Ian Harris
This one create opportunities there with AI again, humans are terrible at detail of large quantities of text as you're reading through code and you're trying to debug it and understand what's going on, you're trying to understand the meaning in the detail or someone else has put into this code. Having an AI be able to look at software and explain it to you is a fundamental shift in the way we interact with software. It's often the case that the person that is originally wrote the code that you're looking at is not there anymore, for whatever reason, and have something else explain it to you in a way that makes sense, that it's relevant to the topic that you're talking about, understands the context of the code as a whole, understands the particular language, and then can say, but by the way, there's a few errors in the code, which under these circumstances, you'll have these bugs that you might not have realized before, be able to spot those edge cases that humans are we really struggle with discovering these things ourselves. They'll be able to spot things that we can't see, and they'll also be able to refactor and be able to reorganize the code, make it more obvious to see what's going on, even from a human perspective. Being able to review the code like that. I've heard a lot of good work going on at the moment with AI helping out with pull requests. So being able to find the changes of software that's going into an organization and and being merged into it, into the main branch, being able to describe that in a way that is clear and understandable and repeatable and contains all the information you need. Again, humans are terrible. I have to add all this stuff in, I have to add this detail, and I have to describe something I've forgotten to even reviewing pull requests I've seen AI do fantastic. Oh, you forgot to describe exactly what the use cases is then. What description doesn't actually match that code change in fact, or your code change does more than what you've described in the pull request. Fantastic. Well, this level of detail and being able to do basically do a better job at software because we've now got an AI that really deeply understands through millions of lines of code. The software itself, probably better than any human would be able to do in terms of that level of speed and accuracy and being able to get that information out of that code itself. So there are a couple of examples. But I'm also saying now we're starting to look at ways of describing software in not so much architecture diagrams but documents. So taking product documents that describe how software should function and what its limits do, and then being able to read the document and say, okay, well, that's what you want to do, then you're going to need these sort of software modules, and they're going to need to interact in these sort of ways. You're going to need a whole bunch of APIs that you can go and get it to write the APIs for you. Fantastic. And then you can say, okay, I've got these APIs, but we need to store the data, obviously, and it can create the database schema for you. Great. Okay. That's another mundane task that slog through it and work out what the different fields are going to be or AI can do it for you and do it much more accurately than I would, and I'm going to get there, but I'm going to make mistakes along the way. And again, it's back to that force multiplier. Humans have created all this stuff on average overall eventually, but if I can do it in 30% of the time and I haven't had to sit there, and even if I've got the idea and I know the fields and I know what to do, I've got to sit there and type it. And no matter how good a typist you are, if you're thinking and typing at the same time, it's slow. It just takes time. And then you got to review it and you got someone else to check it. Meanwhile, the AI just done it. You're done and you move on to the next thing. And so you see that developers will get faster. And this is great, because the fun part of software is the thinking, the designing, the overarching architecture. It hangs together, the typing it up, the code, the writing, the code bit. That's actually not that fun. The fun bit is getting it, working it, debugging it, and seeing it all connect together and seeing the whole thing function as a whole. That's the exciting part of software. And so I think it's going to make software more fun, and it's going to make it faster. So you'll be able to get much more done with less people.