Is AI an invaluable partner for marketers or a creativity challenger? Piotr Bombol discusses the future of AI and marketing

fot. mat. prasowe

As the AI solutions continue to evolve dynamically in 2023, the marketing industry is now at a pivotal juncture, facing both a breakthrough and the challenge of adaptation. In this context, we engage in a discussion with Piotr Bombol, an economist, strategist, and expert in digital marketing and AI, about the future of artificial intelligence. We explore whether AI will become a crucial ally for marketers in crafting effective campaigns and delve into the innovative tool, adaily.

How do you assess the current impact of AI solutions on the marketing industry, and which new tools from 2023 have caught your attention the most?

I think the marketing industry, like other creative industries, will undergo significant changes. I often use the phrase that in advertising, there was a time before and after ChatGPT. Previously, it simply wasn’t possible to create such a large amount of content, to analyze and synthesize such vast amounts of sources in such a short time.

However, paradoxically, these changes are happening quite slowly. The marketing industry is quite traditional, even „outdated.” Most work is done the same way it has been for the last 50 years. AI is slowly penetrating agency processes, but the pressure is coming from outside: firstly, from the clients, and secondly, from the holdings, which see AI as an opportunity for greater efficiency.

As for specific tools, it’s definitely those in the creative field. Firstly, textual ones: mainly ChatGPT, but also Jasper, Writesonic. Secondly, graphical, where the most advanced Midjourney leads the way. Thirdly, animation generators, especially Pika Labs and Runway, which alternately introduce top market solutions.

However, I haven’t observed wider adoption of other tools. I suspect that 2024 will be the real turning point in this regard.

And how do you assess the quality of content, ideas, and concepts prepared by AI-based tools? Can we expect in the near future that AI „creations” will become comparable to human ones?

I agree with the view that, for now, AI is not capable of being exceptionally 'creative.’ Can it be in the future? I am not convinced of this. Some experts believe it’s only a matter of time, but I see more challenges. Firstly, the inherent characteristics of AI models (especially LLMs) present certain limitations. Models are neural networks that can 'compartmentalize’ (i.e., adopt a specific context) as they receive instructions from the user (prompts). Humans, however, can jump from context to context very flexibly. Another thing is feeding selected knowledge: language models are based on diverse resources, and it is impossible for them to determine which of these is exceptionally good.

Therefore, from my perspective, the only chance to achieve a better level of creativity is to create dedicated models that receive appropriately selected sources. This is a process we are currently testing in adaily. We have the luxury of having the world’s largest database of recognized advertising campaigns and can test this hypothesis.

But how to test whether AI concepts will be as good as human ones? Perhaps only through 'blind tests’ of advertisements. For example, by showing a client two proposals at a tender, one human, the other 'artificial.’ Or testing such two concepts with real respondents. Otherwise, we will return to the unresolved discussion of whether one idea is better than another.

Looking at it from the perspective of improving quality, rapid AI development, and its takeover of more and more responsibilities, how do you think the role of humans and the human factor will change?

Like many people at the beginning of 2023, I had concerns that the role of humans would be marginalized. However, remembering the earlier challenges, I am now convinced that the human factor will still be needed. One of the greatest values of a human expert is that they can make authorial selections. In the first step, they can assess what is a good reference point (benchmark), in the second – what they need (input), and in the third – whether the answer is sufficiently valuable (output). We can, of course, leave all this to AI, and partly the models are able to take on these roles. For example, ChatGPT will be able to write a prompt for itself for a specific query or evaluate its own proposal. However, this is fraught with the risk of feedback looping. The more we 'loop’ AI in conversations with itself, the more we will tend to regress to the mean.

Of course, I am fully aware that in the end, there will be a business confrontation. Is what AI creates good enough for me, and does it provide concrete savings? If so, I decide to compromise. So, we end up with what is the general market consensus: repetitive intellectual work, with an acceptable average level, with small consequences for errors, will be taken over by AI. The more difficult, specialized, and responsible work will be carried out by human experts. But they will also be supported by various AI tools – that’s certain.

What do you think are currently the biggest challenges, problems, and objections related to the use of AI in the work of marketers?

I mentioned that the industry is, contrary to appearances, quite conservative. Technological innovations on the agency side are not adopted as quickly as on the client side. The only leap in the digital transformation of agencies occurred in 2020 due to forced remote work. Added to this is the overt aversion to artificial intelligence by some in the industry, fueled by fear-mongering propaganda, which can be found especially in industry media. 'AI or someone with AI will replace you at work’ – this is a message built on negative emotions. It is not surprising that some people see AI as their enemy, not an ally.

As if that were not enough, there are also very important ethical and legal issues. Some marketers do not feel comfortable with the fact that AI tools were trained on the works of other authors (which applies to 100% of models) and outputs may contain very strong inspirations or, as some claim, even copies. Legal issues concerning the possibility of using AI content as a work with the transfer of rights to the client, data security placed in AI applications, or the use of queries to train models. There are corporations that fundamentally prohibit their employees from using AI tools until there is clear regulation, for example from the EU (which is already happening under the AI Act).

Does this mean that people in the marketing industry do not use AI? Of course, they do. According to studies, most of them simply do not admit it to their superiors.

Do you think artificial intelligence can in the long term limit creativity, or will it rather serve as a tool to support creative thinking?

It’s difficult to assess the net effect at this moment. I lean towards the option that AI will 'boost’ the creativity of people working in the marketing industry. Already, AI tools, like ChatGPT, are used for acquiring inspiration, conducting virtual 'brainstorms’, suggesting insights. Additionally, we can already use tools that directly relate to real content from the Internet. Here it’s worth mentioning Perplexity, a unique alternative to Google Search. In this tool, we can ask specific questions and get concrete answers prepared by AI based on real content. We approached the creation of our adaily in a similar way.

In terms of AI, you created your own solution called adaily. Could you explain what it is and how it can help marketers in their work?

Of course. Adaily is an AI assistant for people working in marketing. We provide easy and quick access to the best industry knowledge, so you can create better marketing projects yourself. We currently offer free access to browse 2,500 winning advertising campaigns from the last 6 years (including Cannes Lions, Eurobest, Golden Drum, MIXX Awards) and an interactive version of 52 tools from the acclaimed Deck of Brilliance with the consent of its authors. This significantly shortens the time to prepare campaigns – from our observations and conversations, even by tenfold.

How did the idea for creating adaily come about, and what were the main motivations for its development?

I spent the previous 10 years in the advertising industry. I had the pleasure of working both in network agencies (DDB, Deloitte Digital) and running my own agency (Gameset). I gathered a lot of thoughts and observations from this period. Returning to the observations about the 'traditional’ agency work model, I couldn’t reconcile with the fact that so many processes were carried out inefficiently. And worse – distracting people from the essence of work, which is creating and implementing concepts for clients.

Starting work on my tool, I wanted to make AI an ally in work and relieve marketers. And let’s be honest – since 2020, work in the industry has become increasingly demanding.

Additionally, while working on the tool, I realized how unfair the marketing industry is. I understood this best during a conversation with Juggi Ramakrishnan, the creator of Deck of Brilliance. His goal was to provide the best knowledge that network agencies have, also to those 'excluded’ from this circle.

However, that’s not the only barrier. In reality, 90% of the best knowledge is in English. How are people from Poland, Chile, Bulgaria, or Vietnam supposed to fully benefit from it? It’s a very important question. That’s why we also decided to natively support as many languages as possible – currently, it’s already 14.

What kind of data does your solution use and base on? Can users use the generated ideas, inspirations, and content without fearing that they are based on non-existent or low-value information?

Our tool uses only publicly available information. Every time, we also provide links to the sources. This way, we can give a 100% guarantee that the information in adaily is accurate. We wanted to eliminate one of the most annoying features of AI models, which are hallucinations. They do not occur in our tool.

Regarding the use of materials from our app in professional work, we leave it to our users in terms of scope and purpose. They have the freedom here. And I know that this purpose can also be different – it can be a list of benchmarks for a presentation at a client meeting, a marketer checking the quality of what the agency brings, and even as input for a thesis in studies.

How does adaily stand out among other AI tools? Aren’t you afraid that similar solutions may quickly appear on the market, especially as the threshold for creating them decreases with the availability of new advanced tools?

Adaily is based solely on selected industry data and provides access to it in an efficient way. Honestly, I haven’t yet come across a tool that offers what we do. We have either generic content creation tools (often overlays on GPT or Stable Diffusion) or tools for general research (like Perplexity or Bing Chat). We want adaily to be a complementary tool to these solutions, focused on the specific needs of marketers. And from messages from users, I can conclude that we are succeeding.

Regarding potential competition, we are not afraid of it. Mainly because building such a database as we have would require a lot of time and expert knowledge. That’s probably why we have the largest collection of industry knowledge in the world, which I heard from people involved in creating AI solutions for the largest advertising corporations.

What skills should marketers currently develop to effectively use AI-based tools like adaily?

I’ve already hinted that adaily is inspired by the Perplexity solution. What captivates me and 10 million other monthly users worldwide is the fact that Perplexity doesn’t require prompting. Personally, I believe in a future where AI tools offer a 'promptless experience’. After the big boom in prompt engineering, I see that we are gradually moving away from it. And rightly so – people should have industry knowledge and the ability to think critically, not skills in prompting. Just as we don’t expect people using office programs, such as Microsoft Office, to be able to program.

Considering that briefs for campaigns for similar brands can also be very similar, will the ideas and concepts generated by adaily not be very similar for such brands? Will this not lead to very similar campaigns, activities, content?

I’ll answer paradoxically – exactly this has been happening in the market for a long time. Just recall the classic example. In every category, there is already a company that wants to boast that it offers better, more comprehensive services than the competition. And for this purpose, it uses the slogan 'more than…’. And AI was not needed to replicate such an approach on a global scale.

But now seriously. Of course, this 'convergence’ of client briefs (i.e., similar target groups, similar tools, similar needs, etc.) will deepen, and as a result, marketing communication – standardize. However, to not confuse cause and effect: AI will find itself as an efficient solution in this case. Since certain approaches have already been invented and tested, why reinvent something new? Of course, many brands will require (or expect) a very dedicated approach, and there AI may be insufficient.

Before we allow our users to create full marketing concepts in adaily, we will want to thoroughly test this with our users. All solutions are developed with their needs in mind, so their opinion will be the most important here.

What are your further plans for the development of adaily? What new features do you plan to introduce?

We want to continue our mission of democratizing marketing mastery. We will provide more and more recognized advertising campaigns (including those that will appear in 2024), as well as selections of the best reports (on consumers, media, brands) or collections of proven marketing theories. The next step is to increase the interactivity of this content, for example, through analyzing sets of campaigns or receiving specific tips on how to create a strategy for a brand. The final stage is the implementation of marketing activities, and here our intention is to integrate with top tools available on the market.

However, the most important project we are currently working on is creating our own AI model, which will be 'retrained’ (i.e., specialized compared to the usual LLM) on campaigns from our database. I believe that this will bring us closer to creating the best tool on the market for creating initial marketing solution concepts for the specific needs of all types of business.

In one of the interviews, you mentioned that you prepared the first version of adaily by yourself. How do you currently assess the possibilities of creating digital products, such as SaaS, as a solo entrepreneur? Can everyone create such a product, or at least an MVP, thanks to AI-based tools?

Yes, of course. As someone with almost zero programming ability, I was able to create a working prototype – a website, a user registration system, automation of user queries to the API. All this thanks to no-code tools and the availability of great tutorials on YouTube. Plus, of course, a lot of motivation and perseverance to complete such a project. Honestly, if I hadn’t tried to create the prototype myself, I wouldn’t have been able to conduct interviews with users or meet my current partners: Mateusz and Jakub.

I recently had the pleasure of co-creating the AI track program at No-Code Days. I was hugely impressed by the topics submitted by the speakers and the case studies that were ultimately shown at the conference. I believe we are only seeing the prelude to a revolution that will happen in the coming years. For inspiration, I recommend following the world’s most famous solopreneur, Pieter Levels. Working alone, he manages to generate millions of dollars in revenue each month. At the same time, he fully controls his life, lives in different places around the world, and has time for his interests.

Finally, please tell us, what can we expect in 2024 in terms of AI development, and can we predict anything at all with such a pace of its development? Are there any areas where you particularly see opportunities and possibilities for new solutions?

In my opinion, 2024 will be the first proper year of AI adoption. Why? In January and February 2023, it was too early to risk implementing AI. ChatGPT had just appeared, and Midjourney was generating hands with six fingers. However, most businesses operate according to an annual calendar, having new funds for investment from the new year. Today we are in a completely different situation: ChatGPT and Midjourney have exponentially developed in terms of functionality, and thousands of other tools have appeared. Some of them are absolutely groundbreaking, solving unaddressed problems. At this moment, companies cannot ignore the AI trend and must consider implementing this technology to maintain or build a competitive edge.

At the same time, it doesn’t mean that everyone will go through this adoption. Some companies will still block AI due to unclear legal situations. The question is only how long they will be able to withstand external pressure (from the market and competitors) and internal (business optimization for the management). I bet it will be rather short.