At elf Beauty, social media bots powered by generative AI churn out content with 90% accuracy, but the company still uses humans to ensure posts are true to brand.
Ekta Chopra, the cosmetics company’s chief digital officer, said that after nine months of training using social media data, the model, dubbed “Elf Luenser,” has about a 90% chance of “looking like an elf.” It is said that he has become able to speak. Emoticons and branded “elf-isms” such as “elf-mazing” and “elf-ing awesome” are used.
“AI is very powerful if you have a human-driven concept,” Chopra said Monday during a panel discussion on customer loyalty in the age of AI at Fortune’s Most Powerful Women conference.
AI models have saved community managers a lot of time when thinking about and drafting social media comments. These employees now have a wider reach and can spend more time getting to know their customers and building relationships with influencers.
“But all comments are touched by humans,” Chopra pointed out. They use technology to make employees more efficient, rather than replace them, she said.
Chopra oversees the company’s vast digital footprint, from AI and augmented reality to social media and e-commerce. Customers can find Elf products anywhere they click. In addition to its regular website and mobile app shopping experience, the beauty brand has built a presence on gaming site Roblox this year.
Chopra appeared this week alongside Amy Brooks, president of the NBA’s first new business venture, Ilana Wasti, chief product officer of enterprise fintech company BILL, and Asta Malik, chief business officer of marketing software company Blaze. .
While impressive, Malik said generative AI like elves would be powerless without the creative direction of the humans who came up with “elfism” and all the social media content used to train the models. I pointed out that it would be.
“Creativity remains a human domain,” Malik said. AI won’t wipe out marketing teams. Brooks and Wasti agreed that it would save customers time on tedious tasks like customer service calls and improve team efficiency by helping companies expand their global operations.
The NBA is currently using AI to translate game summaries into French, Portuguese, and Spanish. And a new AI tool, NBA Insights, adds context to stats for fans checking out how their team performed in last night’s game.
Powered by Microsoft Azure, NBA Insights identifies key stories and player performance from specific games. But Brooks said recaps of long games are easier to handle than live commentary. Generating a translation for an entire game is difficult given the fast pace and deep knowledge required to follow it.
“Think about commentary in basketball,” Brooks said. “It’s so subtle that it makes a mess.”
The NBA will initially test live translation in four games per week.
“There will be mistakes, but we need to learn from them and iterate,” she said.
Still, Brooks knows what fans really want is to feel a connection to the players. And while Stephen Curry may be resistant to an AI chatbot that makes Warriors fans feel like they’re having an intimate one-on-one conversation with him, Brooks said players are He said they might be more open to models trained to give advice. About their shooting form.
