A Conversation with Daniel Wu on Strategies for Success

The Gist

• AI Insights at CMSWire CONNECT. Daniel Wu shares tips for implementing AI into your business.
• Meta unveils AI sandbox. A place for advertisers to play with ad design.
• Where’s the beef? Wendy’s partners with Google Cloud to offer the first AI drive-thru.

This year marks the first time CMSWire’s popular customer experience conference, CMSWire CONNECT, is back to an in-person format since the pandemic. For 2023, the conference runs May 10-12 in Austin, Texas, offering a dynamic lineup of speakers, including Daniel Wu, head of AI & machine learning at JPMorgan Chase. Wu presented Friday’s keynote speech on the future of banking with AI and we thought we’d share a few of the insider insights he revealed.

In regard to ChatGPT, while the obvious application is the chat capability, Wu says the GPT models can do a lot more.

“There’s a lot of behind-the-scenes operations, not just within the bank, but any larger organization that AI has been applied, including supporting all these back-office processes, automation and even reinventing how businesses run and how we measure performance and all that,” Wu said. “So, there’s a lot deeper that AI has penetrated in all the industries that I’ve seen, and it’s not just that customer interaction piece or sort of user interaction.”

For companies seeking to implement generative AI within their brand, Wu offered some advice.

“It took OpenAI years to get to the performance, that ChatGPT that we see today. They went through iterations and experiments … and different ways to get here,” Wu said. “So, the framework that I point out for any organization to build up that sustainable pattern, is to look at five important ingredients.”

1.) Data: Figure out what your organization’s data strategy is.

2.) Talent: People trained in machine learning, computer science, data science, etc.

3.) Compute: So far, you see models getting larger and larger. Wu says training these models actually takes months, so it’s a huge amount of compute, and not every organization has that strategy.

4.) Operation: Wu said, once your data scientists and your AI team develop a solution, how do you take that into production and integrate it with the rest of your software ecosystem and build an operations team to support it.

5.) Governance: According to Wu, this area has been become increasingly important as AI regulations and rules are debated in governments across the world.

“There’s a really imminent requirement for every organization getting into AI to think about how they can develop this technology responsibly,” Wu said. “Because what you don’t want is to invest so much and push something into production and later you find out that they’d cause some kind of harm. It’s a real reputational risk to any organization. So that’s kind of the framework that I’ve been focusing on in my public speaking … advocating for attention on the governance and responsibility.”

In other AI news…

The Brussels Effect: Does the EU Have AI Superpowers?

AI developers across the globe are paying close attention to the European Union’s recent passage of a draft negotiating mandate, that if passed, would usher in the world’s first laws on AI by a major regulator anywhere — because what the EU does impacts everyone.

Professor Anu Bradford of Columbia Law School once coined the term, the “Brussels Effect,” referring to the EU’s seemingly unilateral power to regulate global markets. She even wrote a book about it, called “The Brussels Effect: How the European Union Rules the World.”

“The EU is one of the largest and wealthiest consumer markets in the world. And there are very few global companies that can afford not to trade in the EU. So these companies need to comply with European regulations as the price for accessing their 450 million (consumers),” Bradford said during an interview with the Institute of International and European Affairs (IIEA).

“That’s not surprising, where it gets interesting that often these companies conclude that it is in their interest to extend the European regulations across a global production and they obey global services and their global conduct because they want to avoid the cost of complying with multiple different regulatory regimes in different places, so, all the EU needs to do is regulate the single market.”

Related Article:Ex-Googler Godfather of AI: Machines, AI on Dangerous Path

Advertisers Invited to Play in Meta’s AI Sandbox

Meta has announced the introduction of its AI Sandbox, a testing ground for advertisers to try out new tools and features, including generative AI-powered ad tools such as text variation, background generation and image outcropping. The platform aims to learn what works for advertisers and make these features available in its ad tools.

The company also revealed new features for Meta Advantage, Meta’s portfolio of automation products that use AI and machine learning, and Advantage+ shopping campaigns including, the ability to easily switch from manual campaigns to Advantage+, the ability to import and use creative like branded videos or customer demonstration videos, as well as access to new reports and performance products.

Currently Sandbox is being tested within a small group of advertisers in order to gather feedback for possible improvements. The company will gradually open it up to more advertisers beginning in July and add some features into its products later this year.

Google I/O: A Very Big Year for AI Development

On Wednesday, Google hosted its annual developer event, Google I/O, a much-anticipated one-day conference that lifts the curtain on all the tech giant’s latest new products and updates. Google CEO Sundar Pichai admitted it was a very big year for AI as he announced several new innovations including the release of PaLM 2, Google’s largest AI language model to date, and the debut of Pixel Fold, Pixel Tablet and the Pixel 7a Android phone.