report from The Information claims that OpenAI plans to launch a new AI, codename Strawberry, as part of a chatbot this fall. The outlet suspects that Strawberry could possibly become a part of ChatGPT.
What’s special about Strawberry?
If you follow OpenAI news closely, you may have heard of Strawberry before. In the past, this AI was known as Q* (pronounced Q Star). This is an AI that was started by OpenAI’s former chief scientist, Ilya Sutskever, and later improved on by researchers Jakub Pachocki and Szymon Sidor after Sutskever left the company. This is the same AI that led to a technical breakthrough and raised subsequent safety concerns that caused turmoil in the firm back in 2023.
Its defining feature is advanced mathematical reasoning, capable of solving math problems it has never seen before. This is something that today’s chatbots tend to struggle with. It also has advanced programming and other skills that allow it to answer questions on more subjective topics, like marketing strategies for example.
The version of Strawberry destined to become part of a chatbot is a smaller, simplified version of the AI. OpenAI’s goal is to have it maintain the same level of performance as the larger model, while being cheaper and easier to operate. According to the publication, it’s unclear if this version of Strawberry will provide any performance boosts to ChatGPT or GPT-4 this year.
The role of the bigger Strawberry
OpenAI has found that it can use the large version of Strawberry to generate higher-quality train data to feed into its LLMs. This training data generated from AI is known as “synthetic” data. With this synthetic data, it reduces the company’s need to use text and images from the internet to train its AI models. As an additional benefit, this higher-quality data can also help reduce the errors, known as hallucinations, AI is known for.
It’s expected that Strawberry could help OpenAI’s problem with obtaining enough real world data for its LLMs to consume. As a result, the company is reportedly using the large version of Strawberry to train the successor of GPT-4, codenamed Orion. It’s also believed that the AI could be used to also improve the firm’s agents.
OpenAI is confident enough in Strawberry that it has shown off the technology to American national security officials. The outlet says that the demo could be related to OpenAI’s initiative to be more transparent to policymakers who feel threatened by the power of AI. This is especially important for OpenAI to advertise after multiple safety leaders recently left the company.
It’s unknown exactly when the smaller version of Strawberry will launch this fall. We’ll have to wait to find out the date and if this add-on will give OpenAI a step up on its rivals.
Got a tip? Talk to us! Email our staff at [email protected]. You can stay anonymous or get credit for the info, it's your choice.
English (US) ·