Did you know that even artificial intelligence (AI) engineers don't fully understand how large language models (LLMs) think, even when they appear to give the right answer?
As uptake and experimentation with AI increases, there are too many business leaders that don't really understand what drives the technology. How it is put together, and what truly determines its output.
To truly get to grips with the tech and its capabilities (and challenges), it's not enough to tinker with different applications and experiment with a string of prompts. You need to fundamentally grasp the mechanisms underpinning the technology to know how it can impact your business. Otherwise, you're playing a risky game-after all, there have been plenty of high-stakes, Al-related errors over the years.
Conscious of this, I decided to build an AI from scratch over the past months. A do-it-yourself (DIY) Al to discover what makes it tick, and to share first-hand the lessons every business leader or global organisation should know.
Built without pre-existing, reusable code, my Al is a simple thing, nothing fancy. I limited it to eight-bit to work with small dimensions, inputting simple visual data that would be easiest to learn. Satellite images seemed fun-if my DIY AI learnt enough, perhaps it could create a slice of a brand-new world for me.
Read the full article here.
