Hope you like chatbots

“Four-score and seven years ago, I snagged a great deal on Marketplace!” 

That’s our best guess at what Meta’s new AI Abraham Lincoln chatbot will say as we peruse for online deals. The company is rolling out a new suite of AI chatbots as soon as next month to retain, assist, and collect data on the ~3.9 billion users across its platforms. 

  • Each chatbot has a humanlike character called a “persona,” like one that speaks as Abraham Lincoln or another that gives travel tips using surfer lingo.

Zoom out: Big Tech is turning to AI chatbots as a solution for any and all problems—be it falling user engagement, flagging products or services, or fixing lacklustre customer service. 

  • Amazon is turning Alexa into a voice-powered ChatGPT-equivalent after the product’s performance came under heavy scrutiny last year.

  • Uber is working on an in-app chatbot to enhance its customer service. Delivery competitors DoorDash and Instacart are also working on bots.

Yes, but: Bots are still imperfect products. Companies rolling them out at full-scale is setting off alarm bells over misinformation, privacy, and safety. Look no further than the response to Snap’s My AI, which is under fire for inappropriate messages and secret location tracking.

  • A recent study also uncovered a type of attack that offers “virtually unlimited” ways to bypass safety rules baked into major large-language model chatbots like ChatGPT.

Why it matters: Pandora’s Chatbot Box is open. From Google to Apple, and everyone in between, more chatbots are coming, each with their own set of risks. Companies are working hard to mitigate these risks, but it’s a build-the-plane-as-you-fly-it situation. —QH