Creativity Ex Machina

“Hey there, marketers! 🚀 Ever heard of Generative AI? No worries if you haven’t – it’s a game-changer in marketing. In this blog post, we’ll explore Generative AI and break it down for non-techy folks like us. Generative AI is like having a 24/7 creative genius. It powers AI-generated content that feels human-made, from images to text. It’s your secret weapon to supercharge content and engagement. No tech jargon here. We’ll cover the basics, real-world uses, and mind-blowing examples. After reading, you’ll be ready to harness Generative AI like a pro. So, let’s dive into this AI adventure! 🤖🚀“ The paragraph above wasn’t written by a human, but generated through ChatGPT 3.5. In my prompt, I specified I wanted an informal style, and that the blog is aimed at marketers without in-depth technical knowledge. I think ChatGPT succeeded in its tone, but this text strikes me at overly optimistic when compared to the daily reality of Generative AI. For instance, I will show you some examples below that are certainly mind-blowing … but not in a good way. First things first though:

How does Generative AI work?

Generative AI is a relatively new form of Artificial Intelligence. It is learning how to communicate like a human, using Large Language models (LLMs). The artificial brain mimics human creativity and can produce text, images, video and music. The crucial thing about it is that AI can process an enormous amount of information is a short time, much easier than the human brain can. Processing such large amounts of data has been made possible by a formidable increase in computing power. This has its drawbacks, mainly the fact that it uses a lot of energy. Another problem with Generative AI, is the lack of transparency about the algorithms and data used to train the models. This largely takes place in what we refer to as a black box. Use of unreliable data and faulty algorithms can result in bias and hallucination.

How useful are Generative AI tools?

New tools that can produce text, illustrations, code and even music at the drop of a hat appear on a daily basis. A dream come true for every content manager! Or is it? When you first generate a text in one of these tools, you may marvel at the speed at which they produce something that would perhaps take you half a day. AI delivers in mere minutes, sometimes even seconds. But when you cast a critical eye over that text, you might notice it’s full of repetetive sentences and bad logic. Sometimes the information it is based on is outdated. Any AI-generated text needs to be thoroughly checked by a human before publication, to prevent gaffes like this and this. For now, Generative AI tools seem to be mostly useful for routine jobs like product descriptions, SEO or text that’s bound by strict syntax rules like code. The more complicated prompts too often result in low-quality texts of dubious veracity. That’s ok though, the time you save on routine jobs you can delegate to AI, you can use to write brilliantly creative and original blogs and newsletters. In conclusion, I wouldn’t trust AI (yet) to create ready to publish content. But an AI generated text can help you find an original angle, or information about a subject you know little about. You can ask one or more AI tools to write a concept and go from there. Adjust your prompts until you get something you can use (and don’t forget to check if the information is correct).

Body horror and copyright issues

You can also ask AI to create images. There are multiple tools for this, both free and paid. Trying to generate images with AI often results in body horror. The example on the right was generated in Adobe Photoshop Bèta from the prompt ‘yoga pose’. Click the image for a larger version (if you dare). And what about copyright? According to the experts, there can be no copyright on AI generated images. On the other hand, artists and photographers are protesting the fact that their images are being used to train the models without their permission. Sometimes this results in notable similarities with original work, including adding the copyright watermark news agencies use in photos. More nightmare stuff: some AI generated videos (don’t watch this right before you go to sleep). However, there are also more useful video tools available, that create perfectly innocent animations you can use in explanation videos.

Generative AI for CRM, Sales and Marketing

Generative AI can be integrated in a CRM, sales and/or marketing platform. Salesforce has been using Artificial Intelligence for a while now. Under the trademark name Einstein, it’s built into all their clouds and applications, providing analyses and predictions. EinsteinGPT, now in pilot phase, is the next step. This AI can provide engaging subject lines and personalised content for your emails, summarize your conversations and quickly look up answers you can use real-time during chats. Salesforce uses the same OpenAI technology that’s behind ChatGPT, but created its own AI models. They make it possible to use natural language prompts to generate content, and write automations, code and queries for the polatform. Salesforce spends a lot of effort to make sure its AI adheres to their ethical guidelines and security mechanisms. Within the platform, you work with your own, secure data and algorithms that are proprietary for the platform. This is much more transparent than any online Generative AI tool.

Legislation is underway

To limit the negative impact of AI and risks of bias and hallucination for its citizens, the EU is currently preparing legislation in the form of the EU AI Act. Although this is a lengthy process (like with the GDPR), it’s a good idea to anticipate on this future law when using Generative AI (and AI in general) in marketing. This will also help to maintain customer trust. Would you like to know about the possibilities of Generative AI for your marketing and sales? Contact us and we’ll tell you more about it.