The goal of this blog is to help you gain strong intuition about AI and navigate through the AI world with confidence.
How it all started
I am sure like me you woke up one day at the end of 2022 with the deafening buzz around OpenAI’s ChatGPT. First skeptical about a new gizmo. The more you played with ChatGPT though the more you got mesmerized by its conversational potency. Merely a year passed since then and now we are witnessing an unprecedented AI revolution unraveling in front of our eyes. People talk, breathe, and live AI these days. It is promising yet a bit confusing, if not scary. AI excitement comes with endless possibilities and opportunities accosted by lots of speculations and worries.
Why bother with another blog about AI?
Today the internet is replete with all kinds of sources about AI, ranging from approachable narratives for the uninitiated, to recipes for AI practitioners all the way to profound (and pretending to be such) scientific papers catering to picky academics. Every day massive new information gets uploaded to the internet with a mix of marketing, technical and academic hype. It becomes increasingly hard to discern what is empty calories and what would be useful proteins to consume in order to gain solid footing in AI.
I very much hope that the posts you find here will help you stay comfortable with foundational AI why’s while navigating through the sea of what’s and how’s.
Who are you?
You are eager to stay relevant and somehow to matter in the fast paced AI world. But it is a bit overwhelming to be in the midst of a barrage of what’s and how’s. You start getting anxious, feeling desperate that you can’t catch up and stay abreast with all the latest trends in AI. Perhaps you are lamenting that with no Ph.D degree in STEM you are hopeless. After all, there are only 24 hours a day. We have day jobs, families, hobbies, and other enjoyable things to attend to.
Yet if we don’t gain solid intuition – what AI is about, why it works and how to make it enhance us professionally and personally – we indeed one day may end up relegated to irrelevance.
Goals
One of the main goals of Enter AI blog is to help the readers of all walks of life to gain intuition about why Large Language Model(s) work, why the magic of Generative AI is not quite a magic rather based on solid and cutting edge science.
- Why does it work?
- What are the scientific underpinnings for it to work?
- What are the limitations?
I will make sure not to bore the reader with fancy terminology and make scientific concepts as approachable as possible.
It is important to keep in mind though that AI is deeply rooted in interdisciplinary scientific landscape – physics, mathematics, probability and information theory, computer science, computer vision, linguistics, dynamical systems to name a few foundational domains. I promise I will help very gently soak your feet in the relevant scientific topics as needed. Necessary links and references will be provided for further research for the interested.
The AI landscape is huge. Another main goal of this blog is to help the reader choose a specific AI focus area for further exploration.
Methodology
My methodology is rooted in prioritizing building strong intuition in a rather complex AI domain. Usually novel concepts in traditional education are tackled with a bottom-up approach: you need to know arithmetic before learning algebra, which you need to learn before learning calculus, which you need to know before learning anything else, and so on. When learning such a broad and deep domain as AI, taking an academic bottom-up approach could be a very lengthy, frustrating and counterproductive if not intimidating experience.
I am leaning toward describing mechanics of why AI works from a top-down point of view before getting buried in nitty-gritty details of what’s and how’s. I am in favor of non-linear learning methodology as opposed to linear one; think big vs. get bogged down with details way too early. Of course, we can’t ignore the fact that some key concepts require knowledge of some other fundamental building blocks. But we can’t get carried away by the endless number of dependencies on prior knowledge. I will help manage filling this gap.
Once you are solidly convinced that AI is deeply rooted in science and is not some weird manifestation of voodoo alchemy of select AI researchers you will be more confident in getting your hands dirty with practitioner’s aspects, nuts-and-bolts of AI implementation when addressing real world problems.
About your host
My name is Nicos Kekchidis. I have majored in Applied Mathematics and Computer Science since the late eighties of the last century. It was right around the time when academia was dissuaded by Expert Systems’ direction of AI and started exploring new ways of implementing AI; ushering in a new Deep Learning AI era that we currently live in. It was then Geoffrey Hinton, Yann LeCun and Yoshua Bengio among other key contributors who laid out scientific and engineering underpinnings of modern AI as we know it today.
I’ve been in the computing industry for a long time, and have done several AI and Machine Learning projects. But even having solid related education and exposure to AI in the 2010s behind my belt, I admit, like many others I was caught by surprise with the arrival of ChatGPT. Immense potency of modern Large Language Models and Generative AI as a whole disarmed me. How could I have missed what was coming? 🙂 But, hey, it is not all lost. We are at the very beginning of the AI revolution.
My friends, enjoy your amazing journey in the fascinating AI world with confidence! I am here to help you. Let’s learn together.