AI-powered technology doesn’t just have to be about refining existing processes, it can also be about creating new ideas and content
Last month I attended Faculty’s 17th Fellowship Demo Day. Young data scientists who are part of the fellowship programme get a few minutes each to walk through a project they have worked on as part of the fellowship.
It was fascinating to get a look inside 12 completely different projects where AI can have such a direct and meaningful impact.
The projects covered a range of activity, including:
- Helping restaurants fine-tune their offering on Just Eat
- Helping the Advertising Standards Authority (ASA) to weed out scam ads
- Helping carers understand better which medications are likely to lead to adverse events for their clients, via the Birdie app
- Help the London Fire Brigade better allocate their staff between fire stations
- Help price optimisation company, PriceFX, to understand exactly how much value they are adding to their customers
The thing that struck me though was that ultimately, all these projects were about efficiency — doing existing things better, quicker and cheaper. This obviously has huge business value.
But none of them were about increasing impact or creating new concepts.
They use AI-based technology and machine learning to do things differently.
“We want to harness it to make things look interesting, push people out of comfort zone. We see AI as a tool to kick-start creative process.
“In our experience, clients often start with efficiency gains and once that’s done they want to know what’s next, ‘what can we do now?’”
(Kerry’s dreamy flowers are at the top of this article, which she created using a Generative Adversarial Network or GAN)
This more creative role is something that AI-based technology is not widely used for.
But could it be? Should it be?
There are a number of ways that machine learning can be a part of a creative process rather than just making things more efficient
- Combining old ideas in new ways (“concept transfer”)
- Inspiring through generating new images/concepts
- Quickly discarding ideas that don’t work to allow more focus on those that do
All of these require a human in the loop to curate and pick the outputs that matter, but all of them could lead to output that would previously have been unthinkable.
Pencil uses one of the world’s most advanced language AI generators to create dozens of copy options
New AI-powered agency start-up, Pencil claim to make “Facebook ads that perform, made by machines in minutes”.
They use an AI-based language generator to create dozens of copy options. Their system then predicts the most effective copy for video captions and then matches it with relevant visual clips to generate finished ads.
Certainly their system comes up with novel and interesting copy and video.
In an early test for toothpaste brand, Closeup, it came up with the line “the goddess in your mouth”. It’s a line which a creative director or client might have ruled out as being a bit, well, odd — but it performed well above average.
It’s this sort of innovation which proves the power of AI to surprise and genuinely create something new, not just refine something old.
In his book, The Artist In The Machine, Arthur I Miller points out that, “Much of the art computers are currently creating is of a sort that has never been seen before or even imagined. It transcends the merely weird to encompass works that we might consider pleasing.”
If AI-based systems can do that for art, they can sure as hell do it for the creative industries.
I would love to hear from others, what are the examples of creative, generative AI that you’ve come across?
If this was interesting, please leave a comment or get in touch
And please comment or share this article if you have enjoyed it as it helps other people find it.
If you’d like to be on my (100% spam-free) mailing list to be updated when a new article is published, you can sign up here.