Open AI’s Sam Altman has admitted his agentic bots keep escaping and he doesn’t know what to do about it while Anthropic’s Dario Amodei admits AI poses an “existential threat” for the whole world. Both companies are preparing $1 trillian IPOs (Anthropic currently loses $8 billion on a $4bn turmnover.) Well follow the money..
Would you trust an ad agency or media buying company to unleash these little devils on a previously unsuspecting world. Even WPP’s Cindy Rose remarked the other day that she had reservations about the way some of her staff used AI.
Apart from enabling more and cheaper production (that is, even more terrible ads and social media posts) it’s hard to see what AI brings to the ad party. Nobody has so far really managed to explain its benefits.
Yet the dangers are obvious. You unleash a bot for, say, Dove, one of WPP’s bigger clients, and it decides it wants to promote something else: AK-47s for example. How do you rein it in?
It’s superficially reassuring for agencies and other to say they’ll keep a tight hold on what they do with the technology but who’s to prevent some over-eager idiot keen for advancement going too far? And nobody really knows yet where that will end up.
Every week we read about about some client, usually in the US, having to make grovelling apologies about, usually, a social media ad that’s offended somebody in that easily offended country. (See McDonald’s below.) As its most damaging this can damage a brand beyond repair.
With AI it won’t just be a reputation that ends in ruins.
2 thoughts on “Is AI really a good idea for agencies?”
I think the concerns raised here are valid, but they also illustrate how narrowly we sometimes define AI in the agency world.
If the question is whether agencies should use AI to generate more copy, imagery and creative output, then concerns about originality, quality, judgement and the erosion of human craft are entirely reasonable. Making more content, faster, isn’t necessarily creating more value.
But that’s only one application of AI.
Agencies create enormous amounts of intelligence as they work: client briefs, research, audience insight, strategy, creative decisions, media performance, campaign results and the experience of people working across different clients, markets and disciplines.
Much of that intelligence is fragmented. Some remains with individual teams or people. Some disappears into presentations and project folders.
And when freelancers and external specialists are involved, valuable thinking may only touch the agency when the finished work is submitted.
There is a very different role for AI here: helping organisations securely connect and make use of what they already know.
That doesn’t mean asking AI to replace the strategist or creative. It means giving those people better context — what has worked before, what hasn’t, what audiences responded to, what was learned in another market and what the organisation collectively knows about a problem.
And importantly, that intelligence can remain proprietary to the agency rather than becoming part of the public information available to everyone else.
Capture the learning from the next piece of work and add it back, and the agency’s knowledge grows. Over time, it compounds.
So perhaps the question isn’t whether AI is a good idea for agencies. It’s where AI should be used — and where human judgement, creativity and strategic thinking become even more valuable because of it.
This is absolutely spot on. An enormous proportion of the use cases at my HoldCo are exactly this. Using AI to scale the collective intelligence within the organisation. And its transformational in many many ways.