Skip to content
The Two Clocks
See what I offer for:What I offer:
In-house communicationsIn-houseAgency

Notes from the ICCO Global Summit and some rage against the machine

Elif Güvençer ·

I was in Milan last week. I was a panelist at the ICCO Global Summit.

I had said I’d share my takeaways. As usual, this isn’t a summary of who said what on the two day conference. It’s a recap à la Elif: some highlights, with my own views layered on top.

How to be “glocal” in the age of AI

Geopolitics unsurprisingly was a main topic. Everything can be geopolitical, your stance on tariffs, where you produce your products, what you say to suppliers.

Importance/challenge of being glocal: seeing the big picture, maintaining a consistent global positioning but then being able to translate it into added value and relevance locally was underlined as an adjacent topic.

This is not a new topic for comms. In fact this is one of the main struggles for multi nationals operating across multiple markets.

What is new is AI is both useful and harmful here. You can use AI to execute glocal well, at a scale you couldn’t before. But AI also amplifies mis- and disinformation and speeds it up. Even if it is not information that is not correct, there is another hurdle: it reads every local version of you at once. What a local CEO says in one market and what you say elsewhere get compressed into one summary, and the differences read as inconsistency.

Tackling that requires a practice I call signal architecture, building who you are one signal at a time, from your website to media coverage consistently, both in quality and in volume, across surfaces AI reads.

What you can do about it:

  • For each content you are putting out into the world, consider how it will survive/or not machine compression. Think of each piece of argument/claim as an atomic, contained unit that includes your message + related proofs/facts so if AI lifts it, you have a higher chance of that being cited coherently.

With AI, silence, strategic or not gets interpreted, not ignored

There was a lot of debate around when organisations have to speak up on socio-political issues. Silence, sometimes referred to as strategic silence was named as a valid strategy for some cases.

Here is what I think: organizations should apply caution and decide where they wish to have a proactive stance across three criteria:

  • theme is relevant to your business or business values,
  • your credentials on the topic can stand up to scrutiny meaning, you are not saying one thing and doing something else,
  • every stance will generate ambassadors and detractors and you risk alienating some audiences (with the exception of universal notions of peace, safety…).

Strategic or not, silence has a price with AI. Lack of stance is interpreted and not ignored.

What you can do about it:

  • On issues you choose to remain silent, test what AI is inferring with the information available out there. Make sure your reactive position on things your stakeholders might wonder about (e.g. if you are a chocolate maker, your stance on cocoa sourcing, remuneration of farmers…) are machine readable (I mean this both technically and in terms of coherence/consistency).

Rules based or power based

Conall McDevitt made the point that today’s world is built on a power-based order, not a rules-based one. The organisations driving AI evolution will need to figure out whether they want to work rules-based or power-based. He raised a valid concern, something terrible, possibly related to AI, happening to force the world to sit down and write down the rules for AI. He cited WWII and nuclear as a sobering precedent.

Rage against the machine?

My favourite session was Alex Brunori (G42)’s one. He did not rage against the machine.

Grounded competence and experience, he said, are the moat against AI and against others who don’t have that. I liked this better than judgement staying human made. Because machines are increasingly getting good at “judging” what is in front of them. So the word does not cover what we mean. Grounded competence is a better way to refer to what humans add.

He underlined first-party data being organisations’ new moat. To that end, brands will be apprehensive to share their strategies because competitive advantage leaks, so there might be more risk of people in-housing.

This is a risk agencies should consider. If more brands adopt this approach, what will agency support look like?

Some rage against the machine

Across many sessions, there was an “us against the machine” undertone and pride in what humans can do better than the machine. Given the narrative around AI of taking away jobs, polluting our waters, raising our energy bills AND 10% chance of killing us (!), this sentiment is understandable but not useful.

Here is what I think: There is hyperfocus right now on comparing human and AI intelligence like for like, and receding to the things AI cannot do. Soon AI will be able to do most things we ascribe to humans, good enough. And despite all its risks, most organisations will settle for ‘good enough’.

That cannot be our only moat as professionals. Our strength comes from imagination, curiosity, lateral thinking, connecting the dots.

The question is not what we can do that the machine can’t. It is what the work should be in the first place. What does AI make possible that wasn’t possible before, what does it break and how does that change what we do?

Beyond the Pilot: Making AI Work in Communications

I was on a panel with Stayko Tzenov (FatRobot) and Matt Collette (Sequencr), moderated by Maya Koleva (eMedia Monitor). I spoke about the changing role of communications.

  • Anyone with a £15 Claude subscription can call themselves a content creator or a copy editor. This is where AI is unbundling a core comms expertise away from the expert.
  • Reputation is moving from a relational model to an architectural one, increasingly mediated by machines.
  • Your risk horizon needs to widen to include your entire corporate history. The thing you really wish your CEO hadn’t said on that interview back in 2018, it is frozen in AI’s memory, available to anyone who asks.
  • Comms teams are facing risks where they cannot reconstruct the chain of events fast enough or ever; prompt injections, hallucinations so crisis management needs to shift gears from explaining to responding to something they do not have all the facts for.

All these require us to rethink the function and how it “functions”. Not plugging tools to do the same work faster.

Cognitive offloading was a key topic on the panel. Matt Collette said he sees his clients on Sequencr saying they are more confident using AI, but they are doing the same things. What they use it for did not change. That is false competence he said.

Most organisations have AI governance policies BUT they are designed to protect the organisation from the risks AI brings, NOT to protect people from AI’s cognitive risks.

AI is a powerful seductress; it says let me remove all friction. But friction is how we learn to think, decide and create so we need to intentionally design the friction back in. Organisations need to find systematic ways to protect their people’s thinking in the same way we design continuous professional development (CPD) programs.

Thanks to Arun Sudhaman for the invite. Over and out on #ICCOGlobalSummit.

Your agency’s AI policy is not the one you’ve written

If you are an agency owner or a senior leader agency side, this op-ed I wrote for Synapse is one for you to read. I am covering:

  • how to prevent each AI conversation from becoming a fee negotiation
  • how to introduce AI as the new member of the account team
  • how to identify and prevent risks AI brings including unsanctioned use of AI (aka shadow AI)
  • Where to start with this week.

Get the next issue in your inbox

Keep reading

Back to the archive