microsoft recently presented a new The version of its software with the addition of artificial intelligence (AI) capable of performing a variety of tasks. For example, Copilot can summarize oral conversations on online meetings in Teams, present the arguments for and against a certain point of view based on oral discussions and respond to part of your emails. He is even able to write a computer code.
This rapidly developing technology brings us closer to the future, where AI facilitates our life, eliminating boring and repeating tasks. However, despite impressive opportunities, we must be careful when using such large language models (LLM), including ChatGPT. They require certain skills for effective and safe use.
Why excessive dependence on AI can be a problem
Despite the apparent intellectuality, we cannot blindly trust LLM in matters of accuracy and reliability. It is important to carefully check their conclusions, making sure that our initial requests are reflected in the answers provided.
This becomes especially critical in situations where we use LLM to fill in gaps in our own knowledge. Here, our lack of knowledge can lead to a situation where we simply cannot determine whether the conclusion is correct.
The use of AI to participate in meetings and the preparation of their protocols is obvious risks in terms of reliability. Meetings based on transcriptions are still generated in the same way as another text from LLM. They are still based on the language templates and the probabilities of what was said, so they require verification before they can rely on them.
There is also a problem of verification if we use AI to generate computer code. Testing code with test data is the only reliable method for confirming its functionality. But this does not guarantee that his behavior corresponds to real expectations.
LLM, such as ChatGPT and Copilot, are powerful tools from which we can all benefit. But we must be careful not to trust their results blindly.
We are at the initial stage of the Great Revolution based on this technology. AI has endless possibilities, but it must be formed, checked and verified. And so far only a person is capable of this.