Interfaces play a crucial role in enabling interaction between humans and machines. A common observation is that, although the keyboard and screen are the main means of interaction for most communication use cases, they are both very inefficient and impractical. Why?
- Inefficient because the average person types slower than they speak.
- Impractical because we can’t type while doing something else. In comparison, we can safely talk and drive at the same time.
The most natural and practical way to communicate is to talk, which allows us to do something else at the same time. This is why the most promising AI use case seems to be allowing humans to interact with machines by talking without the need for a keyboard and screen.
This seems to be one of the most promising business applications for AI to bring a fully natural and seamless customer interaction with machines using our language, in its spoken or written form.AI is therefore invisible when it works well behind the scenes.
More than a decade ago, the iPhone introduced the virtual keyboard, which made the physical BlackBerry keyboard obsolete. Similarly, the natural language interface powered by AI makes the screen obsolete in some communication use cases: the screen becomes an option.
In this article, you will learn in which situations it is worth using artificial intelligence and in which it is not.
What is a great AI use case?
Any use case where there is interaction in natural language, made possible by AI behind the scenes, is a great use case. Suppose you would like to get a new bike for this summer. Now, talk to your AI agent:
I’d like a Dutch bike, either new or already used but in good condition, with at least 3 speed modes, and a basket. I would like the first price, please compare among the offers online from well-known online shops with numerous strong and good reviews, as well as offline in the nearby stores, within a 5km range from my home. Please call each local store to confirm their offers, prices, and opening hours. Please email me the first 3 best options by tomorrow evening.
Which leads the AI agent to act, specifically to:
- Search for online offers.
- Autonomously search and call local shops in full autonomy: an AI agent making a phone call to a local shop, interacting with a human.
- Consolidate and rank the results.
- Notify the best offers.
As of today, the common online shopping customer experience is not there yet. Sooner or later, this will become a standard, and that’s where the development opportunities are, to gain differentiation, attract, and retain customers.
The consequences are acceptable if the AI outputs are incorrect, regardless of the quality or formulation of the prompt. At the end of the day, you will decide whether to buy this bike or not, among the recommended options.
With this in mind, let’s look at the wrong AI use cases.
What is a wrong AI use case?
If a third-party vendor makes a mistake, the vendor might be obligated to bear the consequences and to provide a fix depending on the type of contract. If AI fails, not only are the chances low that AI could fix a problem it caused in the first place, but additionally, you bear the full responsibility if you configured the AI agent by yourself. Since AI is not a responsible entity, it becomes your fault, unless you externalize the risks, but that’s another topic.
An AI agent might give the impression it has human reasoning capabilities, hence a sense of responsibility, but that’s not true. This is precisely what brings AI into the wrong use cases. Specifically, giving too many responsibilities to an AI system and expecting it to be smart enough to make the right choices are wrong use cases for AI.
Summary
The AI natural language mimics human behavior so greatly, which is both its strength and weakness. Its strength is that it makes it a perfect human–machine interface. Its weakness is that it is tempting to project that it is smart enough to make the right choices responsibly. We are recalled back to reality when it hallucinates.
Let’s start the conversation and let us know how you think AI might shape the future?
Leave a comment