29 Apr 2024 | Omar Oakes

Everyone’s suddenly gone quiet about generative AI

Opinion: 100% Media 0% Nonsense

Now is the time to set standards and rules to avoid the many arguments that will ensue unless we get a grip over who owns what when it comes to AI.

Stop using it — now.

That’s the clear instruction from many advertisers when it comes to using or experimenting with generative AI. Brands are issuing either a flat “no AI” directive or are asking agency partners to provide a much more comprehensive set of policies and guarantees over how gen-AI tools are used for marketing and media planning.

This explains why, if you attended our Future of Brands and Future of Audio and Entertainment events this month, you would have found a much more circumspect tone about AI in media and advertising compared with a year ago.

We seem to be in that awkward phase when a new technology is adopted, businesses start playing with it and suddenly realise we need to adopt rules and standards. Those norms will be extremely influential in a creative economy where success will depend on your skill and knowledge in working with AI tools built on large language models (LLMs).

These rules and standards are being created and adopted now.

Agency holding group investment

You may have missed an interesting bit of marketing put out by WPP a year ago.

“We have created our own set of guard-rails,” said WPP’s top lawyer Vicky Brown. “We have our own policies for working with creative AI; and we have a security and privacy charter. We have to get the regulatory issues right on behalf of clients; and we also have to understand clients’ risk appetites.”

Fast forward to January 2024 and CEO Mark Read pledged that the company will invest £300m annually in AI technology. Within days, his counterpart at Publicis Groupe, Arthur Sadoun, made a similar announcement.

And yet several GroupM leaders have confided in recent weeks that many marketers have either asked them to pause or completely stop using AI in any creative or media services on their brand’s account.

That means WPP and Publicis’ £300m-odd annual investment in gen AI can’t all go on hiring “prompt engineers” — a growing job role in which experts navigate trendy LLM tools like ChatGPT. Agency groups will need to devote significant resources into putting systems into place that protect their clients.

Otherwise they are effectively bringing about the Wild West of the internet.

Who owns what?

Thankfully, the UK trade bodies are catching up and can legitimately claim to be world-leading.

Last November, the IPA and Isba launched 12 industry principles that media and creatives services should adopt when using gen AI.

And, earlier this month, Isba quietly put out a meaty add-on to its Media Services Framework, a members’ document that advises advertisers on how to draw up contracts with agencies. It tends to ruffle — the latest version in 2021 told advertisers that they need to pay agencies more.

What is fascinating about Isba’s approach is that it’s clearly meant to provide specific guidance for media agencies, as opposed to creative ones, that are using gen AI.

As I wrote in this column earlier this month, the prospect of creative and media “coming back together” is a trend that is happening within media agencies, not traditional ad agencies. Media agencies have the bigger budgets, the specialist channel planners and the data analysts. Ad agencies generally don’t and are destined to “stay in their lane”.

So the Isba advice, drawn up by its legal advisors The Quarterback, is fascinating not just because it tells you where the industry is heading on gen AI, but it also shows a direct path for how media agencies might use AI to deliver integrated media and creative.

Compare this with Isba’s US counterpart, the Association of National Advertisers, which last June included one cursory line about AI in the latest version of its media services framework.

Shadow AI and IP bunfights

Chief among Isba’s guidelines are rules around using so-called “shadow AI”.

This is when you may use ad-hoc gen AI for project work, such as asking ChatGPT to come up with a meeting agenda, create a social media calendar or suggest headlines and thumbnails for an ad campaign. You know, all the “grunt work” that we’re eagerly hoping to eliminate from our jobs so we can focus on the real work (whatever that is!).

Agencies are reportedly willing to pay six-figure salaries for AI “prompt engineers”, who you can expect to become part of every account pitch very soon.

Broadly, the guidelines insist that advertisers demand a policy from agencies over how their brand and their data are used. Of course, there could be great value in using an open-source AI tool to automate work, but not if it’s compromising their data.

There’s also a growing fight to be had over intellectual property. Take a global FMCG advertiser that works with various agencies in different markets and allows one media agency to experiment with AI on a particular campaign for a single brand.

Let’s say the agency does a fantastic job in using AI to create a media plan by training a model with synthetic data, or creates a social media plan with various prompts, or has a valuable database of discarded prompts (it’s very useful to know what not to do…).

At some point, the brand owner will ask for all of this AI work to be replicated and tailored to its other brands across the world. To which the agency might argue: “But this is using our AI software and personnel — we own this.”

Knife to a gunfight

There is clearly no one-size-fits-all approach to be had here.

We all see the potential that AI has to revolutionise media services and to integrate creativity into automated online processes. Gen AI, whether it’s using something made by Google or Microsoft, or is developed in-house by the likes of WPP and Publicis, should be central to this revolution.

But now is the time to set standards and rules to avoid the many arguments and potential legal wranglings that will ensue unless we get a grip over who owns what. We’ve seen publishers either take legal action or try to retroactively sell their content after their vast archives were scraped by AI to train their LLM tools.

While we should welcome initiatives like the guidance from Isba, that’s a trade body that represents UK advertisers. As usual, when it comes to regulating Big Tech, we’re kidding ourselves if we can stop powerful global companies with piecemeal national policies.

Omar Oakes is editor-in-chief of The Media Leader

100% Media 0% Nonsense is a weekly column about the state of media and advertising. Make sure you sign up to our daily newsletter to get this column first in your inbox every Monday, as well as key updates with what’s happening at The Media Leader and our upcoming events.