Wednesday, June 19, 2024

Computers Are Learning To Tell Jokes

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

Teaching a computer to tell jokes may seem like a silly exercise. But humor is an effective way to engage with people, and it forces regionalization of marketing concepts and messages. Computers can do a lot of things, but they have, until recently, been unable to execute in ways we believe are creative other than through scripting.

Humor isn’t a purely numerical problem. To create a good joke and tell it effectively you need to understand a lot about a culture—particularly what areas are taboo. You need to understand timing and nuance, and you certainly need to understand what that localized culture currently thinks is funny. On this last, if you aren’t careful, something that might be funny one day could be incredibly upsetting another based on current events.

This is an incredibly difficult task—just ask anyone who writes comedy. The old saying is, “Dying is easy. Comedy is hard.”

If you can program a computer that can create comedy, you have the elements of a system that can create custom, individualized, selling pitches. Effectively, you can automate and create a better outbound telesales system, one that uses an AI and not human salespeople.


One of the big problems with telesales is that to keep the cost of sales down, the jobs are often sent offshore. These folks are focused on volume, not quality, and the call lists are purchased from the lowest cost provider and are generally not of very high quality. Most of the focus is on getting the cost of the operation down and the volume of calls up. It’s a numbers game for the firm executing the effort—and a nightmare for us as these annoying calls often come in at the most annoying times. I don’t think they do the related brands any good either because annoying people in mass can’t possibly do anything for the image of the company.

Even if you have an excellent team, salespeople can become tired and irritable after a few hundred rejections, which forces monitoring and tools that alert on tone or content. These tools are far from perfect, and the end result is that these sales efforts often do more damage to the related brands than they generate unique sales, and they certainly don’t create product advocacy.


Telesales would seem to be an ideal way to use automation, and clearly computerized systems have been applied to this problem for years. The end result is canned pitches that sound like they are coming from machines delivered identically to people from every region, modified only for language (and often those modifications really screw up sentence structure). This increases volume and reduces expense, but it not only doesn’t address the brand damage and annoyance, it might actually increase it.

But if you have a system that can create humor, you have a system that can better emulate a good salesperson and one that that can treat each sales opportunity as unique. It doesn’t get tired, it doesn’t get angry, and it can learn from each caller, changing a bad call list into a powerful advocacy tool over time. This not only keeps costs down, it increases engagement and shouldn’t damage the company’s brand.

We’ve often talked about expert systems like Watson being able to become a super-Siri by scaling a far more capable database based query/response system. But this is the first time I’ve seen this idea applied to an instrumented sales process. This could not only work for sales but for staying connected to customers, doing quick checks to see if everything is ok, or responding automatically to an alerts related to customer satisfaction and loyalty.

The Emergence of the Funny AI

As we move forward with this idea of computers like IBM’s Watson being able to replace people at scale, particularly for highly repetitive and relatively unrewarding jobs like telesales, we see a very different future. From the customers’ perspective, this is a future where we might actually appreciate the telesales call rather than looking at caller ID and dodging it. From a company’s perspective this promises both higher sales and higher customer loyalty.

When we reach a point where computers can create good individualized humor on the fly, we will likely enter a period where nearly all of our telephone interaction with a firm will be through computers. In fact, we might actually begin to prefer the computer solution, particularly if we are getting the call.

Photo courtesy of Shutterstock.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles