AI Briefing: How Priceline and other e-commerce companies are approaching generative AI

Beyond adding just a few new generative AI tools, some companies are using large language models to overhaul entire platforms for customers.

Last week, Priceline added more features for Penny, an AI travel assistant that debuted last June. When it first was launched, Penny could only help with finding accommodations, but now it can help find flights, car rentals and vacation packages. Built with OpenAI’s GPT-4 and Google’s Vertex AI, updates in the online travel agency’s “trip intelligence suite” include using Penny find destinations, generate itineraries and modify plans directly with the chatbot. 

According to Priceline’s svp of strategy and brand marketing Lesley Klein, the company is testing generative AI in other parts of the business including paid and organic search, messaging formats and various creative assets.

“We as a company were very early and enthusiastic about embracing the possibility of generative AI,” Klein told Digiday. “There’s no denying that it is going to change every facet of business and life, so we didn’t want to be shy about it.”

Penny is trained on a range of data sources: high quality content, first party data, search data, on-site behavior, etc. It’s also plugged into various APIs to help answer questions beyond Priceline. It might also help with other parts of Priceline’s marketing strategy. For example, Penny might someday help identify psychographic and behavioral data that wouldn’t have been previously clear.

Instead of beginning with the early steps travelers take when planning a trip, the company started at the end, according to Priceline chief product officer Kevin Heery. When people check out, Priceline asks if there’s anything else they want to know. Are pets allowed? What’s the cancellation policy? Do nearby restaurants meet dietary needs? 

The approach helped to “get to the real heart of what people might be worried about,” Heery said. And while Priceline expected a conversion deficit from the distracting questions, that never happened.

“It really helped us as we move up the funnel from checkout,” Heery said. “Then we moved up to the details page where people start to think about the position of a hotel. We’ve now gone up to the home page, and who knows what people will ask on there?”

Giants and startups alike are adding more ways to help consumers, enterprise customers and employees. One startup that debuted last week is Sierra, co-founded by former Salesforce co-CEO Bret Taylor, which promises to help people use AI agents to perform various actions. Another, XGenAI, is focused on building new generative AI features for retail and e-commerce. However, some companies also have had bot-related setbacks. Last week, a Canadian tribunal ordered Air Canada to issue a refund after a chatbot gave inaccurate information in a 2022 conversation with a customer.

Generative AI is already being adopted by a number of e-commerce businesses. In a new report from Junglescout, 50% of businesses on Amazon have used AI to manage e-commerce channels. Around 34% are using AI to help with writing and optimizing listings, 14% use it for creating marketing and social content and 7% are using it for helping with keyword and SEO research.

Prompts & Products: Other AI news and announcements

  • OpenAI announced a number of major updates including giving ChatGPT a way to remember conversations. It also previewed a new text-to-video model called Sora, which lets people make life-like videos. (The model is likely to compete with other AI startups like Runway and Stability.)
  • The open-source platform Hugging Face added a list of detection tools for mitigating the risks of AI.
  • Cohere AI research nonprofit released a new large language model that supports 101 languages, which could help with driving AI adoption across borders and cultures.
  • The Federal Trade Commission has proposed new rules for protecting people from harmed by AI impersonations. In a separate announcement, the FTC warned it might investigate companies that try to use personal information to train AI models.
  • Salesforce announced new generative AI features for Slack to help workers summarize conversations.
  • Alembic, an AI startup focused on marketing analytics, said it’s raised $14 million from investors including a VC founded by Dreamworks co-founder Jeffrey Katzenberg. 

Other stories from across Digiday

  • Nearly two dozen tech companies are making more commitments to address harmful AI content related to global elections.
  • Qualcomm has crafted a new sonic logo for Snapdragon, the mobile processor platform powering a wave of new AI-enabled smartphones, PCs and headsets. The “Snap Beat,” which debuted last week, will show up in an array of marketing efforts from Qualcomm and its various partners. 
  • Publishers like The Washington Post are making it harder for AI crawlers while others like Politico EU are welcoming them.
  • Digiday+ Research: Most brands and retailers employ generative AI, with chatbots topping the list of uses.

https://digiday.com/?p=535284

More in Media

Publishers’ Privacy Sandbox pauses settle into a deep freeze following reports of poor performance

Publishers aren’t ready to press play yet on dedicated Privacy Sandbox tests.

AI Briefing: Senators propose new regulations for privacy, transparency and copyright protections

A new bill called the COPIED Act aims to pass new transparency standards to protect IP and guard against AI-generated misinformation.

Media Briefing: Publishers reflect on ad revenue midway through 2024 

Some publishers say ad revenue is pacing 15% up year over year while others are still managing their expectations for how 2024 will shake out.