The Rise of Large Language Models (LLMs)
Unless you’ve been living under a rock, with no wifi and industrial ear defenders, you’ll already know all about the power of Large Language Models. In particular – Chatgpt. And in 2023 Microsoft launched its own LLM, Copilot, which uses their Prometheus model.
This new wave of AI is allowing developers to produce code at scale, faster than they ever would have imagined possible. Tasks that would have taken days can now be completed in minutes. And as such, the value of highly skilled developers has been yet further magnified.
The mistake is in thinking that these LLMs offer a similar value to less skilled developers. Ultimately, the LLM is only as smart as the person directing it. If the human at the helm inputs low quality prompts or doesn’t know what “good” looks like, the value of the output is going to be inconsistent and unreliable.
In fact, even with skilled developers, these are tools that need to be treated with caution. LLMs are regularly criticised for their tendency to “hallucinate” (generate data accurately and present it as fact) and these errors are easy to miss when producing code at scale.
The other issue is that of Intellectual Property – who owns the code that is being scraped and aggregated to generate this information? It’s a legal minefield that’s yet to play out.
That said, we should be under no illusion – while it’s far from perfect, the role of AI and LLMs is here for good, and the tools we’re using today are the worst we will ever use. This is a trajectory that only knows one direction.
The Continued Evolution of Low Code / No Code
As with LLMs, we are still in the early stages of low code / no code solutions. With each month that passes, they make coding simpler than ever before, with enhanced usability features like visual interfaces, easy drag and drop functionality, and ready made templates. This makes it more straightforward for new developers to get up to speed and even opens up accessibility to those with no coding skills whatsoever.
As of today, there continue to be lots of things low code / no code solutions cannot do, but as with LLMs and AI, this is a one way journey. And it’s not hard to imagine a world in the not too distant future, where entire solutions could be designed and built by anyone, no matter what their development experience.
