AI in communication: how we work with it

Whether texts, videos or audio: generative AI models are changing the way we think about communication and content. Large language models (LLM) and diffusion models, i.e. artificial intelligences that can train themselves using neural networks, are becoming increasingly powerful. Technically, there are various forms and approaches, but the results are usually equally impressive: with the help of almost countless information points and data, which are often collected from the internet, these programmes can generate code, write essays, generate images or imitate voices - with astounding authenticity. They are sometimes so convincing that even experienced software engineers have mistakenly assumed that they have developed a programme with self-awareness.

 

While the question of self-awareness is more of a philosophical digression, most companies are more interested in the practical side of AI: How can we become more productive? Where are the opportunities for us? Where are the risks? Because one thing is clear: software that can take over entire communication channels speeds up work processes and could take over the work of many people. Chatbots have long been widely used in customer service, but now this change could fundamentally change the core of companies.

 

The World Economic Forum expects around one in ten jobs worldwide to be taken over by an AI within five years. The bank Goldman Sachs also expects 300 million jobs to be performed by software in the future. However, both studies come to the conclusion that, overall, more jobs will be created than will be lost.

 

The possibilities of AI for companies are exciting: new tools will change our work processes forever.

Corporate communication with AI:
New technology requires new skills

As a digital and creative agency, RYZE Digital is itself in the midst of this change. We are also looking for ways to use this new technology to improve our processes and develop creative potential more easily. Because AI has many advantages: For example, it can help structure and analyse data, use it to create texts for various channels and even translate them if required. Whether artificial images, self-generated audio or edited video sequences - AI offers a lot of potential in media production.

 

However, it is always noticeable that current AI models are not yet good enough to take over production processes independently. For example, there are problems with the accuracy of texts, as AI can invent facts. The underlying data can also be out of date. And finding the right tone of voice is an abstract process in which even humans often need several rounds of coordination before the finer points are clearly defined. Because as convincing as AI may seem, behind the prompt line is still an algorithm that works with probabilities - and has no real understanding of the content.

 

Working with AI therefore means that skills such as criticising sources and questioning facts and contexts are needed more. After all, a faster work process does not help if it is more error-prone.

 

The creation of content, such as this graphic, can already be done entirely by AI. For abstract planning, however, there is still no way around humans.

What are the advantages of artificial intelligence in
communication?

However, this does not mean that AI cannot play a role in creative tasks, as the mass of data on which the models are based opens up a wide range of other possibilities for creative processes. Artificial intelligence is a sparring partner that can generate text suggestions or image ideas. If the blank page in Word or the white background in Illustrator seem overwhelming rather than liberating, AI can help to find the right start to the project.

 

This can start with research, for example, in which a chatbot such as ChatGPT briefly summarises the most important aspects of a topic, or image generators such as Midjourney create initial sketches for a graphic. Even if there is no "perfect solution" that can be used directly with copy-and-paste, with a little trial and error and the right input, AIs often provide suggestions that, with a little manual editing, hit the nail on the head.

 

There are also constant developments in this area that suggest that in the foreseeable future, artificial intelligence will be able to take over creative tasks completely by itself. Then it will depend much more on what instructions - i.e. prompts - users give the software. There are already add-ons for ChatGPT that are much more competent in the creation of text and can also match tonality with a company's published data.

 

In this case, however, customised solutions are required, as public AI models such as ChatGPT use user input to further develop themselves. This could expose internal company information to the public. Accuracy can also be a problem because, as researchers at Stanford University and UC Berkeley recently discovered, ChatGPT is making more and more errors. It is not yet clear why this is happening, but tests involving code generation, recognising prime numbers and comprehension questions clearly show that the AI is losing performance. Companies should therefore rely on their own models that they can use themselves. In this way, no internal information is leaked and errors can be found and rectified more quickly.

 

AI is suitable for this in areas such as communication:

 

• Automated process control
• Sparring partner for creative tasks
• AI-supported search queries
• Automatic generation of SEO attributes
• content creation
• Automated product descriptions
• Chatbots and consulting tools
• Analyses of data and trends
• Lead generation and scoring