AI and Aldi ideas

When people talk about AI they often talk about using machines for what machines are good at and letting humans do what humans are good at.? That sounds like good advice but when you look into it, it doesn’t actually help.

The problem is that it isn’t yet clear what AI is good at.? The academic and writer Ethan Mollick described AI as having a ‘jagged technological frontier’.? What he meant was that AI is surprisingly good at some things and bad at others – and that line is not always clear.? And, as it is constantly evolving, we are finding out more about its weaknesses as well as its strengths.?

AI is an emergent, bottom-up technology so there is only so much you theorise about it when even its designers don’t fully understand how their LLMs work.? The only way to find out about AI is to experiment and use it - so that is what I’ve been doing.

In one experiment I’d been exploring the emotions people experience when pitching and I created a game based on Douglas Adams’ ‘Book of Liff.” to help give meaning to the specific emotions of pitching.?

The broad idea is that Adams took names of little-known English towns and matched them to concepts that previously didn't have a meaning. This helps create a new emotional vocabulary and I was looking to illustrate my thinking with some additional examples, so I turned to Chat GPT.

I was pretty pleased with the results.? Chat GPT did a good job of understanding the basic rules of the game and creating some good ideas around situations and emotions you might experience in a pitch.

At first glance it was impressive.

But it didn't always use the rules. I gave it instructions to 'use town or village names in the UK that don't currently have a meaning.' but it still used Dull, Beer, Crackpot, Scratchy Bottom and Little Snoring as examples.

But my biggest observation was the ideas were a bit of a rip off.? Not a complete 100% steal but if you knew what you were looking at you could tell its heritage, For example, it came up with this definition:

Throcking?– The surge of adrenaline when you realize your computer has frozen mid-presentation, and you’re about to wing it.

This is a great definition.? Everyone will be familiar with the feeling of their computer packing up on them and having to somehow muddle through a presentation.? But, if you are familiar with ‘The book of Liff’, you might remember Douglas Adam’s original description:

Throcking - The action of continually pushing down the lever on a pop-up toaster in the hope that you will thereby get it to understand that you want it to toast something.

It is interesting to note that the original definition is also concerned with being let down by technology. But this definition contains a little more specificity –continually pressing down the lever on a toaster will be familiar to everyone - and the real genius is to link it to the word Throcking.? It has such a wonderful onomatopoeic feel to it - It sounds like the action it is describing.

It is such a brilliant description that is why it has stuck in my head for 30 years.

So, what do you think of the Chat GPT version now you have seen the original?? Is it as good?? Are you still impressed? Do you think they are related/ a rip off?

When I read it, it raised a smile.? I thought the observation about the computer freezing was amusing but most of all I remembered how brilliant the original was.

It reminded me of Aldi brands.? Those quasi- rip off designs that look like more iconic brands but are just different enough to not infringe copyright.

The ideas Chat GPT produced were variations on a theme.? They were executional thoughts based on an original idea but were different enough to not be straight rip offs.?

But the AI missed some of the specific context and richness of the examples others shared. For example David Meikle suggested this definition:

Bournemouth - the feeling of intense relief that a pitch presentation is over, similar to the sense of relief and achievement of managing to file your self-assessment.

This creates a richer image by linking the sense of relief to completing another unpleasant chore – your tax return.? It also shows a more general understanding of people’s emotional repertoire. It takes you out of the world of pitching and makes you consider the broader human condition.

This is not to say that what AI produced wasn’t impressive – and it did so quicker and cheaper than a human would have done.? But what does this experiment tell us about what humans did better?? What are the human elements that AI lacked??

I would say there are broadly 3 elements that stand out for me.

1.???? The specific – and often very personal - nuances of human behaviour that add colour and relevance to an idea.

2.???? A more general view of the world and how different experience link together and relate to each other to provide a different perspective.

3.???? The creative genius and ability to bring together seemingly unconnected elements - obscure place names with feelings and situations that didn't yet have a name – to create something new and unique.

What are your experiences of using AI and what are your observations on the specific strengths of people vs machines?