ChatGPT, Copilot, and other generative AI applications Generative Artificial Intelligence (GenAI) applications are trained with huge masses of text. Apps can be prompted and respond to user-entered text input. In other words, it is as if you can have a dialogue with applications on a topic that the user wants. They generate responses based on the texts they learn based on a statistical model. Simply put, apps can be compared to predictive text input on your phone. So they know nothing.
For example, the free ChatGPT does not have access to the internet and cannot be used as a data source. It also cannot judge the reliability of the text it produces and can claim to be false or self-invented things (hallucination). OpenAI has not disclosed any information about the data used to train ChatGPT. It is also worth remembering that AI models are biased.
Combining the language model with the internet and a search engine enables the language model to answer based on internet searches. E.g. ChatGPT+plugins, Bing Chat, Google Bard, Perplexity.ai, You.com and Scite.ai.
Policy:
Even in this case, the language model can hallucinate or misinterpret sources. So be source critical. Most tools require you to sign in.
Orientation to the topic
Microsoft Copilot (formerly Bing Chat) is a conversational AI application. The language model of the Copilot is GPT-4 and the image model is DALL-E 3. The app is available in the sidebar of the Edge browser or at https://copilot.microsoft.com/.
When you sign in with your organizational account, you gain access to the Enterprise data protection in Copilot version ensuring better data privacy. Once you have signed in with your university credentials:
When using AI applications, always consider how they handle the data you enter. As a rule, all material entered into AI services is stored outside the EU/EEA and cannot be removed from it. For example, personal data, business secrets, confidential information or confidential information may not be entered into the services.
Learn how to limit the use of your feeds to train AI in ChatGPT.
When using AI based information retrieval tools, the terms of use of e-resources must be taken into account.