Key Points
Since the company launched its Strawberry AI model family last week, touting so-called reasoning abilities with o1-preview and o1-mini, OpenAI has been sending out warning emails and threats of bans to any user who tries to probe how the model works...
Unlike previous AI models from OpenAI, such as GPT-4o, the company trained o1 specifically to work through a step-by-step problem-solving process before generating an answer..
However, by design, OpenAI hides the raw chain of thought from users, instead presenting a filtered interpretation created by a second AI model...
In a post titled Learning to Reason With LLMs on OpenAI's blog, the company says that hidden chains of thought in AI models offer a unique monitoring opportunity, allowing them to "read the mind" of the model and understand its so-called thought process..
Exposing o1's raw chain of thought would be a bonanza of training data for competitors to train o1-like "reasoning" models upon.. Willison believes it's a loss for community transparency that OpenAI is keeping such a tight lid on the inner workings of o1..
You might be interested in
OpenAI addresses 'chat title' bug and global outage, reveals root cause
25, Mar, 23The company has provided an explanation for the recent global outage and the chat title bug, stating that the issue has been resolved. The company has successfully restored both the ChatGPT service and chat history feature, except for a few hours of chat history that could not be retrieved.
ChatGPT bug leaked users' conversation histories
23, Mar, 23Social media users had shared images of chat histories with AI conversations they said were not theirs.
OpenAI says ChatGPT users can now turn off chat histories
25, Apr, 23The company has allowed users to withhold their ChatGPT conversations from being used in training the artificial intelligence models
OpenAI to enable more customizations for enterprise and individual users
10, Mar, 23The Microsoft-backed company is working with enterprise clients to train its models in particular domains.
OpenAI enables customized GPT bots; offers cheaper, more powerful models
06, Nov, 23OpenAI will enable ChatGPT users to build customized bots to handle specific tasks, and slashed costs on more powerful models for developers, the artificial intelligence lab said at its first developer conference on Monday.
OpenAI’s ChatGPT Plus with GPT-4 access now available for Indian users
17, Mar, 23The company took its official Twitter handle to make the announcement. Indian users will now be able to have early access to a variety of features.
OpenAI announces GPT-4, the new generation of AI language model
15, Mar, 23OpenAI says the new model will produce fewer factually incorrect answers. In fact, the company claims GPT-4 performs better than humans on many standardised tests.
OpenAI Offers a Peek Inside the Guts of ChatGPT
09, Jun, 24Days after former employees said the company was being too reckless with its technology, OpenAI released a research paper on a method for reverse engineering the workings of AI models.
OpenAI rolls out ChatGPT plugins that let chatbot browse internet
24, Mar, 23The company said that it will start providing access to plugins with a small set of users and plan to gradually roll out largerscale access as they learn more