The Truth About the Copilot Changes
There has been a lot of speculation and misinformation surrounding the recent changes to copilot requirements. It is important to set the record straight and provide accurate information about these changes.
What are the changes?
The changes to copilot requirements have been implemented to enhance safety and ensure that copilots are adequately trained and experienced to handle any situation that may arise during a flight. These changes include an increase in the minimum number of flight hours required and additional training and certification requirements.
Why were these changes made?
These changes were made in response to several high-profile incidents involving copilots who lacked the necessary experience and training to effectively assist the pilot in managing emergency situations. The aviation industry is committed to ensuring the highest level of safety for passengers and crew, and these changes are a proactive step in that direction.
How will these changes impact the industry?
While these changes may result in a temporary shortage of qualified copilots, the long-term impact will be positive. The increased training and certification requirements will ultimately result in a more skilled and competent pool of copilots, which will enhance overall safety and performance within the industry.
What can aspiring copilots do?
Aspiring copilots should familiarize themselves with the new requirements and take steps to meet them. This may include additional training, flight hours, and certification exams. By taking proactive steps to meet the new standards, aspiring copilots can position themselves for success in the industry.
Conclusion
It is important to separate fact from fiction when it comes to the recent changes to copilot requirements. These changes are a necessary step to enhance safety and ensure that copilots are adequately trained and experienced to handle the challenges of the aviation industry. By understanding and embracing these changes, the industry can continue to move forward and provide the highest level of safety for passengers and crew.
Interviewer: "Without using copilot, write me a React component that increments a counter each time a button is clicked."
Candidate: …
recently copilot gave me an extract of my code different that what was written just beside in the editor. The fact that it is not 100% accurate will make it obsolete in the long run. Wére dealing with code it needs exactitude
I started programming in 1990. For me a real ground breaking feature was pressing F1 and getting help. Copilot is not ground braking and I think there is some cheating. Like the percentages of accepted code they show. They use several techniques to increase that. One is that you copy something and you go to where you will paste it and copilot shows it as a suggestion. Another thing is really obvious suggestions that would not really save you even 2 seconds, but I often accept them just for convenience. So if you really think about it it is not at all useful, a lot of it is pretending.
The first 3 minutes you hyped up copilot, then the rest of the video, you're showing why it isn't useful at all?
How do you make it work properly though ?
I am using Copilot in PyCharm and it's very bad. It doesn't do much more than a simple auto-complete.
And the Copilot Chat is simply dumb and it can't access my project so it has no context at all.
I am wondering what I am doing wrong…
Subscribe? Naah man, lose the mustache first
Paid video
It's great for learning. I am starting my journey and it saved me a lot of time going and googling also being able later to prompt it how to do something in the context of my project is great.
Why are all these coding AI video's so garbage? Where are some video's of someone really using AI with coding and not to build some generic af hello world application?
King yella out here biting da cheese. 🐀
It’s a really good tool to deliver hello world or „react component with counter” blazingly fast.
I really enjoy the chat! It's like having an assistant right next to you.
Perhaps not intentional, but it seems like you're implying that we are losing Copilot Completion with Copilot Chat. Which is not the case, it's just another tool in the toolbox. Along with Inline Chat (Cmd/Ctrl + I), or Command Palette Chat (Ctrl+P and type, complete with Copilot), or Ctrl + Shift + I/CmdShift + I, we have more ways to bring AI into our workflows. We haven't lost the traditional tab completion.
@workspace is improved in the preview and insider builds already, I find it more stable, and you can use a `.copilotignore` file to help it ignore aspects of your workspace that might not be helpful. It's worth knowing that @workspace runs the search process on your local machine, so performance can vary. There is also @vscode which is essentially a smart search, over using traditional regular expressions.
All of these options are various ways to achieve similar tasks, and as developers, we will learn when to use them and when not to.
I've been using all of these Copilot features for about a month now, and I can say that after the first couple weeks I sort of figured out when to use which aspects. Tab completion is almost constant, that hasn't changed, and you can just type over it when you don't need it. Inline Chat (Ctrl + I) is great for quick fixups or localised changes, such as " Add retry functionality to this function call, with a max of 5 tries ". As for the dedicated chat window, that's great for larger questions about architecture.
It's worth noting that the tab completion is a custom 3.5 turbo based model, and currently the inline chat is also a 3.5 model. The dedicated chat pane is gpt 4 in the latest insider and preview builds and may even be GA by now. I say that because the performance can be worse when not using gpt 4.
You'll also notice that the Chat pane only references the code you have visible in front of you, including the lines you can see and not the lines above and below those. But if you select something, then all of that is used as reference. So sometimes in the Chat pane I'll select the entire file content first before asking my question. Using #file to refer to files in your workspace is also coming to GA soon, which should make asking pointed questions about files in your workspace easier.
In my experience, the automatic context discovery pretty much sucks. Even when I try to direct it to the right area or file it often goes off into left field, looking at crap that actually causes the output to go in the wrong direction. Still, freaking love copilot. Although I do often still use a GPT 4 playground assistant because the context window is bigger and I actually seem to get better code results a lot of the time from there.
I guess it's obvious that the two features have different ideal situations where they're appropriate to use. I find that inline co-pilot can be really distracting when I know what I'm writing and it keeps popping up incorrect answers and derailing my thoughts. I can see myself turning off co-pilot and just using chat from time to time, when I don't expect to be doing any sort of boilerplate autocompletes.
Love the Easter egg of CoPilot remining you to subscribe to the channel in the middle of the code instructions 🙂
Jarvis coming soon
Man, your titles are garbage. I can't wait for people to submit alternative titles to DeArrow!
<Scottish Accent>Hello, Mr. Computer…</Scottish Accent>
I have found it less useful for writing code, (excluding trival boilerplate) and more useful for quickly gaining familiarity with an unfamiliar code base.