Will.i.am talks AI in music, business and life

William Adams, commonly referred to by his stage name will.i.am, is most well-known as the cofounder and frontman for the Black Eyed Peas -- but he has also shrewdly invested early in companies such as Beats and OpenAI. He has served as an advisor for several Big Tech companies, as well, including Intel and Salesforce, and has his own AI company.

Will.i.am publicly chastises Silicon Valley investors and the region's financial hierarchy for passing on nonwhite tech leaders. That includes himself, who found support for his own products in India after finding rejection in California.

Will.i.am made part of his fortune as a musician, producer and collaborator with the likes of former Black Eyed Peas member Fergie, as well as Britney Spears, Kendrick Lamar, Miley Cyrus and Rhianna. His AI company, Raidio.FYI, which uses generative AI to create customized two-way interactions and entertainment programming between the app's AI personas and users, aggregates creative content and works as a collaboration tool. He's also a regular Dreamforce presenter and was interviewed by Salesforce CEO Marc Benioff on stage in September.

In this Q&A with TechTarget Editorial, the artist and tech entrepreneur discusses where generative AI is taking the world of business, the arts and education, and how cloud companies can make it more equitable for artists and inclusive for all audiences.

Recently at the MIT Media Lab, you did some impromptu demos of your app, Raidio.FYI. It was striking that the voices in it sounded so much more human than, say, Siri or Copilot (formerly Cortana). So, what's wrong with AI from other sources right now?

Will.i.am

Will.i.am: It's how we how we do our text-to-speech (TTS). Large Language Models (LLMs) don't know they're speaking. It's "just being read," right? Maybe you have four symbols that represent expression. You have a question mark, period, exclamation point. Now add a comma there, and that's really all the English language -- or pretty much all languages -- have to express text.

Can you truly be expressive with text? Yes. You have fonts and italic and bold, right? All caps means you're yelling at me. Now we put some imagination on what all caps might mean. Really, it doesn't mean that's how we speak, but when it comes to expressing, you have to invent [something] -- now that there's something else speaking it out, you have to use your imagination? How do you put a giggle on top of stuff? Up to this point, we never had to do that. We never had to write and assume that somebody's going to read it and giggle it, as opposed to having brackets and say, [speaking] [giggling as speaking]. It's sort of why emojis sort of arose as a new language. But the LLM and TTS is not reading the emoji, right? So, we had to invent some s--t to change how we write, to alter the LLM's output, to tell the TTS how to express itself.

Raidio.FYI seems part project management and collaboration software, part entertainment. But it's also AI infused. How do you describe it?

Will.i.am: When we first started FYI, it was like, "Hey, let's put the digital asset manager and the messenger [together], because why should any creative have to work across four or five different platforms?" Why are the messages here on WhatsApp but then all my files are there in Dropbox? If I have a larger file, we have to put it on WeTransfer because we can't send it on email and I can't even open those large files in the messenger. Why is that? Why is that normal? And why can't I just send large-ass files on the messenger? Why do I have to use Zip? I don't want to have to do these things. That was the first thing we did. Then it was like, if we have this presentation project, workflow or artwork, we could all work remotely -- it was built during COVID, and we could see it all and participate in it all at the same time.

We saw that if we put AI in this project, the AI will understand all of our work, all of our conversations on this project. And then we decided to innovate radio and use the digital asset manager as the content for this radio architecture. So, we'll do articles and segments in segment form, and then put links in the article to reinforce what we're writing about and then aim it at credible sources that correlate to what we're writing about. So, humans are still in the mix of curation for AI to present it in the form of radio when we use the messenger for new media. That was the "Aha!" moment.

If automobile companies adopt this architecture for their vehicles to transform people's relationship with information, then we truly innovated in the radio space. No one's gonna buy a car without radio. And now we've innovated in radio, because radio [was always] super helpful, even at the height of the streaming era -- and that market is flat now.

Oracle, Salesforce, ServiceNow and many other companies have released their first autonomous AI agents. What's your take on those, especially Salesforce's because you were involved with Dreamforce?

Will.i.am: Marc Benioff mentioned Waymo [autonomous vehicles] and how Salesforce differentiates itself from Waymo. Because for all the Uber drivers, cab drivers, bus drivers, truck drivers and delivery drivers, Waymo is like, "Hey, watch out." As autonomous AI is unfolding, Waymo is going to mess stuff up for a lot of folks that depend on those jobs.

Marc said, "We're gonna augment the workforce." It's not a one-to-one comparison like, "We're the Waymo of the enterprise, right?" That would be like, "Wait, what do you mean? You're telling me thousands of jobs are gonna get obliterated, like Waymo [could do with human] drivers?"

That was the part [of Agentforce] that I love -- that it's augmenting jobs rather than it's going to be a whole new path. You still need humans to curate, you still need humans to task check. You've got these agents that are doing tasks, who's doing task check? We know what fact checks are, and we didn't execute social media with fact checking. What's gonna happen in freaking, like, action check -- who's the action checker and the task checker? The goal checker? As a matter of fact, when it comes to power efficiency and the trillion synapses that we have here [points to head], humans are pretty f------g awesome. AI is awesome, and it's gonna be great. It's gonna push us to use more of [our brains], which we're only using 10% of.

What can customer service leaders from, for example, large credit card or health insurance companies learn from the way you set up the natural voices on Radio.FYI?

Will.I.am: They should rock with us. We could build them some personas. If you think of all the people in the [Dreamforce keynote] audience, there are lots of different types of people in the audience, including [yogi and mystic Isha] Sadhguru, and they all have a different way of tackling problems and making people feel acknowledged. Some of them do it better than others, and Sadhguru is on the side of "does it better than others," giving you some sense of peace. Music -- some musicians do it better than others, too. Some musicians also agitate. But the whole concept of music is to make sense of noise. And I think our approach to AI, because I've been dealing with voices and tonality my whole career, and applying that methodology to what type of personas are we building, what's the tone of the voice, how at the core, it's there to serve?

So, what else? What other type of color should we bring in these personas? Does it have stances? Does it stand for anything? What does it say to remind you of the importance of [what] things? Or is it just like robotic servitude? At some point it's gonna get to a point where it condescending. Does it pretend to care to like you, or does it truly care? Because customer service is caring. Well, the good ones are like, "Oh, wow. They really do care, yeah," and it has to get to that; some version of that.

Don Fluckinger is a senior news writer for TechTarget Editorial. He covers customer experience, digital experience management and end-user computing. Got a tip? Email him.