In 1976, Stevie Wonder caught a segment on NBC's "TODAY" show featuring a blind man like himself demonstrating a machine that Wonder knew he immediately needed to have.
The machine was able to read text on a page and speak the words out loud, and it had been made to work specifically for blind people
The device was the Kurzweil Reading Machine, named for its inventor, Ray Kurzweil.
So the blind singer contacted Kurzweil's company and went to its headquarters. After a quick demonstration, he became the company's first official customer.
So began a long friendship between the inventor and the musician. The two would eventually collaborate to produce a groundbreaking musical instrument that uses artificial intelligence to create sounds.
Indeed, some form of artificial intelligence has been at the foundation of just about every one of Kurzweil's inventions. Kurzweil, 68, has spent much of his career building technologies that can learn and think in the ways humans do.
One of the unique human capabilities is the ability to recognize patterns — this is still something humans do better than machines, and it is crucial to many of the tasks only humans can do. Building machines that learn patterns can, in Kurzweil's mind, create inventions that augment human intelligence and help us overcome the challenges such as disabilities. This conviction led him to invent the reading machine that would later captivate Wonder.
Science and technology luminaries such as Elon Musk and Stephen Hawking have publicly warned of the challenges and threats humanity faces from artificial intelligence. But Kurzweil says artificial intelligence in some form or another is already all around us, and it has made us wealthier and more productive.
"AI, artificial intelligence, machines doing tasks that used to require human intelligence, is deeply integrated into our infrastructure," he said in an interview with C-SPAN in 2006. "Intelligent algorithms fly and land airplanes, guide weapons systems, make billions of dollars of financial decisions," among many other examples.
Kurzweil said he believes that machines will achieve the full range of human intelligence by 2030, but "it won't be an alien invasion of intelligent machines that compete with us. I mean it really is amplifying our own civilization. And we are going to literally enhance our own intellectual capabilities by merging with this technology."
His own work in the field grew out of a desire to use machine learning to solve real world problems.
"What is exciting for an inventor is not just an abstract theory, but that leap from formulas on a blackboard to actually changing people's lives," Kurzweil said in the 2006 C-SPAN interview.
The "Kurzweil Reading Machine," which was announced to the public on Jan. 13, 1976, had that ability.
After the first prototype was developed in 1975, Kurzweil approached the National Federation for the Blind about the possibility of working together.
"They provided seven blind engineers who collaborated with me to create a product that was optimal for use by blind users," Kurzweil told CNBC in an email.
The machine made use of a number of technologies that were new, many of which Kurzweil himself developed. For instance, computers had previously been quite bad at reading text. Software at the time Kurzweil was building his Reading Machine could only read text in certain types of fonts — not nearly as many as appear in books.
When the product was unveiled it created a media stir — newscaster Walter Cronkite used the machine to read his signature signoff at the end of one of his broadcasts.
It also turned Kurzweil and Wonder into friends, and eventually, collaborators.
In 1982, Wonder was giving Kurzweil a tour of his new studio, Wonderland, and together they hatched the plan for Kurzweil's next invention.
While showing Kurzweil around, Wonder found himself "lamenting that there were two worlds of musical instruments," Kurzweil told CNBC. "There were nineteenth century acoustic instruments such as the piano, guitar and violin that were still the instruments of choice, but they were difficult to play." Also, one musician could not play several instruments at once.
On the other hand, there was the "computerized world of instruments, which could remember what you played and allowed the musician to edit a multi-instrumental piece the way a user could edit a document on a word processor."
But these computer-based instruments could not replicate the beautiful sounds acoustic instruments could make.
Wonder challenged Kurzweil to "combine these two worlds of instruments, and create a computerized instrument, which would have the desirable deep sounds of acoustic instruments with the powerful control methods of computerized instruments?"
Kurzweil again thought that pattern recognition and machine learning could help bridge that divide.
The two formed a company — another one of the several companies Kurzweil has founded over the years for his inventions.
Wonder became the musical advisor, providing extensive input on the project over the next few years. In 1984, the company released the Kurzweil 250, widely considered the first computerized instrument that could realistically mimic the grand piano and other orchestral instruments.
The important thing to note about the Kurzweil 250 is that it doesn't play back recordings of instruments. As Kurzweil said in an interview with NBC News when the original K250 was released, recording all of the 22,000 sounds a piano can make would have required about 10,000 different computer chips.
The synthesizer instead uses a computer model to create piano sounds, violin sounds, human voices and potentially any other sound a user would want to make.
He sold the company, Kurzweil Music Systems, to a Korean company, Young Chang, in 1990.
The project with the National Federation for the Blind has continued. Over time, the original reading machine, which was so big that it took up much of the back seat in Stevie Wonder's cab, has shrunk to the size of a simple app on the iPhone.
Blind users can carry the Kurzweil reader app anywhere and use it to read just about any print they might encounter in a day. A user simply takes a picture of the sign and it will read it. Since users cannot see, the app will tell them to "take a step back" or "move the camera to the right" to get the best picture.
These days he works as a director of engineering at Google — his first job working for someone else — where he heads a team trying to teach pattern recognition to computers.