How LLMs Keep Getting Better And Why Most People Are Misusing Them
How LLMs keep getting better and why most people are misusing them comes down to one fundamental distinction - a tool should extend your thinking, not replace it.
"If I had asked people what they wanted, they would have said faster horses."
- Henry Ford
The argument that AI is inherently bad is wrong.
You can’t condemn a knife for cutting vegetables just because it was used in a stabbing. In fact, you keep using it even after a cut. Take a hammer or screwdriver, both can be misused as weapons, but does that remove their primary function of making work more efficient? I can’t drive a nail with my bare hand. That’s what the hammer is for.
You learn algebra first, and then schools introduce calculators for complex calculations. Tools don’t replace design thinking; they extend it. They complement each other.
AI agents are not software engineers.
As human beings, we have used tools long before medieval times. Remember watching The Flintstones? I did. They would carve stones and sticks into more efficient tools. The premise was never “what if this is used to hurt someone?” The premise was pure efficiency.
That said, you cannot outsource your thinking to AI. Why would you? Imagine being deprived of the ability to think and make decisions for yourself.
You become a pickled onion.
The legitimate use of AI is in the eye of the beholder.
AI is for execution, driving efficiency and delivering productivity gains. Should we blame the tool or the lack thereof of education on how to use it? Blaming technological advancement is nonsensical because these contradictions tend to surface the moment AI touches upon someone’s specific area of expertise.
Describe what you need AI to do or build, but stay in control of the output. For SMEs and large-sized businesses especially, the goal is to increase efficiency in your processes. What you need is a team that understands your business and can suggest concrete, customised solutions to scale it.
What is your advantage?
Working with a team that knows the hidden pitfalls of AI, and can help you avoid them.
AI is extremely fast at iteration, it's getting better by the day, and what it can accomplish is a seriously cool feat. Whether AI is good or bad lies in the grey area of how it is used, what it is used for, and how well it executes.
Be at the front leveraging AI with help from a team of experts, not at the back blaming it after choosing yet another AI product that fails to do the job while your competitors take the lead.
Laramate GmbH has delivered over 100 projects, built +50 custom solutions and is currently helping several others get out of technical debt.
Have you heard of the Ghost Tech Debt ?
Don't wait for your next AI feature bolt-ons.