Posted by Over the last couple of years, I’ve been giving talks demystifying AI which is a totally joyful labour of love.
I started soon after ChatGPT began hitting the news and we were reading headline after headline about how AI is going to take over the world. I had also attended a talk that, at first glance, seemed credible but was utter science-fiction.
The speaker had been a former guest on a so-called “techbro” podcast where no one speaking into the microphone possessed a formal background in computer science or AI. Yet, they delivered a series of confident, persuasive statements predicting that ChatGPT would imminently spawn robots and take over the world. The speaker went on to repeat this fiction, almost verbatim, the night I was listening to her. It was compelling but entirely false, highlighting the gap between popular perception and technological reality, which inspired me to begin offering fact-based, grounded perspectives on AI.
My talks are jam-packed with how AI actually works, from the Good Old-Fashioned AI (GOFAI) of Knowledge-Based Systems of the 60s and 70s, to the highly structured architectured designs of modern day ‘deep learning’ artificial neural networks, such as ChatGPT and AlphaFold.
Contrary to popular belief, neural networks are a major piece of software design. They haven’t spawned themselves at all, they cannot rewrite or update themselves as they are not designed to do that. And, throughout the history of computer science, self-modifying code has never had much in the way of success. Even the ones that were declared a success, such as Doug Lenat’s AM, then turned out to be not so successful. It was a bit of hopeful interpretation, a bit of wishful thinking.
In the same way, most of the code that generative AI writes is not very good. Think of a bell-curve of the code some AI has written, one end is very good, the other end is complete rubbish, and in the middle, the majority of the code is uninspired and not very good for all sorts of reasons from semantic inaccuracies, to just not very good code, from inefficient to inelegant. And, yes I know everyone says: Ah, but once the AI catches up… but honestly, that is just wishful thinking too. There is a lot of terrible code out there and the AI no doubt has read it all.
I love to talk about the technical debt of such systems, how human-computer interaction is crucial, and how there is as much storytelling around the fields of computing and AI, as there is anywhere else in society. In short, I just love creating and giving these talks.
There’s usually a gap between each time I give a talk, so when I return to one I find myself once again shaping the story I want to tell, asking myself: Ruth! Why didn’t you document this better? So, instead of future me nagging past me, I decided I would do exactly that. So, over the past couple of weeks, I finally took the time to go through each talk, review my scrawled notes, half-written scripts, researched cut and pastes which seemed crucial at the time to keep, and the many references I have noted, in order to pull everything together into my very own little book of talks.
When it arrived from the printers, I was delighted. It has a lovely front page (which I created from the pictures I’ve made on my website). It is A5 spiral bound, has a table of contents, and at 151 pages all told, it’s a joy to hold, flip through, and read.
I’m already in the process of organising new dates for talks, and having this book makes the preparation time so much more inviting and a lot more fun. To say nothing of a few new talks I have planned. I already love my little book of talks.
It’s a tangible way to revisit the stories I’ve told and scribble down new ones, ready to bring them to life for the next audience. As JFK once said:
We choose to do … this thing… not because it is easy, but because it is hard….
This is a quote from a famous speech JFK gave at Rice University which I like to quote during talks on AI because the Space Race helped fund so many projects. At my last talk, some bloke knew the whole speech and recited it correctly (I’d edited it a bit) so we all gave him a round of applause.
Anyway, I do talks because I love them as much as I do the top comment on the NASA video of JFK in the link above, that says: This speech is so powerful to me I listen to this before the gym.
I only wish to inspire my next audience that much too, though I hope they don’t run off to the gym before the end of my talk.
