Conversational thinking - How to design gen AI-enabled services

Beep. Beep. 

You're a submarine captain 50 meters below the surface in the Baltic Sea. Suddenly, you receive a notification from your onboard AI assistant that something is wrong.

The oxygen recirculation system onboard has a malfunction. Speaking to your headset, you ask the system a few follow-up questions, and you're able to troubleshoot the issue in one of the filters. 

The old paradigm of designing digital services 

For most of my career, we've designed digital services that people use with graphical user interfaces. Under the hood, developers have programmed specific orders to the service with code.

Continuing our submarine example, you would go through a data dashboard to find an issue and browse an extensive manual to find a possible solution. 

What is conversational thinking? 

Flash forward to today. With the emergence of generative AI, we'll have to switch our mental model completely.

Enter conversational thinking.

Conversational thinking means that for many non-trivial problems, people can solve them in a conversation with an intelligent assistant.

It's the conversation - not a visual abstraction called the UI - that will help us unlock value from data.

Conversational thinking also means looking for opportunities where intelligent AI assistants can help solve problems by giving people instant access to relevant knowledge. 

Conversational thinking in the real world 

Here are some examples of conversations in action:

1. Unlocking internal knowledge

You're a wealth advisor at Morgan Stanley with a tricky customer question. Instead of sifting painstakingly through your intranet, you'll ask the question from your company's chatbot.

Your friendly AI assistant has been studying all 100,000+ of your company's unorganized PDFs and will pull just the correct information - paired with all the world knowledge and reasoning of a modern LLM.

2. Insights from feedback

Valuable insights often hide in mountains of unstructured feedback from customers. The feedback can be in social media comments, online reviews, or customer messages.

Before, leadership could do a crude semantic analysis of the most repeated words in such feedback.

With gen AI, companies like Viable help decision-makers to talk to the data directly, like "What are our top 5 complaints?" or "How could we improve our deliveries?". We can get the business value directly through a conversation without putting it together from graphs and tables. 

3. Office renovation with a conversation

Say you want to renovate your office space. Instead of browsing different office setups, you share a picture of your space with an AI assistant and have a short conversation about your work.

With a short conversation, an AI assistant can give specific recommendations on the office setup that best fits your life and home.

Look for conversational opportunities

Take a look at any meaningful problems for your customers or employees.

How could a conversation with an intelligent AI assistant make it easier? Is there some knowledge or insight that could be helpful but hiding inside internal filing cabinets? 

Could all of your employees and customers benefit from direct conversation with the best expert on the topic at hand?

Companies that I've worked with are just starting to experiment with conversational thinking that can unlock mind-bending amounts of value in almost any area.

Only some things will be a conversation in the future. We'll still prefer to browse and click for some things - especially when we need visual inspiration or if the task is dead simple.

The key is to look for these conversational opportunities around us.

What next for AI conversations? 

Two new capabilities are making conversations even more interesting: multimodality and agency.

Models like GPT-4V can already chat with images - both receiving and generating. Soon, gen AI models will improve at taking action based on the conversation - like changing your website according to the feedback or guiding your submarine.

Next time you think about a thorny issue in your business, apply conversational thinking.

How could a conversation with an intelligent assistant help your customers or employees? It could be as dramatic as an oxygen issue on a submarine or more ordinary, like finding the most ergonomic office chair for your space.