Fear and Misunderstanding: The Real Story Behind the ChatGPT Hype

Claims are emerging such as “Artificial intelligence (AI) has already surpassed humans, it will replace them, and everyone will lose their jobs!” or from the opposite spectrum, “Artificial intelligence is useless, it’s impractical, it just makes things up, and no one knows how it works. We must ban it!”

 

The Truth Lies Somewhere in the Middle

It’s the same as with fire 🔥. A good servant but a bad master. I need to understand how fire works to evaluate what it’s good for and, conversely, what to watch out for and when to protect myself from it. ChatGPT is no different. It’s still too new for most people to say, “Aha, this is how it works, this is how I can get burned, and this is how I can use it to heat my food!”

ChatGPT is like a parrot on steroids with a photographic memory, shown all the books in the world. It can mimic conversation, and it does so convincingly. It has no consciousness, no opinions, no personality. It can only chatter, and its motivation is not to tell the truth but to convince the other party that it can respond relevantly. It’s that simple.

However, new models are already being developed with priorities different from ChatGPT. For example, they serve to verify the facts and credibility of what ChatGPT returns. Everything we see now is just a technological demonstration of a chat talker (literally).

 

How Do We Use Artificial Intelligence?

I often hear questions about how GPT (Generative Pre-Training Transformer) models can actually be used in the real world. We’ve been working with various AI models since 2017. Over the years, we’ve learned to use them in real practice to increase the added value of human work.

For example, we have trained a neural network model to visually recognize technical drawings and detect different types of objects in them. Thanks to this, we can estimate the amount of material needed or the final price of a product. This helps technicians with the routine processing of price offers based on technical drawings. Using another model, based on the same transformer architecture as GPT, we can extract information from digital documents and enter it into ERP systems.

Currently, we are working on two pilot projects with two customers where we use the GPT model for processing and generating texts. It’s not science fiction; it’s reality.

 

You surely have plenty of ideas on how AI could be used in real practice. If you don’t know who to turn to for help, email me at miroslav.bucek@aiviro.com.

 

Miroslav Buček

CEO Aiviro