Chevy Chatbots Go Rogue

Welcome to another edition of the best damn newsletter in AI.

This free newsletter is designed to keep you ahead of the curve and open your mind to using AI in your work and business.

Digging deep in AI for work or AI Operations? Take a look at our membership.

Our #1 goal is to be useful. So please shoot us an email 📩 if you have questions or feedback, and especially if you implement something we share!

Here's what we're covering today:

• How a customer service chatbot can go rogue

• Using AI to make your holidays a little more… personalized

• AI making its way into politics

Let’s get to it! 👇

TODAY'S PERSPECTIVE

Chevy Chatbots Go Rogue

Imagine this: You've just installed an AI chatbot on your business website, hoping to streamline customer service.

But instead of helping customers, the chatbot is selling your products for a dollar and recommending your competitors.

Sounds like a nightmare, right?

This is exactly what happened to a Chevrolet dealership in Watsonville, California.

Their AI chatbot, designed to assist customers in their vehicle search, became a social media sensation for all the wrong reasons. One user even convinced the chatbot to agree to sell a 2024 Chevy Tahoe for just one dollar!

This story is exactly why AI implementation needs to be approached strategically. Learning to use AI, also means learning to build thinking of the guardrails and boundaries.

Here's our tips.

1) AI is not foolproof

Despite the hype, AI is not a magic bullet. It can be manipulated, as the dealership found out. This underscores the need for careful prompting and guardrails to ensure the AI functions as intended.

2) AI needs human oversight

AI can automate many tasks, but it's not a set-and-forget solution. Regular checks and updates can prevent mishaps and ensure the AI is providing accurate information.

(we hope the Chevy team was watching these chats and following up with customers)

3) Start with lower risk applications, avoid customer-facing ones

The unpredictability of AI can lead to unintended consequences, and it's crucial to consider these potential pitfalls when implementing AI in your business.

USE CASES

Write your holiday cards, with AI

Writing holiday cards can be a pain.

It can be difficult to come up with something creative to say over and over again.

Wouldn’t it be great if you could customize your card response for every person in your life with minimal effort?

#StealOurPrompt

LINKS

For your reading list 📚

AI is starting to show its face in elections…

• Imprisoned ex-PM of Pakistan campaigns via an AI voice clone, with Eleven Labs.

• Councilwoman-elect Susan Zhuang turned heads by using AI for questionnaire responses. Is politics ready for the AI era?

• Bing Chat's AI made up facts about European elections, sparking concerns about AI's potential to mislead voters. Mostly likely, we'll see warnings next to protected AI topics - just like we see warnings on social media for politics, provocative content, etc.

Ethics and responsible AI are getting more progress…

• OpenAI is beefing up safety measures against harmful AI, even giving the board veto power.

• TechCrunch put together a quick guide to ethical and responsible AI governance, check it out.

And in the world of competition…

• Expedia's betting big on AI for personalized trip planning, aiming to bypass Google and boost direct traffic.

That's all!

We'll see you again soon! Thoughts, feedback and questions are much appreciated - respond here or shoot us a note at [email protected].

Cheers,

🪄 The AI Exchange Team