Is it just me or does the Edge integration fail at almost every task I'd want it to perform?
More than once I've had a long document open and I've been like, oh Edge has GPT now maybe it can give me a summary. It will read the page, produce something vague, I'll ask a follow up question and it will completely lose track of what we're doing and like search Bing for generic information not at all related to the webpage I'm on. It also seems inherently unable to say, "oh the page doesn't include that information" rather it just makes things up so I always end up just having to read the document myself.
It really feels like an LLM being slapped on for the sake of having one rather than trying to solve a real need. Which is super frustrating because I have a real need and this technology in theory could solve it.
> March 16, 2023, Microsoft announced Microsoft 365 Copilot.[32][33] This tool, designed for Microsoft 365 applications and services, Edge, Microsoft Bing and Windows, leverages the advanced capabilities of OpenAI's GPT-4 large language models (LLMs). It also incorporates Microsoft Graph to transform user text input into content across various Microsoft 365 applications, including Word, Excel, PowerPoint, Outlook, and Teams.
> GitHub Copilot (not to be confused with "Copilot X", the name of GitHub's "vision" for next-gen Copilot features[1]) is a cloud-based artificial intelligence tool developed by GitHub (owned by Microsoft) and OpenAI to assist users of Visual Studio Code, Visual Studio, Neovim, and JetBrains integrated development environments (IDEs) by autocompleting code.
The five-message limit isn't a result of the context window, but an arbitrary constraint applied after early results of the Bing chatbot were, uh... troubling[0]?
I believe that the limit is five for people who are not logged in, and 30 for people who are logged in, hence the discrepancy between what you guys are saying.
More than once I've had a long document open and I've been like, oh Edge has GPT now maybe it can give me a summary. It will read the page, produce something vague, I'll ask a follow up question and it will completely lose track of what we're doing and like search Bing for generic information not at all related to the webpage I'm on. It also seems inherently unable to say, "oh the page doesn't include that information" rather it just makes things up so I always end up just having to read the document myself.
It really feels like an LLM being slapped on for the sake of having one rather than trying to solve a real need. Which is super frustrating because I have a real need and this technology in theory could solve it.