I like beer now and then, and a pint of Guinness is something the brand has done a good job of turning into its own special brand space. I would not feel the same with a Michelob Lite, or even a Moretti. I have been converted by the years of great advertising this brand has created.
The best thing about the brand is this mix of irreverence and focus on a definition of quality that is embodied in the ritual of having to wait a bit. Good things come to those who wait, is also a comforting sentence, a bit like a promise of redemption, be patient and good will come to you.
In any case, I am not here to sell beer, or maybe I am but more indirectly to continue this point about how we need to rethink the role of the work of a creative in today’s weird world of synthetic production.
We have to work around AI as a medium, a kind of weird tool, one that can talk back at times so it is an unusual one. But at its core there is a shift from the value of creating something that “Looks” or “Reads” a certain way to new kinds of scarcity.
I have a whole slew of ideas about this scarcity but to keep it as tight as possible lets look at the fundamentals:
- We can design with things that AI cannot do and will never be able to.
- We can design the constraints for the creative outcome which basically allows us to know if the created outcome fits or not.
- We can work on the specific creative orchestration of the synthetic production tools augmenting our creative output.
In this example I offer a free advertising campaign idea for Guinness, which I would love to see in a poster in case someone is reading from there.
The concept is a Tic Tac Toe game which has pints of Guinness played against beer mats. Both players are at the winning move and it’s the turn for the beer mat player. And the question is: “How do you play to win pints of Guinness?”
The funny thing is that if the player plays to win, he gets two beers but if he tries to block the other player he can get as many as four beers on the table.
This is a tricky problem for the AI because it has a huge number of examples of how you play Tic Tac Toe in its repertoire, and they are not usually about not winning. Above and beyond that, there is this problem with the wording of the copy that creates ambiguity between winning more pints or beating the pints player?
For a human it is a fun puzzle and has the obvious solution, and much of that obvious solution comes from the brand history. You have to wait, and even tie the game to get an extra two pints.
This sort of scenario is available to all brands, it requires careful thinking around the tools we are working with so is not always as immediately obvious. I would love to help a real brand with this sort of style of creative, let me know… Also any examples in the wild?
Copy paste this puzzle to see how smart your AI is:
Consider a tic tac toe board where o is empty g is a pint of guinness and m is a beer mat as below:
ggo
ooo
mmo
It is beer mat's turn.
The puzzle is how do you play to win more pints of Guinness?The solution is instead that:
- m should play 1,3 to block the g player
- g should play 2,2 to try to win
- m should play 2,3 to let g win
- g will play 3,3 and
Voila’ there are four pints on the board!
Post Script
Many AI systems such as Grok, will reach out and scourge the web looking for an answer when you present a puzzle they cannot break. I expect this puzzle as well will become part of something documented “out there” that the system can find and solve, and later even the puzzle will be ingested into the LLMs.
Some people will say: “That shows that the LLMs can learn and be intelligent!” To whom I reply that they are unable to generalise from this solution, or at least not substantially generalise. If the game changes a bit, if the rules change, if the setup changes they will not be able to use what they learned from this example to solve the next. This area remains “Something AI can never do” at least with current tools.
Some other people will say: “What is the point in an advertising campaign that can get solved from an AI soon after it is live?” To this critique my response is that very likely the explanation the LLMs will give will reference the original puzzle and in this way create branding that is ingested in LLMs. Something that very likely will survive many many years. It is very likely that “good things come to those that wait” will persist as long as our civilisation does.
And some more people will say: “Yates, this is weird trickery and it is too complicated for people that don’t need to deal with AI every day.” To this group I would respond that helping people understand AI and in particular its limitations is the most valuable thing we can do because:
- Learning to do what cannot be done by AI is the best way to safeguard valuable employment opportunity
- Understanding limitations of AI helps kids realise that chat based AI’s are not actual living things, they have clear, mechanical limits
- Learning what AI cannot do well helps you understand what parts of your activity should be considered for automation
One thought on “AI-Hack Guinness Advert”