The second day of OpenAI's 12 Days of OpenAI focused on less spectacular and more entrepreneurial interests compared to the general launch of the OpenAI o1 model on ChatGPT on the first day.
Instead, OpenAI announced plans to launch Reinforcement Fine-Tuning (RFT), a way to customize their AI models for developers who want to adapt OpenAI algorithms for specific types of tasks, especially more complex ones. This release marks a clear shift towards enterprise applications compared to the consumer-focused day one updates. You can think of RFT as a method of improving the performance of AI models through their reasoning of responses. Using a data set and a developer's evaluation rubric allows the OpenAI platform to train its specialized AI without much costly reinforcement from subsequent experiences.
RFT could be a boon for artificial intelligence tools used in law and science. OpenAI highlighted in its livestream the CoCounsel AI assistant built with RFT by Thompson Reuters and how RFT helps researchers studying rare genetic diseases at Berkeley Lab. However, commercial partnerships won't make much difference in the short term for average users of ChatGPT or other OpenAI products.
Today we announced boost tuning, which makes it really easy to build domain-specific expert models with very little training data. Live streaming now: Program starting now, will be released publicly in the first quarter.December 6, 2024
Company or consumer
If you're more interested in the consumer side, don't give up just yet. While the business tilt is in contrast to the first day, it's easy to imagine OpenAI wanting to have as wide a range of news as possible over the 12 days. There will almost certainly be much more consumer news to come. Maybe alternating days or some other pattern.
Still, at least OpenAI's final joke was a little funnier than yesterday. The AI described how autonomous vehicles are popular in San Francisco, and Santa is interested in making an autonomous sleigh as part of the trend. The problem is that it keeps hitting the trees. What is the problem? He did not fine-tune his models. Perhaps the image ChatGPT created for TechRadar editor-at-large Lance Ulanoff sells the humor better.
You may also like…
