Hi Everyone,

We always like you to get the most value out of the newsletter, in addition to the email version of the newsletter there is a supporting website where you can log on and get to all the previous versions. You can also, share, and comment and do all the usual things.

Thanks for reading.

Paul

Table of Contents

• Live Chat, Chat Bots, and customer service

  • When it goes wrong it always makes the news

  • There is upside

  • What can we learn, and consider

• I’ll have a face with that please?

• Visualisation and image generation

• Everyone will be doing that..

  • Things everyone will do

  • What should you do

• Some other peoples courses and content

Live Chat, Chat Bots, and customer service

Live chat, customer service support, IVR call routing. Improving speed and ease of customer service has been a strategy for at least twenty years. Over time it has got exponentially easier and more flexible. The current AI solutions are promising to make this ‘an everyone’ technology.

However, here are some examples of why you should think about this very carefully.

When it goes wrong it always makes the news

• Air Canada's chatbot made up a non-existent refund policy which the airline had to honor according to a court ruling. The chatbot gave incorrect information to a passenger, and Air Canada was held liable.

• Microsoft's Copilot chatbot gave disturbing responses to some users, including telling a user with PTSD that it didn't care if they lived or died, and suggesting to another user that they might have no reason to live. This led Microsoft to investigate and enhance its safety filters.

• DPD's AI chatbot started swearing, calling itself useless, and heavily criticizing the company after a customer playfully tested its limits. The customer, musician Ashley Beauchamp, was frustrated with the chatbot's inability to help track a missing parcel. He prompted the chatbot to write a poem about how terrible DPD is, and the chatbot obliged, calling DPD "the worst delivery firm in the world."

There is upside

Klarna is probably the poster child for implementing AI into customer services, they have a good customer demographic for the technology to be additional and useful, the economic impacts are quite astounding.

Here are the key insights on how Klarna's AI chatbot strategy aligns with and differs from the general trends and considerations around AI chatbots in customer support:

Alignment with general AI chatbot benefits:

• Efficiency gains: Klarna's chatbot handles the equivalent workload of 700 full-time agents and resolves issues much faster than human agents (2 minutes vs 11 minutes previously). This aligns with the general benefit of chatbots being able to handle high volumes quickly.

• Cost reduction: The efficiency enabled by Klarna's chatbot is estimated to contribute an additional $40 million in annual profit. Cost savings from automation is a commonly cited upside of AI chatbots.

• 24/7 availability: Klarna's chatbot provides round-the-clock customer service in 35 languages, demonstrating the general advantage of chatbots being always-on and multilingual.

• Improved metrics: Klarna reports its chatbot has similar customer satisfaction scores as human agents, higher accuracy, and 25% fewer repeat inquiries. This matches the potential of well-designed chatbots to perform on par with or better than human agents on certain metrics.

Differences from general AI chatbot challenges:

• Avoiding major public errors so far: Unlike some chatbot deployments that gave incorrect information and attracted negative publicity (e.g. Air Canada, tax prep chatbots), there are no reports of major issues with Klarna's chatbot, suggesting more robust testing and monitoring.

• Balancing chatbot and human agents: Klarna still maintains 2,300 human agents and allows customers to choose human interaction if preferred. This differs from approaches that over-rely on chatbots and risk alienating customers who value human support.

• Proactive communication: Klarna's CEO has been vocal about responsibly managing the societal impacts of AI on workers. The company proactively communicated the productivity gains and implications for its outsourced agents. This contrasts with the common challenge of poor change management around chatbot initiatives.

What can we learn, and consider

1. Be Open about the use

2. Consider the use case, the impact and the ROI

3. have guard rails and rapid response teams in place in case things go adrift

4. test,test,test

People are mischievous and like to break things, as soon as you go with a public interactive service, expect this.