AI chatbot for companies: OpenAI releases ChatGPT Enterprise

In the future, companies will be able to train OpenAI’s AI language model GPT-3.5 Turbo with their own data using an AI chatbot. This makes it possible to better adapt the system to the respective needs, for example, to improve performance in specific use cases.

AI chatbot
image: OpenAI

The message has been updated, the update is at the end of the news.

So what should be improved is the quality of the answers in chat tools like ChatGPT. For example, fine-tuning can be used to specify that answers should always be of a certain length – i.e. particularly brief – or in a certain language. The reliability of the results can also be improved, which according to OpenAI is helpful when programming, for example.

In addition, a specific tone can be set based on your own data. So answers appear in a style that is typical of the company. With fine-tuning, it’s also possible to get targeted answers with fewer prompts. Employees should therefore be able to achieve the desired results with simpler commands.

AI chatbot: The fine-tuning approach is still limited to older GPT-3.5 Turbo

The fine-tuning solution is currently still running with the GPT-3.5 Turbo language model, and an offer with GPT-4 is to follow in the fall. According to OpenAI, a GPT-3.5 Turbo customized with your own data should already achieve better results than GPT-4.

OpenAI will not use the data transmitted in the fine-tuning process to train its own models. This is what OpenAI does when customers use the company’s APIs. The price distinguishes between training and usage costs: Training costs 0.8 dollar cents to handle 1,000 tokens, and usage level requires 1.2 dollar cents for input per 1,000 tokens, 1 for output, and 6 cents per 1,000 tokens.

The data is checked to ensure compliance with OpenAI’s security standards. This is done using the moderation API and a GPT-4-based moderation system. The company wants to sort out potentially harmful content in advance.

Individually adapted AI models as a vision of the future

One of the future versions of the industry is that companies and organizations train the models with their own data in order to improve the quality of the expenditure and to adapt it to the respective needs. Microsoft also wants to use the business data of the customers as a basis for the Office 365 Copilot, which is based on the GPT-4 language model.

In general, various approaches are possible to customize the models. The spectrum ranges from settings for the entries to fine-tuning existing models to training your own models.

OpenAI AI chatbot expands the portfolio for companies. ChatGPT is now also available in an enterprise version, which allows unlimited access to the GPT-4 language model and places special emphasis on security and privacy. The blog post states that OpenAI does not use any of the inputs to train the AI ​​models. Encryption processes and admin tools are also integrated, which are intended to make member administration and data handling easier.

The Enterprise version also enables the entry of longer prompt commands and documents, the context has been increased to 32,000 tokens and thus quadrupled. OpenAI does not name prices for ChatGPT Enterprise. Companies will receive an offer and access on request.

Since the chart in November, 80 percent of the Fortune 500 companies have already done test runs with ChatGPT, and OpenAI reports. Security concerns have been an issue that has often been discussed in recent months, with numerous corporations – including Apple, Google and Samsung – restricting access for employees.

Also Read,

ChatGPT Equipped Robot Dog Unitree Go2 Came out

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top