Chatbots Can Ruin Your Reputation: The Case of Facebook and Lessons to Learn

Chatbots look set to play an important role in marketing and customer service activities for businesses in the coming years.

Research has shown:

• Chatbots are the fastest-growing brand communication channel
• Seven out of ten white-collar workers are expected to interact with some form of chatbot in 2022
• Only 12.8% of consumers reported negative experiences with chatbots in 2021

However, as with any new technology, it's important to be aware of the potential risks of chatbots, particularly when it comes to how you communicate with your audience and people's perceptions of your business.

In the most extreme cases, chatbots could be actively damaging to your brand, as Facebook and its parent company Meta recently discovered.

What happened with the Meta chatbot?

Meta rolled out a new prototype chatbot, named BlenderBot 3, in August 2022 and invited members of the public to test it out. The company said in a blog post that BlenderBot 3 was "capable of searching the internet to chat about virtually any topic", but also stressed the AI had the potential to make "untrue or offensive statements".

When asked by the BBC to share its views on Mark Zuckerberg, the bot said the Meta CEO and founder doesn't care that his company "exploits people for money", adding that this "needs to stop".

The chatbot also said Mr Zuckerberg - who in the past has faced Senate hearings about his company's record on privacy, personal data and foreign interference in elections - did "a terrible job" when testifying before Congress. It added: "Our country is divided and he didn't help that at all."

This isn't the only time BlenderBot 3 has been less than complimentary about the creator of Facebook. Business Insider reported that the chatbot gave a variety of responses when asked what it thought of Mr Zuckerberg, including calling him "creepy and manipulative".

The unpredictable and often controversial statements made by the AI reflect the fact that its responses are based on searches of the internet and it 'learns' by analyzing conversations and opinions expressed online.

Other companies have suffered bad publicity in the past after launching chatbots that develop based on user interaction and feedback. In 2016, Microsoft's Tay system began issuing racist and inflammatory statements just hours after being launched, forcing the company to make a public apology.

Make sure chatbots work for you

So what can businesses learn from the experiences other brands and users have had with chatbots?

A clear lesson to take from Meta's BlenderBot 3 is that conversational AI systems operate on the data and input they receive, so make sure you think carefully about how you program and 'teach' your chatbots.

The rules and information you feed into the system will dictate what comes out in customer conversations. It's crucial, therefore, to ensure your AI assistants are being equipped with the right language, communication styles and tone of voice to align with your overall brand identity.

This is likely to require close cooperation between marketing and technical teams to minimize the risk of faults or anomalies in conversations having a negative impact on the customer experience.

It's also important to think carefully about your reasons for incorporating chatbots into your marketing and customer communications. If your primary goal is to cut costs by reducing the need for human customer service assistants, this could prove to be a false economy.

Making workforce cutbacks and relying too heavily on technology could result in you frustrating and ultimately losing customers, especially if your chatbots haven't been given the proper investment and are prone to errors.

As the recent Meta story shows, chatbots are still very much in the early stages of development, and are a long way off being able to hold the sort of natural customer conversations that human assistants have multiple times every day.

To make the most of AI and other cutting-edge innovations, ensure your focus is on finding the balance between tech-driven efficiency and the uniquely human skills and experience that only your workforce can provide.