Why Responsible AI Adoption By Advertisers Matters

The adoption of artificial intelligence (AI) is likely to be the most revolutionary change the advertising sector has experienced since the advent of the digital banner ad - even more so. Its influence is already widespread, even in its early stages, as marketers work to harness its potential while reinforcing their brand communications. At the same time, they and their agency partners are integrating it into their tech stacks to enhance operational efficiencies.

Despite all the soundbites from business executives aiming to quell concerns about AI's impact on human creativity and employment opportunities, caution is still being sounded around what it could mean for the advertising sector, and wider society too.

Recognising this, the Advertising Association's AI Taskforce has published its first report; 'Advertising and AI: Showcasing Applications and Responsible Use' outlining the context and potential impact while stressing the importance of responsible adoption.

One contributor to the report, Dr Alexandra Dobra-Kiel, innovation and strategy director at Behave, part of the Total Media Group, discusses her views on how well the industry is handling AI and why it's so important that it gets it right.

What does responsible AI, adoption entail when it comes to advertising?

There are two issues here; the first one is that each industry tends to think that there are different ethical standards that should apply to how it uses and develops AI. That's wrong because Ethics is universally valid, regardless of your sector - some safeguards need to be implemented. The second aspect is that we are so fixated on putting these safeguards around AI that we confuse Ethics with regulation. It's not the same. And given that AI is moving so fast, putting hard boundaries around it will stifle our capacity to develop and use it. And then we also need to be reminded that Ethics often involves ethical dilemmas, and we are only becoming aware of ethical dilemmas because they're quite nuanced once we know how to use and think quite well.

What are the main challenges that the advertising industry right now is facing when it comes to responsibly adopting AI?

The first one is really around accountability and transparency. We often talk about this black-box nature of AI that makes the decision-making processes very opaque, and that is true. I mean, we cannot expect everyone to be a guru in AI and to be able to explain from A-to-Z how the algorithm works. But we should at least make sure that we vaguely understand how it works, because then if we don't understand how it works, we are not really able to exercise any critical spirit on the outputs, and we just take them for granted.

The second aspect is about fairness. I think that you have a lot of people who are saying, 'Oh, AI is biased, and it's so dangerous.' I mean, yes, it is biased, but humans are biased as well. Yes, there are fairness issues about AI, but there are fairness issues about humans as well. And AI is not something that would exist in the absence of humans, so we really need to be much more careful in the way we develop and use it so that we become much more aware of our own biases and fairness issues.

There are two more things: the third is around privacy. So a lot of companies are really jumping on this idea of personalisation and hyper-personalisation because it really speaks to our need for customisation as humans/as clients - fair. But then there is a clear privacy intrusion that can emerge.

And the last aspect, which we are nowhere near tackling, is around the sustainability aspect of AI. It takes a huge amount of energy to run AI. And on the other hand, we're all saying that we want to save the planet and reduce emissions, etc. So how do we align these two? Statistics show that one AI model alone can emit 284 tones. CO2 which is five times the lifetime carbon emissions of an average passenger car, so it's huge.

One of the recommendations you’ve made in the report is to focus on the Socratic method of adoption. Can you explain why you think that method is best?

I chose that method because of the nature of AI. It is very complex. It is also evolving very fast, much more fast than most of us as employees and companies can grasp. And so, rather than coming with hardcore rules, regulations that restrict the use and the development and that have almost a punishing aspect to it, the Socratic method makes you inquire AI and have a very inquisitive relationship with how you develop and use AI.

It's really about posing these sorts of thought-provoking questions, then thinking about what assumptions how you would answer with, so that narrows it, and then you open it again, because you are asked to actually think about all the biases that could have gone into how you answer it with questions. And so it is quite an iterative process, but it really obliges you to think and to question yourself and your assumptions.

Whenever I've had conversations with marketers, especially around the adoption of Gen AI, we've spoken about how they need to train the tools to understand and follow their brand's ethos. Does that solve the ethics problem?

Considering that a lot of brands, for instance, when it comes to more ethical aspects, are engaged in greenwashing, etc, training your AI to follow the brand is not a guarantee of safeguarding ethics. That takes quite a static approach to the brand. We know that brands are actually evolving quite fast themselves, and they will need to evolve even faster in the future. So training something on a static view of your brand doesn't seem to make commercial sense.

Where do you see the most impact coming from in terms of the adoption of AI within advertising?

I would hope for the biggest impact would be around creativity and strategy, because this is where this industry really needs to show progress. But again, AI often is only as good as your prompts. It is only as good as you can use it, especially when it comes to creativity and strategy. And so I would expect that we would see the most impact, actually more in; 'Yes, make a new slide, make a new graph', or things like that. But it's quite superficial creativity. What you'd hope is to have a creativity that is much more in-depth, in the sense that it is a creativity that is there to solve issues, rather than just to look good and exciting. For now, this industry seems to hop on what is glamorous and what is glitzy, rather than how can we help solve issues and it's a shame because it is an industry that could really be at the heart of influencing people's emotions, which is what matters to drive change. But it seems to miss that opportunity by going after the shiny new thing.

What impact does AI have when it comes to media agencies?

So obviously, there are a lot of efficiency changes that this can make, not least in terms of how you buy media and all of these programmatics.

But whilst the conversations of most companies seem to be fixated on efficiency, I think that it should also move pivot towards excellence, because if AI is really your co-pilot, then it is also to achieve excellence, not just efficiency. After all, if it is just efficiency, then we will all compete for the same things very quickly, and AI will no longer be a distinctive competitive factor.

Where are you seeing the decision-making coming from on what the ethics of advertising are that must be complied with? Regulation still feels a long way off.

It does feel a long way off. I did a lot of work in my past life with financial services companies - in contrast to the advertising industry, they're heavily regulated, and it has been even more so the case since the financial crisis in 2008. So for them, regulation is core to their operating model or their strategy. They cannot ignore it. Here in the advertising industry, it feels that it's quite an incipient stage. People are scrambling into; "Let's put a task force together. Let's put initiatives together to have a coordinated response." But ultimately, the regulation should come from the government state and regulatory bodies. So there is a question of whether trade bodies that look after our industry should be granted more regulatory powers. It's an open question for the future, but it is about that careful balance between regulation and innovation. I'm very much thinking that with AI, regulation shouldn't stifle, it should permit, it should allow, but it certainly shouldn't restrict, and that is something that we haven't solved yet, and that we are still far away from solving.

The other aspect I would raise, also about why regulation should move in terms of permission-to rather than restriction-from, is that when you look at the regulation of financial services, it has often played out as a cat and mouse game. The cat was the regulator, the mouse was the financial services company. So there are a lot of unintended consequences from restricting, and maybe there are lessons from that industry that our industry could learn in terms of how it envisages regulation.

What would you imagine will be the trigger that will accelerate regulation?

I think that there are a lot of questions around IP. Who ultimately owns IP?

IP is also a big part of our operating models - the more conversations and issues will arise on that, the more likely it says that regulation will need to kick in, because something will go wrong, and not least in terms of how companies can continue to make money out of their IP and there is the whole question about the black box and the transparency we have recently seen. For instance; the riots in the UK. Such civil unrest can also be affected by social media, and social media is part of our industry, so I could imagine that there will be a strong regulatory response if there are not more safe guards put in place by our industry in terms of how the messages are spread, communicated and all of that.

Why does this matter to society that the ad industry gets the adoption of AI right?

It matters because of several factors. The first one is that when you look at outside of this industry, you do have a lot of big consulting companies, Accenture, Deloitte, etc... who have heavily invested in buying advertising companies. And whilst these advertising companies within the Accentures and the Deloitte, you could argue, are still part of our industry, they have also moved into the consulting industry, and they benefit from a level of professionalisation and optimisation that our industry simply doesn't have. So it matters, not least because of a survival factor. If you want to continue to survive and thrive economically as a standalone advertising industry, then you need to look at what your competitors are doing in that consulting space.

What advice would you give to the advertising industry in terms of adopting AI and getting it right?

I would say, really adopt rather than implement. There is a big difference in between the two. Implementing is just that you put all the tools in place and you expect people to use them. Adopting is that you take the people with you and that you understand what the obstacles of the people are. So adopt and then really think about how you can open a new realm of imagination with AI. Really think about the future and have foresight because only then can you achieve excellence rather than efficiency. And then it's really about question, question, question because this is the only way that you have ethics in place.

The ‘Advertising and AI: Showcasing Applications and Responsible Use’ report can be found here.