Carlos Garcia
Head of AI Strategy, Irish Manufacturing Research
Large language models (LLMs) hit the world unlike any AI technology before. Suddenly, with widespread accessibility, extreme views poured in: some fear AI will enslave humanity; some believe AI will solve all our problems.
One thing is clear: AI excels at many tasks that do not require human-level understanding, reasoning or planning.
AI in software programming
Possibly the first instance of programming, in the sense of writing and running instructions on a programmable computer, happened in the mid-40s. For many years, the procedure involved punching holes in a tape or later paper cards, something reminiscent more of a manufacturing task than coding.
From there on, things moved fast. First, assembly languages, which still required a lot of low-level tasks not related to the algorithm itself — then, increasingly higher-level, more abstract computing languages that allowed the programmers to focus on high-level concepts instead of tedious mechanical tasks.
Today, we have AI assisting us to put our ideas into software, enabling a much broader population, even those with minimal programming knowledge, to create applications.
AI is not replacing humans
but augmenting our abilities.
AI augments human capabilities
This evolution from a mechanical endeavour punching cards to real-time AI assistance reflects a shift from mechanical precision to creative exploration — a journey that can be extrapolated to many other work-related tasks.
AI is not replacing humans but augmenting our abilities. This is a trend that will only increase: AI will take on more of our mundane tasks but not replace humans who, at least for a few decades more, will still hold the high ground on mental activities like reasoning, planning and creativity. In my opinion, the AI we have today is a sparring partner for the more abstract, creative tasks and an able assistant to which we can delegate routine tasks.
LLMs as probabilistic aids
We can see this in the current wave of LLM hype. Many people use them as a one-shot intelligent tool and expect a perfect task result, completely replacing our work. However, these tools are probabilistic in nature, thus, there is always a chance that the output they provide is wrong. The best use of LLMs is using them for ‘knowledge sparring’ instead of leaving the work to them without any human supervision.
AI is a tool that complements human-only skills, taking on repetitive tasks and enabling us to focus on higher-level, creative work. By understanding and embracing this dynamic, we can navigate the complexities of AI with a balanced perspective.
