Which developers will survive Microsoft GitHub Copilot AI?

Which developers are going to be okay in the coming AI tooling wave that Microsoft is bringing with Copilot? My thoughts on the topic, plus a discussion on Developer IQ.

0:00 Who are Microsoft's actual customers and what do they want?
1:22 Developer IQ and tooling's impact on it
4:44 Which developers will be most impacted?
5:44 Who else is Microsoft gunning for?
6:33 My advice for different levels of developer for surviving the AI wave

(this is a re-upload as there was a glitch in the previous version - my apologies!)

#csharp #dotnet #githubcopilot

[edandersen]

Thanks for watching! If you disagree with my take (highly likely) please let me know why as I'm trying to figure out the implications still myself 👍

[aaronbcj]

Most enterprise developers hardly code 20 lines of code a day, most of the time spent in "maintaining" it with configuration changes. Startup owners and new product developers will gain the most using AI tools.

[TL]

I am a beginner and I have heard that writing code is only 5% of software development and that replacing writing code is a normal thing that has been happening since the beginning of this industry.

[johnprincipes]

AI tooling is basically useful only if you know what you are doing. Sticking even to open source project collaborations, I have seen AI hallucinate some hilarious things. A scripter or even an experienced developer is not going to be able to compete with a skilled, experienced dev by using tools like copilot. Its the junior dev at risk - you dont need as many if you can hand your senior devs co-pilot and have a bucket of snippets at their disposal to wire together the easy parts.

Also, someone who cant code can just be automated with AI. Same with analysts who just assemble reports. Not sure you need anyone but above average devs if you have AI, especially for MSFT houses. Devops or GH can take care of CI/CD and it all wires up, so i dont think it plays out that way long term.

Also, all products are aimed at the check signers, its just that the Gartner hype cycle is bad right now. I appreciate your fresh perspective, but take a step back - MSFT is here to make money, but so is everyone else.

I do think that companies will get stupid and try to do it on the cheap. Im just saying it wont do - anyone who has implemented an algorithm short on sleep knows the difference between an experienced, trained stance versus an uneducated one, and the fundamentals just aren't there yet. Maybe decades from now... but not yet.

[shafsterk]

Github Copilot is an excellent learning tool if you know how to use it and already have good foundational knowledge as a developer. It can even be used as a debugging tool when something inievatibly goes wrong in your codebase.

[theHenrik]

I was about to smash the like button three times, but after thinking about it, this might make a lot of sense. At least in the short term. In the slightly longer term, I think no job is "safe", but having the skills of general problem solving will for sure help exploit the period leading up to that. I also don't think we need to panic about that long term possibility.

Great content, we need more like this to expose us to challenging thoughts that many would rather be ignorant about today.
Ended up hitting the subscribe button three times instead.

[sialababamak]

I am average developer working in bank and I'm not worry at all. I might be more expensive than junior with copilot but I do understand what I'm doing. When production failure happens know how to fix it.

[ok_comment]

Great video, Ed! What’s your take on low-code platforms like Power Platform and Dynamics 365? It doesn’t feel truly low-code if only basic apps can be done out of the box, and custom development, especially done with C#, is still needed for real client demands. Thoughts?

[FlyFisher-xdje]

Very intersting...

I agree with much of what you said. One thing that will save developers is cultivating other skills along side their technical skills. You mentioned product management, I think you could be on to something. Could there be a melding of product management and software development? What about SDETs? What about DevOps Engineers? If these tools improve productivity couldn't that free up time for software developers to take on tasks that SDETs or DevOps Engineers do now? I think so. As a matter of fact, I think what would be fantastic is enabling software development teams to own their software in all environments and be responsible for operations. I have managed teams that did this and it worked great because it incentivized higher quality code and observability.

Personally, I don't view myself as an .NET/C# developer or a Python developer. I think of myself as a problem solver that uses software development skills as tools. This could be a very good thing, at least for a while.

My suggestion for developers is to work on soft skills, technical skills in adjacent domains, and business acumen.

[ranz]

I like your straight to the point attitude and the ironic smile.
But, can improved tools really increase the performance of a programmer whose level is below average?
As you mentioned, the advantage of a good programmer using tools like copilot is, among other things, the ability to find errors in the code that the tools generate. But what does the lack of such an ability mean? Well, this means that the errors will in many cases be discovered by the end users. Or by good programmers who will have to maintain the problematic code, which will make their work more difficult and reduce their productivity.
I don't think the tools available today, including copilot & chat-gpt, can bridge over the gap between the skills of a below average programmer and the output of a good (or even an average) programmer. My guess is that in many cases disappointed managers will find that the opposite is true, these tools allow bad programmers to hide their poor skills on the short term and add code whose problems will be discovered later on and will be expensive to fix.

[hemant-sathe]

Microsoft since it's early days as a developer tool company has a dream. Business folks know the business and they don't get a product they want for software developers. So they chased code generation as a philosophy in all their tooling. They wanted to make coding easier for business or non IT folks. Most other developer platform companies laughed at them. Microsoft developers were rarely considered as designers or architects simply because they wouldn't understand the design details hidden by the tooling. That was broken to some extent when they started with MVC. That was probably first time when Microsoft developers tried something like a real design pattern, architecture style and were involved in the serious design discussions. MVC still had some magic involved but the developers needed some depth. Microsoft tooling didn't find its mojo of code generation since then. They tried their hands with some code generation in .Net 6-8 but that's a bit complicated. The AI tooling has given them that old trick back. Now they can again go back to businesses and say look guys, now you can code without depending on developers. You can write unit tests, understand the legacy code, deploy your code in much less time. They can once again sell something to business saying you don't need developers.

[nicholasah-loe]

I feel like it's not intended to hire cheaper lower skill developers, the job market is looking more like they want 1 senior dev doing everything and laying off anybody average or below.

[franko]

The same can be said for any language, tool or framework in tech. Do you think tech will always be in high demand like in medicine and engineering or will it become more saturated like finance and law fields?

[fatihozgen]

AI hype sooner or later will hit its plateu with current models. Diminishing returns is inevitable. With current models it'll never hit that %0 mistake. You can think of what %1 cumulative mistakes can do in large projects. Plus tooling is not there yet.

[ahmadkharsah]

nice video, do you think that chatgpt can be a good choice for developers instead of co-pilot?

[freeideas]

I must be a 70 DiQ developer, because I pull my hair out every time I drill down into all these Microsoft Tooling UIs, and I end up writing things from scratch "the hard way".

[muhammadnasirilyas]

Am scare and excited at the same time, I dont know what category I am on the scale. Anyway am pretty good with blazor, running away from js of course, now am learning Python not sure what to do with it. Thanks for the video.

[PticostaricaGS]

This is not accurate, in enterprise scenarios usually developers do not do deployment, and the people in DevOps do lots of work without Visual Studio, since they have to setup pipelines. Also, many MS shops actually pay more due to higher productivity since it directly translates in faster Return Of Investment than non MS shops.

[Chiramisudo]

So they're wiping out the Developer "middle class" eh? Lovely! Given my recent empirical experience of not finding a job for 8+ months (and being unemployed for 16 of the last 21 months), my arguably below average C# and .NET Framework skills are not wanted. F*ck me! I don't know how to get out of this damn pigeon hole I got myself into by supporting (barely writing any code, but mostly maintaining) loads of freaking legacy apps for 11 years at the same company. 😥

[gameboy]

How come you are far smarter than the average tech youtuber ? :))) Thank you a lot sir.