Earlier this week, we kicked off a new episode of our “Tech Connects” podcast with a guest. Adam SeligmanVice President of Developer Experience at Amazon Web Services (AWS) AWS launches a portfolio of AI-powered helpers and tools for developers, showcasing how advances in AI and other technologies are rapidly transforming the tech industry.
Here is a video of our discussion:
The main points from our talk were: Generative AI While generative AI will change the software development landscape, it won't curb the demand for technical talent. There will still be a need for highly skilled, specialized developers, but generative AI will allow people with less experience to enter the field and contribute to generative AI and other technical teams. This is because generative AI tools make it easier for people to try out and learn new technologies. This will create a more democratized approach to software development, allowing more people to participate in creating software.
Here are some excerpts that delve deeper into these issues, and be sure to watch the full episode for more details.
Q: One of the things that was interesting to me was the idea that you were pitching there, that when you talk to most people about generative AI and machine learning and stuff like that, there's a lot of things to focus on. You have a talented upper echelon of PhDs and so on who have very high levels of expertise, and they're busy every day getting training and getting their LLMs, and figuring out how to integrate it into their workflows and their products and all sorts of other good things. But what you were hinting at is that the more junior people on generative AI teams, people who have taken a few classes in AI, are just as valuable to these teams in terms of training and getting their LLMs. That was interesting to me, and I wanted to dig a little bit deeper into what you're saying.
Adam Seligman: I think generative AI capabilities are going to upend a lot of assumptions about being a software developer, working as a software developer, or even becoming a software developer. They're going to upend those assumptions. They're going to allow aspiring developers to discover new technology, ask for a little help, get it explained to them in the way that makes the most sense to them, and try it out instantly like they couldn't before. It's great to have that kind of capability at our fingertips for anyone who wants to work with the cloud and technology.
This is really interesting to me because before the LLM AI revolution really started, there was a push for no-code and low-code tools, democratizing the ability for anyone from startups to large enterprises to sit down and build something like a simple app, with (for lack of a better word) some or no coding knowledge, or even without extensive coding knowledge. This AI revolution is going to accelerate that even more, but when you talk to different people, the timelines seem to be different. When you talk to some people, they say, “Basic coding knowledge is still essential, and the time when you can just type in a prompt and have amazing software is still a long way away (or never will be).” But when you talk to more optimistic people, they say, “Developers will be able to launch software instantly within a few years.” What do you think? [scope]Because you're in the privileged position of being able to see where things are evolving.
Adam Seligman: We learn every day, and thanks to our customers and community, we see generative AI innovation happening everywhere: our customers are some of the most skilled data scientists and ML engineers building and training custom models. Sage Maker It's built on your own enterprise data. Experienced developers use Bedrock to build complex systems and then apply AI to build back-end agents, intelligent automation, and more. With generative AI tools like our Queue Developer product and others on the market, developers of all skill levels, including interns early in their careers, are upskilling incredibly quickly. And there's plenty of choice, as new features are available to everyone, allowing them to push the boundaries of their current capabilities.
Q: It's amazing how quickly things are evolving. Every time I log into a generative AI tool and try it out, there's a new feature added, and I'm amazed.
Adam Seligman: I've talked before about a time when generative AI will replace coding by writing out loads of code, but I don't think that's the right mindset. What we're seeing with our customers and with our developer community is that everyone can move very fast all at once now. They can ask for approaches. They can get help with migrating their old Java code using things like queue developer transformation and code transformation. They can ask for serverless architectures or how to build event-driven patterns to solve their specific problems. No matter what stage they're at or what problem they're trying to solve, they can get help right away. So now, having that help again, just in time, is great in terms of empowering humans to achieve more.
Q: And then there's an element built into the tooling that you're using there, right? So when you ask for code snippets or whatever, you can see that what you get back is, for lack of a better word, relatively…
Adam Seligman: Yeah, software is never done, right? We all know we're on a journey of building stuff, adding features, fixing issues, learning how to operate at scale, running in more places, adding new features. So, software is never done. You have to have that mindset. We believe the best way to equip developers is to help them across the entire software development lifecycle, not just one modality like inline code completion or chat. It could be learning, exploring solution architectures. It could be planning an approach to build scaffolding code for your application like inline, writing tests, suggesting improvements to your code, migrating your code with chat modalities. Like from Java 8 to Java 17, Q products help with that. Debugging operational issues. So, you might have some challenges when you get an error message that you're not familiar with. “Help me understand,” it could be a chat modality, but it's entirely on the operations side, not the software development side. At Amazon, we try to help our customers and developers at every stage of the entire software development lifecycle, and we think this exponentially improves all our developers and their capabilities.
Q: When companies hire junior developers, they go through a whole interview process, of course, and they're evaluated. But even if you already have a lot of skills and experience and so on, when you talk to hiring managers, project managers and so on, there's always this concern: Can people rapidly improve their skills? Can they jump into the ongoing workflow and be ready and willing to contribute? I think AI can really help in this regard. It's like you can get up to speed quicker than before, and I think that's definitely a great potential benefit.
For example, let's say you're part of a development team and you're tasked with building models, training models, etc. Many of the people joining these teams, especially interns and junior developers, may be great jump-start engineers. They may be great at running queries. But what skills do they need to make the leap into the architecture and training side of things and quickly become fully functional on an AI team?
Adam Seligman: I think it's easy for those of us who have been in this space for a long time to have this preconceived notion that early in your career you're less competent and then as you go along you become more competent and you can get really deep in certain areas, and what I think is interesting is that the power of the cloud and generative AI can make you a really valuable resource with really deep areas, without necessarily having years of experience. [training]Let me give you an example. With Sagemaker, you can train or fine-tune your own models very easily, and you don't even need to know assembly language or C. It's pretty amazing what you can do. So, for enterprise companies and large customers, an intern might train a basic model in Bedrock or Sagemaker. The intern might do not just prompt engineering, but quality engineering to ensure that these generative AI systems are high quality. So, a lot of preconceptions are changing in the world of generative AI. I'm an optimist, and I think it's very exciting that people in the early stages of their careers suddenly have superpowers and can really benefit their companies by building great software and solutions to solve problems.
