No, this isn’t some kind of pledge of undying devotion to ChatGPT, the Artificial Intelligence (AI) that has reawakened broad interest in ‘intelligent’ computers. It’s just that I only Googled “You’re my lobster” this morning and for the first time found out what it was about.
So you might be in one of three camps here: (1) you already know what “You’re my lobster” is all about; (2) You have no clue at all and it sounds like gibberish; (3) like me, you know that it was a saying from the popular nineties/noughties TV series ‘Friends’, but, because you studiously avoided ever watching it, never knew what it meant. It was from early in the series and so I remember hearing some of my friends (actual ones, not TV series people) use it. It raised smiles and chuckles.
Early morning, right when I just wake up, is the time my mind is full of ideas and answers to things from the day before. Every day. There’s apparently some good brain science behind this: my dreaming mind has been exploring paths less familiar than those exercised in my daily routines and so makes new creative connections. Like a dream, if I don’t write my waking ideas down quickly I soon forget them.
So this morning one that I woke up with was a need to look up “You’re my lobster”. And it’s because I remembered overnight that I once used this phrase myself, even though I had very little idea what it meant or signified to the millions of others who had watched ‘Friends’ in the nineties. But I have to admit at least I had the context that it seemed to go down ok.
The occasion, quite some years ago, as I remember it, was handing someone a mug of tea and noticing it had a lobster picture on it. Like a push-button reaction I said “You’re my lobster”, denoting the lobster, as I handed it out. I recall two things that happened instantly and in parallel: (1) It went down OK – probably met with a smile; (2) near panic inside as I realised I had no clue what I might have said.
So ChatGPT is the text-generating AI that has astounded the world with its human-like and detailed responses to being asked to comment on or write about all kinds of subjects. Lots of people are astounded. I worked in AI and in particular in natural language understanding right at the start of my career (and it was the start of the path to KM, too) and I’d say I too am astounded just how far the tech has come in addressing lots of the fundamental problems – mind you, we waited almost 30 years for this. Equally, I think there will be a whole lot of engineering to do to cross the gap that still exists between what we have now and what people are, once again, imagining. Science fiction.
As well as the enthusiasts of course there are other voices emphasising all sorts of drawbacks and warnings about this tech – and we should listen to both sides because there is truth in each. And one of the points that I find very valid is that, however well the AI appears to perform (Turing test, anyone?), we must remember it actually has no real first hand knowledge, experience nor context. It’s a machine, a fancy one, but a machine. Part of what it does is that it matches what follows to what proceeds. Yes, like predictive text ‘on steroids’, as the saying goes, it ‘knows’ what is usually said after what was said before. Turns out we humans are far more predictable than we probably like to think (but get over it!). As I mentioned, my daily waking mind follows routines; my sleeping, dreaming mind explores the paths less trodden. My daytime self is somewhat predictable – my night-time mind has unique adventures: You only live twice, or so it seems, once in your life, and once in your dreams, as the James Bond movie theme reminds us! So, yes, as I went to bed I was thinking again about natural language AIs that don’t really understand what they’re saying, but how they may appear to.
So there I am, handing out the lobster mug of tea, several years ago. Innocently, like a robot, I blurt out “You’re my lobster”. I have enough data to know this is a light-hearted thing you can say to a friend and the cue of the lobster seems to make it appropriate, or, at least, possible. I get away with it. There’s no thumbs up or thumbs down button to correct me, but I kind of get similar from the smile reaction, and I might have rushed to look it up sooner had I been met with a frown.
But in that moment I’m ChatGPT. You might think I know all the catch phrases or at least that I watched ‘Friends’. But in fact I have no context, or very little. I don’t know who the ‘friends’ are, nor what they do and I certainly don’t know what all that stuff about lobsters is about. But you might think I do, and you might thing the AI does too.
Well, consider the lobster as we consider when, where and how to use these new technologies which truly do hold potential to finally, at last, automate some of the otherwise manual work with content that KM often demands in order to achieve its goals.
As a footnote: Life is always stranger than strange and the truth isn’t out there (that’s a series I did watch, by the way). So the lobster thing is supposed to be about having a friend or mate for life, based on the idea that lobsters do that. But when you Google it you’ll also find a controversy about whether the premise is really true or not (Friends viewers will take it as fact to their graves, I’m sure – but is it?). And you’ll find out it’s also a very ancient insult to call someone a lobster. I had no idea how risky it was!
AI4KM - Robert Taylor 