OpenAI says its AI is getting better at doing things humans do, like shopping and meal planning
People already use OpenAI’s ChatGPT for things like helping with homework and creating presentations. Now, the tech giant claims its AI models are getting better at everyday tasks largely left up to humans — like shopping and creating a budget. OpenAI on Monday announced a new internal tool for assessing how AI handles tasks like … The post OpenAI says its AI is getting better at doing things…
OpenAI, the technology powerhouse behind ChatGPT, has announced that its AI models are becoming increasingly proficient at handling everyday tasks once left to human hands, such as shopping and budgeting. The company unveiled a new internal tool designed to assess how well its AI handles such tasks, alongside research on the performance of its own models.
OpenAI's GPT-6 Astra model, for instance, achieved a remarkable 93.1% in locating the right kindergarten after a family move, a 98% score in managing community volunteers preparing for a flood, and a 90.4% score in creating a meal plan for a parent recently diagnosed with diabetes.
This development comes as OpenAI faces intense competition from tech giants like Meta, Anthropic, and Google, all of which are striving to develop AI agents capable of performing real-world tasks. The race to create such personal AI agents is heating up, as companies grapple with concerns about AI safety and the need to monetize their substantial investments in the technology.
However, before measuring the AI's proficiency, OpenAI had to create a way to evaluate whether ChatGPT excels at everyday errands like creating a grocery list and filling out online forms.
Standardized evaluations play a crucial role in AI development, allowing labs to train and optimize their models for specific improvements. OpenAI previously released a benchmark called GDPVal to gauge AI performance across various work-related skills. The new benchmark aims to propel ChatGPT from a content generation tool to a more proactive assistant that can handle tasks for users.
Measuring AI performance in real-world scenarios, however, poses unique challenges. The tasks required for shopping or selecting a kindergarten involve broad and subjective criteria, making it difficult to establish definitive tests. To overcome this, OpenAI consulted surveys and direct feedback from ChatGPT users to identify potential tasks, ensuring that the tasks proposed had a solid understanding of the desired outcome and the level of expertise needed.
Written by urgent.news from Egypt Independent's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
