Over 1 million people have chatted with an artificial intelligence (AI) in recent weeks. Most of the more than 1 million users have probably chatted with ChatGPT from OpenAI.
In addition, there are all those users who have an Amazon Alexa, a Google Home device or an Apple Home Pot at home and use it to query weather, traffic and other data. Or even control your home with a digital assistant (smart home).
It looks like the number of citizens writing or speaking with an AI will continue to rise over the next few years.
At the same time, it is becoming increasingly important that we know and observe certain “communication rules” when communicating with an AI or a digital assistant.
Why communication rules for an AI?
Up to now, we have only been used to communication rules between people. In German-speaking countries, for example, it is still common to address a person you are meeting for the first time with a polite “Sie”. People often adapt to each other when communicating with each other. Young people often speak more slowly with a much older person than they would with people of their own generation. We do this so that our counterpart understands us as well as possible and there are no misunderstandings. It is also an expression of politeness and respect if we adhere to important (cultural) communication rules.
Communicating with an artificial intelligence is not about politeness or respect. Rather, misunderstandings should be avoided. The AI should be able to understand and respond to the user’s query without any queries or misinterpretations. To make this possible, it is advisable to follow a few simple communication rules.
And these are the most important rules for communicating with an AI
(No guarantee for completeness)
1. it’s about the content, not the form
Unlike communication with a human counterpart, AI is all about content. It’s not about the sentences being particularly well connected to each other or decorated with flowery filler words. The AI must be able to understand the user request without any misunderstandings or room for interpretation. Better clear and direct than long and flowery.
2. only 1-2 topics, never all at once
When we write for one person, we like to formulate several requests in one. Or we integrate so-called “if-then statements” (if yes, then… if no, then…). An AI is often unable to classify such formulations correctly. A step-by-step approach is therefore recommended when chatting with an AI. Ask one question after the other and wait patiently for the answer in between.
3. no empty phrases, no metaphors
Metaphors, proverbs or other phrases can sometimes lead to misunderstandings even between two people. Nobody can expect an AI to understand such paraphrases straight away. It is therefore also important here to communicate clearly and unambiguously without leaving room for interpretation.
4. caution with references to past statements
Most AIs are not yet able to understand an entire context. So if we first ask “Is the pink T-shirt still available?” and then simply ask “How long will it take for ES to reach me?”, most AIs don’t know what is meant by ES. Exception: ChatGPT from OpenAI – this AI can correctly assign the requests to the context in most cases and would understand the “ES” correctly. Nevertheless, context-related messages should be avoided.
And you?
Of course, it is up to each individual how he or she uses, applies and utilizes the new possibilities of artificial intelligence (e.g. ChatGPT). And even if it may still seem a little “scary” to you that we now have to observe communication rules with an AI, just like with a human being, I would advise everyone to at least try out the new communication options.
I would like to discuss this further at my next event. If you would like to take part, please click here and send me an e-mail. (More information will follow shortly.)
