Designing conversations: principles from ChatGPT to improve human interactions

Dall-E generated art: a computer that is a portal to another beautiful dimension world, digital art

Have you ever encountered a situation where you were held accountable for something which you were never explicitly informed about?

Well, it happened to me, being on both sides of the situation. Whenever this happens and the communicator does not recognize their role in the misunderstanding that this lack of alignment generates, it makes me think — what else is not being communicated, but it is expected? I feel like walking on eggshells, which erodes trust over time.

Being an effective communicator is not a skill we develop overnight. It is also not a trivial one for leaders or designers. It is key to shared understanding, trust, and collaboration in any relationship (be it at work or outside of it).

It requires reflection and intention. And it is okay to fail sometimes. This is a continuous process. When I am the one that fails to communicate well, which I usually see by the outcome of the situation, I recalled a quote from Julie Zhue, in her book Making of a Manager:

“Whenever you find yourself deeply disappointed or disappointing someone else, ask yourself: where did I miss out on setting clear expectations and how might I do better in the future?”

Because even when we explicitly communicate something, it does not mean the message was transmitted. Imagine not communicating at all.

It is like creating self-prophecies, where we think the other person 'should know by now' and leave it to chance for others to guess what we thought. As if everyone has the same background, context, or expectations we have.

Trying to read people’s minds

To avoid this, we can start communicating (or over-communicating if in remote environments). But this is not as simple as it sounds.

This diagram of accurate empathy from the student of the psychologist Carl Rogers, Thomas Gordon, illustrates how this process works:

Diagram from the book “Listening Well: the art of empathic understanding”

BOX A

Meaning: that’s what is happening in the person’s mind and heart at that moment, along with all their background of a lifetime of experience.

BOX B

Spoken words: what the person says, or writes. It has also more involved, such as facial expression, vocal tone, volume, and posture.

BOX C

Heard words: this represents the listener.

BOX D

Meaning (again): that’s the interpretations of the listener about what is being said. “What does the speaker mean?”.

There are several factors where communication can go wrong.

Seriously?

Here are 3 of them based on William R. Miller (an author, Ph.D. in clinical psychology, and professor):

1. Spoken words are a small part of communication: people don’t always say what they mean. The speaker may not be good at putting their meaning into words or might be speaking in a second or third language. Spoken words can come with embarrassment, fear, a desire to make a good impression, or an intention to deceive.
2. What is being said can be hindered by many factors: inattention of the listeners, distance, distraction, hearing impairment, or what is being said in another language.
3. The listener runs an inner dictionary of possible meanings: and from experience assembles an interpretation of what was meant.

Okay, what to do, and what does chatGPT say about it?

As I began using chatGPT and testing different prompts, I wondered if this language model could inspire us to apply similar principles of clear communication in our daily interactions with others.

After all, the more specific and concise our prompts, the more engaging and informative the conversations become.

In the book The Art of ChatGPT Prompting: A Guide to Crafting Clear and Effective Prompts, the author says that what makes a good ChatGPT prompt is:

• Clarity: Avoid using overly complex or ambiguous language, and aim to be as specific as possible in your prompts.
• Focus: Avoid using overly broad or open-ended prompts, which can lead to disjointed or unfocused conversations.
• Relevance: Avoid introducing unrelated topics or tangents that can distract from the main focus of the conversation.

Those are great principles to have in mind during conversations in general.

And here is what chatGPT answered when I asked it about its principles and how we could learn from the way we communicate with it:

• Clarity: When you communicate with me, you have to be clear and concise in your language so that I can understand and respond to your prompts effectively.
• Specificity: When you’re more specific in your requests, you’ll receive more accurate and relevant responses from me.
• Conciseness: As I am a language model, I can process a large amount of text, but being concise in your request or question can help me to understand the context and purpose of your message better and respond in a more efficient way.
• Reading comprehension: Interacting with me can help you develop your reading comprehension skills, as you need to understand my responses to continue the conversation.
• Empathy: When you communicate with me, you need to understand that I am a machine and I don’t have emotions or human experience, so it’s important to be mindful of the language you use and the context of your message.
• Proofreading: As you communicate with me, you learn to proofread your messages before sending them, in order to avoid any misunderstanding or errors.

Those are basic principles, but easily forgotten if we assume that we all share the same context and background. We know this is far from the truth.

As we go deeper into using chatGPT, and crafting the way we communicate with it, we could go further and remember some of these principles in our day-to-day conversations, adapting wherever needed.

Especially because communicating with humans has often more room for interpretation and understanding, considering the context, tone, body language, and emotions.

Different from the current language models, we also give nonverbal cues, which can help to clarify or add meaning to verbal communication.

In conclusion, we can not not communicate, as it is said in the 37signals Guide to Internal Communication:

“You can not not communicate. Not discussing the elephant in the room is communicating. Few things are as important to study, practice, and perfect as clear communication.”

And even when we do communicate, we should aim to make conversations more effective. Observing how chatGPT works and responds to our prompts has been a good practice.

Good communication is not a one-time achievement. It is a continuous learning process. We share, reflect, understand, and iterate.

(And do not forget to use critical thinking and double-check the answers. Just like humans, chatGPT can very confidently give incorrect answers).