ChatGPT, GPT-3 and AI are changing marketing: Who should fear for their job and what chance is there?
How important are data and artificial intelligence in marketing today?
That’s an exciting question that I have to answer separately. With data it’s like with New Year’s resolutions, for years every marketer has been talking about it and very few use it. If you look away from the usual advertising metrics like CPM (cost per mille), CTR (click through rate) and CPC (cost per click), things look really lousy. Even the simplest numbers about website visitors and content cannot be answered ad hoc and if there are numbers, they are scattered around in various Excel files. Sometimes some of them are copied into presentations to justify budgets and to give the impression of at least quantitative growth. Real qualitative data about marketing efforts and their contribution to business success is unfortunately super rare. The second part of the question is where it gets exciting. While artificial intelligence was a topic for specialists, nerds and science fiction fans until a few months ago, this has radically changed with the launch of chatGPT. The new pure text-based communication possibility with a Large Langue Model like GPT-3 aka AI enabled millions of people to access AI within a very short time. The results have been equally human, from AI replacing us all to being derided as a useless toy by a US startup runs the gamut.
As great as access is, the supposed half-knowledge it generates is dangerous. The year 2023 will certainly be the year of AI’s launch for companies in the applied sense. But for many, the application has yet to be defined. chatGPT or GPT-3 is like a brain in a test tube, has seen and learned a lot, but has no connection to the outside world, which some users don’t seem to have noticed yet. Applying AI to enterprise data and processes will be both the challenge and great opportunity in the near future.
ChatGPT shows the potential for automation to go much further in communications than we have seen to date. Are jobs in PR and marketing now at risk?
ChatGPT and GPT-3, currently the best-known model, are far from the only groundbreaking AI applications. Google’s Deepmind or Meta’s (Facebook) AI Studio have an excellent reputation and have made incredible releases in the last year. I would replace automation with assistance systems and expand PR and marketing to include all cognitive activities. Let’s stay with marketing, because here it is already very clear to some how the industry will change. If we assume that AI solutions will not replace humans, but support them, i.e. can increase productivity incredibly, then we can also put aside the fear of the workplace. At a time when the growth of most companies is prevented or slowed by a shortage of workers, I need to ask myself much more the question of whether I can at least somewhat alleviate this shortage through productivity-enhancing AI measures. Transcribing, creating, and converting text is not, as far as I can tell, most people’s favorite work, and given that we’re starting to see an incredible democratization of media creation, hopefully it’s becoming increasingly clear what that means economically. Anyone, regardless of skill, expertise, or means, can, simply put, produce media with text commands in ways that were previously unthinkable. The quality is still determined by the human being, in the future only much more in that he can formulate the right questions and tasks (prompt engineering) and then edit something. But for a good blog post you need then no longer 7 hours but only 1 hour. Image creation is already close to text creation, as some applications have shown well. Video will be possible in the near future, along with music production through simple text entry.
What new career paths are emerging from data and AI in communications?
Here, of course, everything around AI topics itself instinctively comes from the Machine Learning experts, Data Scientists, developers and so on. But I personally believe that many new careers will emerge here at the intersection of AI and business. The right understanding of AI solutions, why there also needs to be more trustworthy AI, how to ask the right questions and how to apply them will be whole new skills in the job market. This is also one of the great opportunities we have in Austria and Europe. Of course, we must not give up the race in research and development of large-scale AI models and must do everything we can to ensure that the gap to the US and China does not widen. But we also have the chance to get involved as quickly as possible in building up expertise in applied AI and to conduct research and development into how we can combine these large models with our very own capabilities in industrial production and business services and gain a competitive advantage through early application, and not just in communications or media production. In communications, there are already easy-to-apply capabilities that just need to be applied and not messed around with ChatGPT.
What tools and possibilities should you familiarize yourself with now?
Here there are already many tools on the market, the biggest danger is “hopping” in the application, where you try new possibilities in a daily rhythm and at the end of the day you don’t understand and master any tool so far to use it in your daily work. The best way, in my opinion, is to define a use case for your work and then implement it with the tools you need to do it. Even though there are many, market leaders have already emerged in most areas. In the text area, you can achieve really great results with chatGPT if you also learn to define the right use case precisely. With Neuroflash there is also already a German provider in the text domain, DeepL, my favorite European AI company (after 506) has released a beta for rewriting text. Dall-e2, Stabel Diffusion, runnwayML, to name a few, releasing new features in image and video every day. Add to that specialized solutions like Synthesia, which lets you create really great virtual avatars and videos, or Tome, which lets you create presentations from text, and that’s just the beginning.
What contribution can enterprise IT make to this? What does the company’s internal role distribution look like?
That depends entirely on what role IT plays in the company. There are still many companies in which IT is primarily responsible for the digital infrastructure and takes care of end devices, software and internal company applications such as file servers, ERP and e-mail. Where IT is also involved in and understands the processes of the business departments, it has an exciting role to play. In our projects, we often experience that infrastructure IT tends to be blocking and complicating, which prevents innovation, trial and error, and learning. The other area is the engine and catalyst of development, with data strategies and associated processes and systems playing a very important role. Not surprisingly, these are most often found in companies that are continuously successful and have had to reinvent themselves a time or two.
How do you think collaboration with external parties is changing? What role will agencies and service providers play in the future?
We are allowed to work for partners such as agencies and internal departments of customers alike, which gives us a good insight into the development. The big trend of insourcing has slowed down, in my opinion, because companies are also no longer able to bring enough experiential know-how in-house. At the same time, complexity has continued to increase and there should be such “wonderwuzzis”, as they are sometimes described in job ads, who know several subject areas inside out. The lead agencies that do everything from new creative ideas to websites to PowerPoint presentations have long since changed. At the successful ones, a development has begun that I personally consider the most sustainable. There, the marketing department has split into a data area and a content area. The data teams are responsible for collecting and evaluating the relevant data and then making it available to the content and sales teams. This applies to all types of data, i.e. paid media as well as websites or internal company data. The content teams, in turn, get all the analytics they need to produce the content for marketing and sales in the best possible way for the respective channels and applications. These companies have already begun to leverage the new capabilities of marketing data science and are also the first to actually deploy AI solutions with demonstrable business success.
What role does 506 play in the AI and Data Ecosystem of CMOs and CIOs?
Thanks to our 506 AI-based ability to analyze first-party data and understand for the first time what visitors and customers really want, we are seeing rapidly growing interest in our AI-based customer intelligence solution. In fact, Gartner.com’s recent study predicted that by 2025, companies using AI across all marketing functions will shift 75% of their operational activities from production to analytics. The fact that I can also use consumed content such as text, audio and video from anonymous users to analyze my marketing, sales and product success is completely new territory for many companies. People are familiar with CRM systems with personal, mostly rule-based data and perhaps anonymous website analytics data. Combining this data and extending it with the complete content component opens up completely new possibilities. It was also important to us that the operation can be learned quickly and does not require any special knowledge in order to achieve an immediate benefit for the company. We still live in a strange world where 99% of all budgets and human resources are used exclusively for content definition and production, and only a marginal part for analyzing and verifying what of it really works and what my customers really want. The AI wave will also bring analytics more to the forefront in the coming months, and with it the problems that some have not yet solved. We already had an AWS-funded research project on AI trustworthiness 2 years ago.
2023 will be the year of applied artificial intelligence in AI.
