Free AI web copilot to create summaries, insights and extended knowledge, download it at here

Abstract

e that it doesn’t produce a new uncopyrighted piece of code. That’s just a derivative work that contains several pieces of copyrighted material.</p><p id="f59a">Another entry in the FAQ that addresses this was one that asked about the ownership of the code produced by the AI. Who owns it? Well, they say that you own it entirely. That is another obvious problem. Think about this for a second: If GitHub Copilot is taking derivative work from copyrighted material and giving it to you, do you really own anything?</p><h1 id="3212">2. It Is Another Crutch You Will Have To Pay for at Some Point</h1><p id="4fda">This tool will not make people better programmers, although it does not actually claim to do that. I mean, it is really no different from copying and pasting from Stack Overflow. This is just the hyper-integrated version of it. Not that it is a bad thing, but Stack Overflow is a crutch too. The difference is that you have to actually put some effort into finding the code you are looking for with Stack Overflow. It will kind of guide you to the actual solution that you want.</p><p id="b04e">As for the cost of GitHub Copilot, I had assumed at first it was free, but after reading the frequently asked questions, I’m not so sure anymore. They address this in a couple of places in the frequently asked questions. The first is where they say that the technical preview is free for some people. Will there be a paid version? They say that they plan to build a commercial version of the Copilot in the future. That means that the core reason for the preview is to learn how people use Copilot and what it will take to operate at scale.</p><p id="7d68">Of course, you can form your own opinion here, but it is 100% clear that they will have a paid version. It is not clear, however, if they will have a free version. If I were making a prediction, I’d say that there will be a free version with limitations.</p><p id="c3d1">In short, it seems like GitHub might be trying to build an entirely new revenue stream around tens of millions of developers who have contributed to open source. I do not want to judge, but you can make of that what you will.</p><h1 id="fd9a">3. Your Private Code Will Be Sent to GitHub for Analysis and Storage</h1><p id="305e">OK, this depends a lot on the nature of your project. I had assumed two things were true from the outset here:</p><ul><li>Your private code sent for analysis is not

Options

actually stored</li><li>The private code sent is not actually used to generate any future suggestions for other people.</li></ul><p id="9006">After reading their frequently asked questions, I found that only one of those is true. This is addressed in a question asking whether the transmitted data is secure. The response given was that all data is transmitted and stored securely. That means they will store it, but then they go on to say that when humans read it, it is specifically to improve the model or detect abuse.</p><p id="4ac8">Does that mean that your potentially private and sensitive code is both stored and/or read by another person? Keep in mind, depending on your project, this could range anywhere from a non-issue (working on open source) to a deal-breaker (working on something super-sensitive). What if you are working on client code for which you signed an NDA, and GitHub Copilot is just ingesting that? The good thing is that they are not sharing that private code with other users — or at least that is what they say.</p><h1 id="ac2f">4. Code Quality</h1><p id="c662">This is not me being all speculative. GitHub themselves have actually tried addressing this issue in their FAQs. They said that the AI does not write perfect code. All it does is try to understand your intent and generate the best code it can. This means that the code suggested may not always work or even make sense. They go on to suggest that the code produced by the Copilot should be carefully tested, reviewed, and vetted like any other code.</p><p id="0e0b">So, the idea that this tool will be able to just write the code you need is probably not going to happen — at least not for now. If it writes anything useful at all, you will still have to check it for problems.</p><h1 id="199b">Conclusion</h1><p id="1e68">I am sorry if I have stepped on any toes with this article, but I think it is still too early to deem this tool useful. I mean, Stack Overflow is still better when it comes to providing useful code because you usually pick the one that has the most upvotes. With the Copilot, you get a piece of code that you have to confirm and test for quality.</p><p id="7b3b">All in all, I do not think it is a tool I am very excited about. I am sure some of the coders out there are excited about it, and I would like to know their thoughts. If that sounds like you, I would really love to know your views.</p></article></body>

4 Concerns I Have About GitHub Copilot

Don’t rush into using GitHub Copilot just yet

GitHub Copilot. Yes, the new tool on the block integrated directly into GitHub. If you follow technology and programming news, you are already aware of this tool. If you have no clue what GitHub Copilot is, GitHub describes it as “your AI pair programmer.” This means that you will get suggestions for whole lines or entire functions right inside your editor. Interesting, right?

Well, I have a few reservations regarding this tool. I thought of sharing the different reasons why I am not excited about it. By the way, I know it has not been out there for a long time, so my views are just mine. I am not sure it is a tool I will be using anytime soon or at all. I am not even sure it is a tool I can recommend to new programmers out there.

Here is why.

1. It May Expose You to Liability

You might be wondering where this code is actually coming from. Well, the answer is quite simple. It comes from GitHub. When they made this machine learning model, they trained it on millions of open source repositories from GitHub.

This, at least for me, is where the first problem exists. GitHub Copilot is most assuredly going to disrespect open source licenses. When that happens, we do not know to whom that liability is going to fall. It could potentially be you, the user.

There were a couple of entries in the frequently asked questions that addressed this issue. One of the questions asked was whether GitHub Copilot recites code from the training set. The answer given was quite interesting. They say that about 0.1% of the time, it will recite something directly from the training set verbatim. What does it mean for the other 99.9% of the time? Does it mean that they may take a bunch of copyrighted pieces of material, put them together, and give you that as a new thing? The problem with that would be that it doesn’t produce a new uncopyrighted piece of code. That’s just a derivative work that contains several pieces of copyrighted material.

Another entry in the FAQ that addresses this was one that asked about the ownership of the code produced by the AI. Who owns it? Well, they say that you own it entirely. That is another obvious problem. Think about this for a second: If GitHub Copilot is taking derivative work from copyrighted material and giving it to you, do you really own anything?

2. It Is Another Crutch You Will Have To Pay for at Some Point

This tool will not make people better programmers, although it does not actually claim to do that. I mean, it is really no different from copying and pasting from Stack Overflow. This is just the hyper-integrated version of it. Not that it is a bad thing, but Stack Overflow is a crutch too. The difference is that you have to actually put some effort into finding the code you are looking for with Stack Overflow. It will kind of guide you to the actual solution that you want.

As for the cost of GitHub Copilot, I had assumed at first it was free, but after reading the frequently asked questions, I’m not so sure anymore. They address this in a couple of places in the frequently asked questions. The first is where they say that the technical preview is free for some people. Will there be a paid version? They say that they plan to build a commercial version of the Copilot in the future. That means that the core reason for the preview is to learn how people use Copilot and what it will take to operate at scale.

Of course, you can form your own opinion here, but it is 100% clear that they will have a paid version. It is not clear, however, if they will have a free version. If I were making a prediction, I’d say that there will be a free version with limitations.

In short, it seems like GitHub might be trying to build an entirely new revenue stream around tens of millions of developers who have contributed to open source. I do not want to judge, but you can make of that what you will.

3. Your Private Code Will Be Sent to GitHub for Analysis and Storage

OK, this depends a lot on the nature of your project. I had assumed two things were true from the outset here:

Your private code sent for analysis is not actually stored
The private code sent is not actually used to generate any future suggestions for other people.

After reading their frequently asked questions, I found that only one of those is true. This is addressed in a question asking whether the transmitted data is secure. The response given was that all data is transmitted and stored securely. That means they will store it, but then they go on to say that when humans read it, it is specifically to improve the model or detect abuse.

Does that mean that your potentially private and sensitive code is both stored and/or read by another person? Keep in mind, depending on your project, this could range anywhere from a non-issue (working on open source) to a deal-breaker (working on something super-sensitive). What if you are working on client code for which you signed an NDA, and GitHub Copilot is just ingesting that? The good thing is that they are not sharing that private code with other users — or at least that is what they say.

4. Code Quality

This is not me being all speculative. GitHub themselves have actually tried addressing this issue in their FAQs. They said that the AI does not write perfect code. All it does is try to understand your intent and generate the best code it can. This means that the code suggested may not always work or even make sense. They go on to suggest that the code produced by the Copilot should be carefully tested, reviewed, and vetted like any other code.

So, the idea that this tool will be able to just write the code you need is probably not going to happen — at least not for now. If it writes anything useful at all, you will still have to check it for problems.

Conclusion

I am sorry if I have stepped on any toes with this article, but I think it is still too early to deem this tool useful. I mean, Stack Overflow is still better when it comes to providing useful code because you usually pick the one that has the most upvotes. With the Copilot, you get a piece of code that you have to confirm and test for quality.

All in all, I do not think it is a tool I am very excited about. I am sure some of the coders out there are excited about it, and I would like to know their thoughts. If that sounds like you, I would really love to know your views.