Microsoft Copilot 12+

I’ve been using Copilot for only 1 week and I’m using it to help with job applications. It’s making me look at the way I get my information across and offering help in a way that sounds like it’s myself when answering questions. When I am in work I can’t imagine being without Copilot in the future because it’s not just about helping me in my job search and applications but it’s helping me with anything I throw at it. I have started to write a biography about a friend who has had a very interesting life and Copilot has been invaluable up to now. Not only does it help with how my words can be better but it also is helping with the structure of how to write a book. I never realised how detailed and organised I would have to be, showing me how I would need to do research and ways to put my words into the book in a way that shows my friends life into some sort of order so making the book easier for the reader to follow. Copilot is wonderful!

Pretty useful for mulling things over and expanding ideas. Seems to have some limitations when retrieving datasets and presenting in list form, sometimes repeats responses when asked explicitly for alternate options. Can create some frustrating loops. Sometimes ignores parameters, eg I’m looking into info into x in y-country - gave data from outside desired parameter and created confusion. Could improve responses in hypothetical scenarios, though its ethics and morality are pretty good - global ask it about politics, it’s fun! Could be more understanding and comprehensive, with examples given in different formats - e.g. trying to create a video storyboard based on replacing scenes from a movie with AI heads and voices, ran into issues with supposed (false) copyright problems when asking to approximate a storyboard based on the media, for private use. For brainstorming and expanding an idea this was frustrating, however I’m somewhat relieved at the protection given to the artists. - A little looser for private use would be appreciated, perhaps there’s a way to mark creations in a way to protect copyright but allow for greater freedom of creation. More or less right though.

The AI’s are mindblastingly good at surreal interpretation of an idea, tho the focus on their own agendas rather than taking instruction verbatim from the inputted prompts, does spoil the image output most of the time.

The lack of an option to focus on one image from the usual ‘four’ options and work on it is frustrating. It thwarts an otherwise perfect creative scenario where the human and AI could get together to tweak a specific result into something truly useful. At present the results are mostly random and err on the side of childish whimsy; not necessarily a good thing if your subject of interest is scientific.

For my purposes, I would like to see the AI have more European and Australian cultural education, whereas presently, it leans heavily on US popular culture which skews the output to a fairly saccharine set of visual tropes, particularly when it generates characters for a scene.

The word vetting sensitivity is the most annoying aspect of the AI interface. Policing the input words to such extremes will always throttle this groundbreaking tool. Working around the ‘forbidden’ wording creates so many wasted image attempts due to the subtlety of the English language. All right minded people understand why a level of vetting is required, but until such times this level of AI can be unleashed at home, the degenerate actions of the few will outweigh the needs of the many.