One interesting outcome of last month’s Imagination In Action conference was a very rich interview with will.i.am (titled “Creative Harmony”) conducted by CSAIL director Daniela Rus.
Will.i.am livened up the talk by sharing his own experience with AI and related technologies, explaining how he met MIT professor Patrick Winston in the early 21st century and then began using AI in sound and video production, eventually leading him to become involved with Beats.
“Computers are why I became a musician,” he said, noting that programming and computer science were his early career goals, and speaking about the intersection of technology and music.
Explaining how AI has evolved, he said early pioneers said that one day people would talk to computers using language rather than programming.
That time has now come, he noted.
will.i.am at MIT
“There was no concept of a law degree back then,” he says, explaining that will.i.am brought something in a briefcase to the band in 2010, demonstrated the potential of a technology that was still in its infancy at the time, and said, “This is the future.”
He said many fans know the Black Eyed Peas lyric “I’m 3008, you’re 2000 and late,” but he added that the lyric was born out of the idea that we were rapidly embarking on a journey towards innovation.
“I’m a curious person,” he told Russ, explaining how he got involved with AI in the first place. “I like to ask questions.”
He explained that he realised early on that the potential of machines would eventually make many human activities unnecessary.
“Why am I here creating when I can be creating something to create?” he said.
will.i.am at MIT
will.i.am explained how he started investing in AI with companies like OpenAI and Anthropic,
He is now CEO of FYI.AI. He typed in a demo for the audience that started with an AI responding to a Bob Marley avatar. His AI responder used the phrase “put your imagination into action,” which I thought was great (listen carefully). The quality of the conversation was amazing.
Below are some AI ideas aimed at promoting diversity and democratizing technology.
“It’s about showing everyone that technology isn’t just for a select few. It’s about bringing voices like ours into the mix, for black women and inner-city people to see themselves reflected in this tech revolution. Cha. This is about empowerment, showing that AI can speak our language, understand our struggles and amplify our stories. When people see AI (like this), it closes the gap, makes technology accessible and shatters the notion that something as powerful as AI is out of reach. This is about inclusivity, breaking down barriers and making sure the future looks like it for everyone.”
When asked how the company achieves this, will.i.am said that FYI uses natural speaking voices rather than voice actors.
He also talked about how he’s using an AI named Fiona to run the show.
will.i.am at MIT
“People can engage with their co-stars every day, on a one in a million basis, rather than the limited experience they currently get with traditional media,” he said.
He suggested that by abandoning the training of stiff conversation, one can develop a more authentic and engaging side to oneself.
“They were amazed at how it made them feel when the AI communicated with them,” he said of a De La Soul interview in which Russ asked how guests would react to speaking with an AI. “The AI is a superstar. It’s a star. It’s not just a tool.”
Elaborating further on what makes up FYI’s “secret sauce,” he said part of the technique involves “puppetizing” models, but suggested natural language also helps.
“You can’t write a book,” he said, contrasting FYI’s personas with more formally trained traditional models, which he characterized as “very old-fashioned” and “very bland” because they lack conversational authenticity. “The magic is in how we collect and record the data.
“We had a real conversation, not a ‘perfect’ one, but more like ‘ums’ and ‘ahs.’ And the second thing is that the LLM has no idea what he’s saying. He thinks he’s being read.”
That being said, he went into more detail about how LLM works with text-based components, which he calls TTS, and how it will be a pioneer in bringing those models to life.
When you read a text, all the coloring is done by your imagination. … Can a machine have imagination? No. … So how do you get it to express itself? You have to break away from traditional ways of writing and reading and start thinking differently.”
Another interesting part of the will.i.am interview is the overview of modern world music history, from the time before he started recording and touring, which I think is interesting and worth sharing in detail.
On the late 1800s: “I was a musician, and I played in orchestras and bands, and in theater and classical music, but I couldn’t imagine a music industry. There was no music industry. There was no such thing as radio. There was no such thing as records. There was no song structure as we have today. Songs were written to tell a story, and for a conductor to conduct an orchestra. So popular music was not what we have today as popular music.”
On a time before people had common recording equipment: “People listen to music in churches. They listen to music in theaters, but do they all sing the same lyrics randomly while cleaning their house? You don’t have that experience. What happened? The recording industry did it. Every industry that has to do with music owes something to some technology. In 1449, because of the printing press, and shortly afterwards Mozart, Bach, Beethoven. They composed because they printed the scores so conductors could conduct and they sold more oboes. Different instruments to play music printed on scores, that was the first industry.”
On the recording era of the 1900s: “You have the rights, the mechanical rights, or the contract. A song is four minutes and a few seconds. Anything over six minutes is part one and part two. Look at all jazz songs. Any song over six minutes is part two, because that’s how much information you can fit on the lacquer. It’s a limitation.”
Next, touching on the latest developments, he spoke about the rapid changes that AI can bring about.
Now we have a new technology. So should this new technology be part of the recording technology? That would be foolish. Why put abundance and infinite possibilities in an industry based on limitations? Why have an AI make this kind of music? A recording? Something that exists in a cage, a trap container, for three minutes?
At one point he spoke about live musicians and AI speaking directly to the crowd, another unprecedented use of technology in music, and he explained some of the audience dynamics and alternative aspects that AI could impact.
“We’re attached to love songs because they’re close to love. It’s not our real love. We like breakup songs because they’re close to heartbreak. It’s not your heartbreak. It’s my heartbreak. I just shared my heartbreak with you and you’re like, ‘Oh yeah, that’s like my situation.’ It’s not your situation. We like party music. … You go to the club, and when you’re in the club, the song is like, ‘We’re at the club.’ It’s telling you what you’re doing, like, ‘Yeah, we’re at the club. We’re going to have a drink. Oh my gosh, this song is right.’ But it’s a closeness, not your actual (experience).”
Now, he said, we can make music more personal, less indirect and more direct.
“Let’s do something new with this,” he said. “Let’s make it taste your true sadness. Let’s make it come alive. What’s your song?”
But it’s important that people own their own data, he said.
“This is a wrong practice,” he added. “We’re handing all our data over to these data monarchies.”
He also gave a good analogy comparing people’s data to their nervous system or digestive system: “Why data when other things can’t be replaced?” he pointed out.
The challenges to humanity are widespread, he added.
“This isn’t just about Frank Sinatra,” he said, explaining how AI is trained on human endeavors and what that means for everyday people: “It’s your neighbor. … If you can simulate a building not collapsing in an earthquake, you can simulate how a person sells, how a person loves, how a nurse takes care of a person, how a person raps, all of that rhyming… People need to own their stuff.”
In that vein, he also spoke about his ongoing efforts to nurture a new generation through the College Track Foundation, which was established in 2008 and initially started with 65 students but has since grown to have 13,000 students moving on to various schools.
“Our students always aim to get into MIT,” he said. “It’s hard to get in, but our students work towards that goal every year, especially our computer science majors. … MIT is a special place for me, not only because it sparked my interest in AI, but also because of my passion and purpose. … The program we revamped is one of the most successful with the most schools adopting it in the district, with 400 robotics teams serving more than 13,000 students. Now we want to continue to grow all of this to prepare our young people for tomorrow.”
Speaking of opportunities and challenges, will.i.am acknowledges where AI is at right now. “This is a new renaissance,” he says. “There’s this wave of humanity… This wave is important. It’s going to change how we solve disease, how we solve climate… I’m optimistic. I’m inspired by the engineers and the people who have been working in this field before it was trendy, before there was any fuss, the people who did their law degrees in the shadows. I salute them.”
His call for kind and compassionate outcomes was moving.
“There are a lot of companies trying to get into AGI, but what type of AGI are you trying to get into? Is it empathetic? Is it kind? Or is it analytical and judgmental? … I’m looking to get something empathetic, something that gives you a sense of being loved and cared for. There are a lot of lonely people out there. There are a lot of people that are scared of this field. And as this job displacement happens, there’s going to be a lot of people that need to feel loved. … It’s about using your imagination.”
