Thanks for confirming in the logs that GitHub Co-Pilot sometimes uses GPT 3.5 Turbo, was pulling my hair out not understanding that sometimes I would have to take my code to ChatGPT to fix the error and then continue again in Co-Pilot 🤘
yeah, it's a little unfortunate they don't always use GPT-4 model... I mean they could at least make it upgrade models when you click regenerate response... either way glad the video helped!
Tabnine is still a great copilot for a free AI tool, it’s not as extensive, but for an intermediate coder that just wants to speed up their workflow, it is a great free option.
Lol… I thought that programmers were totally geeks and nerds, but I think you are an exception. Really nice video, and exactly, VSC is super helpful. But I hate that for now, the Copilot can’t chat with my whole code.
I think it’s ridiculous that this is considered making you “lower end” in development. And “doing it yourself” is considered being “better at programming”. That’s like saying “I walk to work so my commute is more legitimate than your drive.” It’s just intentional self-sabotage in the name of outdated pride or something. It’s like people being proud that they didn’t have to google a historical date. How about be proud of delivering a useful product instead?
It depends on each situation. If you are trying to learn, "doing it yourself" allows you to understand what each part is doing and experience the pitfalls. Let's be honest, many new programmers will happily copy&paste code without batting a second to see if it make sense or why it works. If you are a seasonal programmer, there is not much point to write a similar code yet again for the 8th time. CoPilot would be great, and if the code has potential issue, you can easily spot and update it.
True, the difference is why you are using it, I guess. Are you using it to skip past understanding anything? Or are you using it to aid yourself in learning things deeply? @@fifigaia8451
@@fifigaia8451 Even when learning, I have no shame in letting others show me how it's done. I think this is the difference between someone who learns rapidly and someone who does not. The one who learns rapidly does not attempt to reinvent it. They let others share their inventions so that they can rocket ahead to the more relevant unsolved problems out there. Software engineers seeking to interview at the big companies would be wise NOT to solve the coding questions themselves while learning. First, expose themselves to hundreds of solved problems. The familiarity with their solutions will give you a deep intuition on how to solve the next one. The missing step is "developing understanding". Which is far easier done with the solution in hand. Again, this is why leading companies have a competitive advantage. Because they have found the solution and others will have to reinvent the solution somehow. The advantage of the entire world is in having access to the solution. And what you write about is separate from "shame about using tools". In my interpretation. What you write about is "never actually trying to learn and just cheating". What I am speaking of is focusing on the grand prize of understanding and capability. Not on these false booby prizes of "doing it yourself pride" and "pretending to know copying and pasting".
Hi, i am having 10 years of experience as a Quality analyst in banking, can you please help me to suggest if self-study or doing a master's is a good option for data science
7 месяцев назад+1
Could the conclusion at 1:54 not also be, that because they have access to the copilot, they have the time to go back and fix stuff, that prevously would have remained bad.
Mmm, increased churn could be interpreted very differently: on larger mature products, about 70+% of all online controlled experiments have neutral or negative impact, so it might actually indicate faster implementation/removal by engineering teams, and be a good thing...
But if were a junior developer/data analyst, what insensitive would you see learning what all those lines of code are doing instead of relying on GPT to spit out the code for you? It seems like a slippery slope of forgetting how to learn things and losing the most important quality of a good programmer, patience and hard work.
@@Glotaku You are demanding too much will power my friend :D You talk from the point of a person who have developed the persistency muscle to learn something. Its quite optimistic to think that people of tomorrow will have this muscle developed all that well.
I thought low quality , below average developers use IDEs like VS code, I mean real hard core true seniors use notepad otherwise they are subpar inferior low quality wanna be - but will never be - true devs. Did I kill it ?
Using "AI" to do something you could have done since 1985 with excel with just a few clicks on a button seems pretty useless to be brutally honest. In addition, feeding company data is a big NO if you are working in specific industries such as cyber security where publicly sharing data to the wrong people could have fatal consequences.
@@InnocentiusLacrimosa Of course no one in the cyber security space is doing that. What I am talking about is feeding sensitive data to gpt. Nobody knows where gpt stores its data, just like people uploading sensitive data to the cloud. It is completely insane what people are doing with sensitive data nowadays.
Thanks for confirming in the logs that GitHub Co-Pilot sometimes uses GPT 3.5 Turbo, was pulling my hair out not understanding that sometimes I would have to take my code to ChatGPT to fix the error and then continue again in Co-Pilot 🤘
yeah, it's a little unfortunate they don't always use GPT-4 model... I mean they could at least make it upgrade models when you click regenerate response...
either way glad the video helped!
I really like copilot, it's really good the the broad strokes and helping me see different ways i hadn't thought of.
Tabnine is still a great copilot for a free AI tool, it’s not as extensive, but for an intermediate coder that just wants to speed up their workflow, it is a great free option.
So it's not just me running into limitations with Copilot. Thank you for the video Luke!
Lol… I thought that programmers were totally geeks and nerds, but I think you are an exception. Really nice video, and exactly, VSC is super helpful. But I hate that for now, the Copilot can’t chat with my whole code.
I think it’s ridiculous that this is considered making you “lower end” in development.
And “doing it yourself” is considered being “better at programming”.
That’s like saying “I walk to work so my commute is more legitimate than your drive.”
It’s just intentional self-sabotage in the name of outdated pride or something.
It’s like people being proud that they didn’t have to google a historical date.
How about be proud of delivering a useful product instead?
It depends on each situation. If you are trying to learn, "doing it yourself" allows you to understand what each part is doing and experience the pitfalls. Let's be honest, many new programmers will happily copy&paste code without batting a second to see if it make sense or why it works. If you are a seasonal programmer, there is not much point to write a similar code yet again for the 8th time. CoPilot would be great, and if the code has potential issue, you can easily spot and update it.
True, the difference is why you are using it, I guess.
Are you using it to skip past understanding anything?
Or are you using it to aid yourself in learning things deeply?
@@fifigaia8451
@@fifigaia8451 Even when learning, I have no shame in letting others show me how it's done.
I think this is the difference between someone who learns rapidly and someone who does not.
The one who learns rapidly does not attempt to reinvent it. They let others share their inventions so that they can rocket ahead to the more relevant unsolved problems out there.
Software engineers seeking to interview at the big companies would be wise NOT to solve the coding questions themselves while learning.
First, expose themselves to hundreds of solved problems.
The familiarity with their solutions will give you a deep intuition on how to solve the next one.
The missing step is "developing understanding".
Which is far easier done with the solution in hand.
Again, this is why leading companies have a competitive advantage. Because they have found the solution and others will have to reinvent the solution somehow.
The advantage of the entire world is in having access to the solution.
And what you write about is separate from "shame about using tools". In my interpretation. What you write about is "never actually trying to learn and just cheating".
What I am speaking of is focusing on the grand prize of understanding and capability.
Not on these false booby prizes of "doing it yourself pride" and "pretending to know copying and pasting".
Which software do you use to record your screen. Its very cool the way it zoom to the point. is it camtasia?
Hi Luke! Love watching your video. Thank you for sharing! Keep up!
Thanks for making me feel bad about my lacking programming skills
I am using Copilot becasue I belive not using it in 2-4 years will make you uncompetitive in general IT market (as developer)
This is crazy because now we're going to have more developers who are better, faster, but yet understand less. What is this world
Hi, i am having 10 years of experience as a Quality analyst in banking, can you please help me to suggest if self-study or doing a master's is a good option for data science
Could the conclusion at 1:54 not also be, that because they have access to the copilot, they have the time to go back and fix stuff, that prevously would have remained bad.
So you uploaded a working code of your own to fix the Copilots errors? Genius LOL
Stackoverflow assistant
Mmm, increased churn could be interpreted very differently: on larger mature products, about 70+% of all online controlled experiments have neutral or negative impact, so it might actually indicate faster implementation/removal by engineering teams, and be a good thing...
Wow this is very good
Thanks Luke
I’ll be your copilot anytime
😘
Um… 🙋🏼♀️
Code replacement is incorrect - lines are duplicated, imports are duplicated. The context of the entire project is not available. So so.
amazon code whisperer is free. and works similar. just saying coz you said you don't like to pay for tools
Awsome🎉
That’s a pretty hefty P value!
👀
But if were a junior developer/data analyst, what insensitive would you see learning what all those lines of code are doing instead of relying on GPT to spit out the code for you? It seems like a slippery slope of forgetting how to learn things and losing the most important quality of a good programmer, patience and hard work.
You can easily use it to accelerate your learning in all depends how you use the tool
Who needs learning when we have things just done for us 😂😂😪
@@Glotaku You are demanding too much will power my friend :D You talk from the point of a person who have developed the persistency muscle to learn something. Its quite optimistic to think that people of tomorrow will have this muscle developed all that well.
I thought low quality , below average developers use IDEs like VS code, I mean real hard core true seniors use notepad otherwise they are subpar inferior low quality wanna be - but will never be - true devs. Did I kill it ?
Codium and Tabnine AI are free
So, this is only for below average coders? 😢
No.
Using "AI" to do something you could have done since 1985 with excel with just a few clicks on a button seems pretty useless to be brutally honest. In addition, feeding company data is a big NO if you are working in specific industries such as cyber security where publicly sharing data to the wrong people could have fatal consequences.
Muah. If you are working on cyber security and decide to work on public facing website to show your data, then you have a YOU problem.
@@InnocentiusLacrimosa Of course no one in the cyber security space is doing that. What I am talking about is feeding sensitive data to gpt. Nobody knows where gpt stores its data, just like people uploading sensitive data to the cloud. It is completely insane what people are doing with sensitive data nowadays.
7 minutes out of the 10 were useless fluff, waste of time
Seems like a paid review