OpenAI invites startups to tailor AI evaluations and models for their industries
OpenAI has launched a new initiative to work with startups to create new ways of evaluating the performance of AI, and to develop bespoke models for specific industries.
The OpenAI Pioneers Program invites businesses to apply to work with the company on customising AI solutions, and benchmarks, that are tailored to different sectors and real world applications.
In a blog post today, OpenAI said it wants to ensure businesses have the means to accurately assess the performance of AI as it becomes more entrenched within different industries.
The company is proposing the creation of “domain-specific evals” that would measure the effectiveness of an AI model in a particular context. For example, a law firm might want to measure the legal accuracy of an AI model, while a manufacturing firm might want to assess its material knowledge.
OpenAI wants to work with companies to create a “unified source of truth” across industries like legal, financial, insurance, healthcare, accounting etc. The idea is to create reliable metrics that can be used to enhance the performance of models.
The other side of this initiative is fine-tuning AI models to make them more effective in the top three use cases identified by the startups OpenAI works with. The fine-tuning will be guided by OpenAI’s team and will help the model become more aligned with the needs of that industry.
Businesses can apply to take part by outlining their industry-specific use case and the challenges they face. They’ll also need to provide examples of how tailored evaluations could benefit their sector.
OpenAI is currently preparing to unveil new models in its GPT series and is expected to release GPT-5 in the coming months.
