"Demystifying AI": Recap Internal Communications Conference 2023

AI moves.

The lively discussions throughout the year were also reflected in the number of registrations for our conference. There were around 150 registrations for the topic of the Internal Communications Conference: "AI in IC" - the interest was huge. The ICC wanted to help lift the veil: "Demystifying Artificial Intelligence" was the motto.

 

How digital is internal communication?

The first part of the ICC once again focussed on facts and figures from internal communication. These were not exclusively shaped by AI - digitalisation was once again the big umbrella topic this year, because without the digital foundations, AI is also unthinkable.

 

156 communication professionals were surveyed in this year's Trend Monitor , most of them from internal communication and some from related communication areas. The socio-demographic study data provides a good numerical representation of managers, employees and gender ratios in the industry and also shows that we are dealing with experts with professional experience - a big plus for the content of the study results.

 

They can be summarised as follows for the question of the importance of digitalisation:

AI - game changer or short-term hype?

And what about AI? Most respondents largely agree that AI is an important trend, with its significance ranking even higher than sustainability and purpose. It is therefore not surprising that there is great curiosity about the topic: significantly more than half of those surveyed have a positive attitude towards the topic of AI in internal communication. And many are also taking their first steps with the new technology. In particular, more than a third of respondents are already eagerly experimenting with AI solutions such as ChatGPT, DeepL or various image generators. The creation of simple texts, translations, summaries - these are the contexts in which AI tools are being used. Some of the companies surveyed already have their own AI-supported tools.

 

AI raises questions

Even if the new tools invite you to try them out, there are still reservations among communications experts when it comes to the legal and ethical aspects. Precisely because the vast majority believe that AI has great potential, these points weigh heavily.

 

In addition, employees in communications departments complain that they have too little expertise in AI and too few resources to develop and deepen this expertise. At the same time, there is a lack of strategic guidelines from management: only if these are set will it be possible to better utilise the potential of AI.

 

The results of the study - presented by Jutta Gawenda and Dorothee Bönsch - confirm what many of us experience in our daily dealings with the new technology. This is precisely why we approached the topic of AI from two sides at the ICC with the following specialist presentations: one on the technology and its possibilities - presented by Patrick Weiser - and one on the legal situation, which was explained by lawyer Stephan Schmidt.

 

Disruption or illusion?

Sounds astonishing, but it's true: AI has been around for several years. The first algorithms date back to the 1980s, chatbots were implemented at the turn of the millennium and finally, around ten years ago, we literally brought amazing AI services into our homes: we're talking about Alexa and Siri. They understood and generated speech - but could only recite knowledge. It is only with generative AI that language understanding reaches human level. Chat GPT is probably the best-known application at the moment.

Modern AI systems work on the basis of large language models in the field of speech or diffusion models in the field of images and audio. They determine the probabilities of answers or results on the basis of an infinite amount of training data.

What AI can do ...

AI is fast, can make great combinations and knows rules better than humans. Nevertheless, in terms of our intelligence, it has no judgement or imagination - the technology lacks soft skills. We can conclude from this which tasks AI could take over from us in the future when it comes to editorial work in internal communication:

 

• Text creation, summaries and translation
• Topic identification and research
• analysis
• editorial planning

 

... and what not

But beware: controlling the AI is a tricky business - formulating "prompts" needs to be learnt. And even if the AI can deliver convincing texts, it is essential to check whether they are correct. Creating image series is not easy either - reproducing identical images proves to be almost impossible. And then there is perhaps the most important hurdle: Most of the publicly available tools are not suitable for sensitive data - this plays a particularly important role in internal communication.

 

What AI could be particularly good at in the future is the personalised display of content. This is because the algorithms can calculate recipient interests and then serve them with content from a wide variety of sources. This form of playout is currently being heavily criticised, but could become manageable in a controlled space - such as that found in AI. If you think this through to the end, completely new usage scenarios will emerge, as content will become customisable - not only thematically, but also in terms of form, complexity and scope.

 

The bottom line: more disruption than illusion

• Kl will ensure greater efficiency and quality in many areas.
• The leaps in development in recent months suggest that there is still a lot to come.
• Kl is penetrating all scientific fields (ChatGPT is just one of the most tangible examples).
• Debates about intellectual property, fake news and software ethics will lead to comprehensive social and cultural changes.
• Access to Kl-supported tools and services for the masses is comparable to the spread of the internet at the turn of the millennium.
• The current Kl is smart, but not intelligent in the human sense.
• As yet, virtually nothing serious can be done "just at the touch of a button".

 

Legal tips on the use of generative AI in companies

The last - eagerly awaited - specialist article by lawyer Stephan Schmidt dealt with the legal issues raised by the use of AI. This is because the uncertainties in dealing with AI tools and potential pitfalls in their use are enormous. The speaker gave a guiding principle: AI is a collective term - from a legal perspective, the specific system and the specific use are always decisive. Or to use an example: the rules for AI in intelligent driver assistance systems are different to those for translation tools.

 

The main legal risks that companies should be aware of when using AI in communication are in the areas of confidentiality, data origin and quality, data security and data protection.

 

AI and copyright

Dealing with copyright is particularly important for all communication professionals: according to Section 2 (2) of the German Copyright Act (UrhG), only works that are based on a personal intellectual creation are eligible for protection. AI cannot create a protectable work - copyrights are therefore not applicable. Exceptions can be a complex generation process that reaches the necessary level of creation or an existing work that is modified with the help of AI.

 

Conversely, AI can certainly infringe copyrights:

 

• if the result is very similar to an existing work and no consent has been obtained.
• The modification of existing works can also infringe copyrights. Various lawsuits are currently pending in the USA, for example against Open AI, Midjourney and Microsoft.