The ecosystem surrounding large language models (LLMs) is growing faster than the number of developers with LLM skills.
Does that mean that now is the time for programming students to specialize in LLM development? Should they enroll in one of the LLM courses that have sprouted up in recent months?
For most people, the answer is no. LLM development skills might become a useful asset for some programming careers, but I wouldn't bet my future on LLM expertise. Here's why.
Growing Demand for LLM Development
As IT Pro Today recently reported, the surge of interest in large language models — which are the technology behind generative AI tools like ChatGPT — has spawned an increase in demand for developers with LLM skills.
That's unsurprising. It's hard to deny that generative AI is the most significant disrupter in the technology industry at present. It makes sense for companies to seek software engineers who can help them leverage LLMs to power custom generative AI tools capable of helping them solve unique business challenges — like interfacing with customers or helping to write custom software.
Also unsurprising is that a flurry of courses and certification programs related to LLMs have emerged over the past half-year. Massive open online course (MOOC) providers such as edX offer LLMs courses. You can sign up for an LLM bootcamp that promises to teach developers the fundamentals of LLM programming in a matter of weeks. Professional certificates for LLMs are now available, too.
Two Reasons Not to Become an LLM Developer
Given the clear demand among businesses for developers with LLM skills, combined with the increasing availability of educational resources and programs that teach those skills, you might think that now is the perfect time for people considering development careers to become LLM experts.
I'd caution them against doing so, for two reasons.
The first, and probably more obvious, factor is that LLMs may be a big deal today, but they're probably not going to remain the hottest technology for years on end. Tech trends come and go, and while there's no denying that LLMs and generative AI are a very big deal, there's no reason to think they won't turn out simply to be a trend. There will always be some level of interest in them, but they're not going to be a top skill forever.
Second, I wouldn't place a strong bet on many companies hiring large teams of LLM experts because I suspect that a lot of LLM development needs will end up being met by LLM-as-a-service providers. Azure is already investing heavily in this space through its OpenAI Service, which provides access to ready-made LLMs that developers can leverage to create custom generative AI applications.
These services do the heavy lifting needed to create and deploy LLMs. They make it possible to use LLMs without having to design or train them.
That means that developers who possess the rare skills necessary to create and deploy LLMs aren't actually very necessary.
To be sure, there will be some organizations out there that want to develop their own LLMs from the ground up. They might have complex LLM use cases that generic solutions like Azure OpenAI Service don't address, for example. Or they might want to use unique sets of training data that aren't available publicly. These companies will need to hire LLM developers.
But that's the exception. The vast majority of businesses, I think, will be perfectly happy to consume LLMs as a managed service. They might require engineers with some basic LLM skills to help use those services, but they'll have no reason to pay programmers enormous salaries to create LLMs from scratch.
What Makes LLM Skills Unique in the Labor Market
This makes LLM development different from other major IT trends, such as the shift to the cloud. Cloud computing skills are a vital asset for most developers and IT operations engineers today because almost every business needs to use the cloud, and few companies can outsource their cloud development and management needs.
LLM development is also different from cybersecurity, another domain where people with relevant skills are in short supply. Every company with IT assets needs cybersecurity, and although some can outsource their cybersecurity requirements using managed security service providers, there are enough businesses in need of in-house cybersecurity teams to make cybersecurity a solid career path.
Conclusion
I'm not saying that LLM development skills are totally useless or that no one should specialize in LLM development. LLM expertise could be a valuable addition to your resume, assuming you have strong skills in the core software engineering domains that employers will value over the long term. And if you are intrinsically passionate about LLMs, go ahead and take some LLM courses.
But don't go chasing the LLM development fad just because you think it's the ticket to a high-paying, stable software engineering job. Ask ChatGPT. It says it's not.
About the author
Christopher Tozzi is a technology analyst with subject matter expertise in cloud computing, application development, open source software, virtualization, containers and more. He also lectures at a major university in the Albany, New York, area. His book, “For Fun and Profit: A History of the Free and Open Source Software Revolution,” was published by MIT Press.
