Hello. My name's Richard. Quite a long time ago now (almost 20 years!), I graduated with a Bachelors of Science in a field called Cognitive Science, from a nice little institute called Exeter University in Devon, in the UK. Nobody had really heard of 'Cognitive Science' at the time. The few graduates this degree produced (we only had about 12 people in my year) all ended up asking "what the hell do I do with this qualification?"
That was 2004. Our phones were still dumb and Google wasn't a major player quite yet in the search engine space. Open Ai, I'm sure, wasn't even the seed of a thought in anyone's mind. It was in the before-beforetimes, before smartphones, tik tok and lockdowns.
I actually did pretty well in that degree. I got a Dean's commendation and otherwise very good grades. I became obsessed with this thing called "connectionism," through the course of studying about brains and computers, philosophy, linguistics, logic and machine learning. In the summer of the second year I managed to land a sweet research assistantship placement with a wonderful Professor named Jay McClelland, a man who could certainly be called one of the Godfathers of modern Cognitive Science. Really the work done by him and his team of cohorts in the late 80s, lays the foundations for most of the Ai stuff coming out now, although not a lot of people probably know that (check out the original book here: Parallel Distributed Processing: Explorations in the Microstructure of Cognition).
As it happens, the neural network model Prof. McClelland and I created was published in the journal Developmental Science (McClelland, J. L. & Thompson, R. M. (2007). Using Domain-General Principles to Explain Children's Causal Reasoning Abilities. Developmental Science, 10(3), 333-356. [PDF]).
Funnily enough, the neural network model I created for my Dissertation Project was a language model of sorts. What it did was predict the linguistic structure of a sentence using a recurrent network to step through the sentence one word at a time. I found that it reproduced the type of delays that humans are subject to while processing language (as measured by eye-movement data during reading). I was thus able to make the proposition that, as humans process language in an essentially statistical manner, statistical neural models are likely on the right track in at least approximating the type of processing that happens in human brains. (My supervisor gave me 100% for the project. Not because it was perfect, but because "we couldn't expect any other student to be able to do any better than this.")
Fascinating stuff, I'm sure you'll agree. After graduating with a First and Dean's commendation, I promptly changed direction, and didn't do any more practical work in computing for about 20 years.
If you've done a degree you loved, you'll know that when you live and breathe a topic for 3 years of some of your most formative years, it comes to shape the person you become. This happens regardless of what you go on to do. The same was certainly true of myself and Cognitive Science.
I mention all this by way of explaining what has brought me to this moment. In the intervening 20 years, I have been a teacher, trainer, coach, consultant, salesperson, research writer, video game tester, personal trainer, carpenter & joiner, amongst other things.
But here I am, back at square one... Back at where I began... Staying up late nights at the age of 15 trying to hack some random bit of code together, probably to get MS DOS to allocate the correct amount of RAM to play Doom 2 on my Pentium 133Mhz, or to devise the inner workings of some extremely basic Joke-bot / OP-bot on IRC (if you don't know what IRC is, I can't help you).
The developments in Ai and Machine Learning have advanced to the point I can't ignore them any more, and I've reached the end of a particular trajectory with my most recent career. I'm 40 (well, 41 actually), and I'm starting again.
But not from scratch. At least there's that.
So to the topic of this, my first post... "Can we design intelligent systems?"
In a way I've been asking this since I walked off that stage, scroll in hand, pasteboard cap on head and poorly fitting graduation gown draped over my sloping shoulders. And the answer up until this moment has been, "certainly not."
The recent conglomeration of developments in the field of Ai, Large Language Modelling, Auditory Analysis, Voice Synthesis, Computer Vision, Neural Networks, Robotics and of course Generative Ai, lead me to think that in the next 5 years, the answer to this question will be "indubitably, yes."
Whether or not these systems will also have wisdom, ecological value - in the Batesonian sense, and real intelligence, however, is far more doubtful. Look at the mess people have made of the world with technology so far. Can we really trust ourselves to just "do it right," when we've made so many mistakes already?
Perhaps Cognitive Science is finding its moment. For me, personally, it certainly is. The history of humanity seems to be that technology and intellectual advancement precedes Wisdom. And sometimes Wisdom never arrives.
We can't afford to follow that pattern with AI. If we build without Wisdom, we will fucking kill ourselves. And there's no shortage of experts in AI who agree on this point. Just search Google for something on the lines of "experts who believe AI will destroy us" and you'll find plenty of horrifying data to support this hypothesis.
So for the next year, I'll be re-educating myself in whatever crucial bits of data I may have temporarily misplaced within my cerebrum, and to newly educate myself, in whatever bits of data I'll need to be able not to just ask the question, whether intelligent systems can be made, but to be able to actually make some of them myself, and possibly even contribute something worthwhile to this field.
Not that I know anything about Wisdom myself particularly, but I think it's a worthy thing to devote myself to for a little while, so let's give it a go, hey?
Yes, I feel very behind. I definitely should have been doing this a decade ago. But here we are, and we can't change where we're at, can we?
This blog is going to catalogue my personal journey into the void that is the modern Ai Landscape. I'm almost certain nobody will ever read this, but if you're here doing just that, I welcome you, Namaste.