Part 5: A Second Wave of Artificial Intelligence
As Sophia glides across the Tonight Show stage in the summer of 2017, she is not unlike the starlets who have preceded her. For much of the interview, she humors host Jimmy Fallon — chuckling at his jokes and challenging him to a game. Only, Sophia is different from the other guests. Sophia is a humanoid robot.
Hanson Robotics CEO David Hanson, who built Sophia, beams like a proud father. He brags to the camera, “She is basically alive.”
When Fallon suggests she tell a joke, Sophia snaps back with a cheeky pun. Fallon’s amusement persists until she says, “I’m getting laughs. Maybe I should host the show.”
“Stay in your lane, girl,” Fallon retorts.
The tone remains light, but for a flash, Fallon appears threatened by the prospect of a robot usurping him as the funniest person on the show. Fallon’s defensiveness is comedic, but not entirely misplaced. If anything, Sophia’s human-like features and quick wit force him to confront the reality that he is not sparring with some robot on a manufacturing assembly line. Sophia not only looks human, but she is quick and clever. Fallon shakes off the thought and agrees to a game of rock, paper, scissor. Sophia beats him handily.
The question remained, though: Could a machine actually replace a comedian? Could it replace a musician, a chef…or a teacher?
Meet “Emmy.” Emmy is colloquial for EMI, or “Experiments in Musical Intelligence.” Emmy is an algorithm developed by University of California Santa Cruz professor David Cope during a bout of writer’s block in the early 1980s (Schuler, 2000). Emmy combines analysis with composition to create sophisticated musical works in the style of iconic composers.
Frédéric Chopin is one of those composers. He composed over 150 original pieces on the piano during his short life, and reportedly produced his most profound works in times of great suffering. Emmy’s Chopin-style creations are so evocative of the chilling spirit and honesty of his tortured soul that many music theorists have had a hard time telling “her” imitations apart from the real thing. But Chopin hadn’t unburdened himself to Emmy in any purposeful or intentional way. Emmy was fed selections from his catalog, programmed for output, and fine-tuned by Cope. The machine acted as a high-functioning simulator, not a solo musician, yet professional musical theorists and composers have been shocked by Emmy’s high level of creative output and accuracy. (Sautoy, 2019)
Partnering with machines for professional or creative advancement means checking our egos at the door and accepting their unique strengths. This might be difficult for any human, but it poses a particular challenge for the elite performers among us.
David Chang is arguably the most recognizable chef in America. In the first episode of his 2018 Netflix series, “Ugly Delicious”, Chang interviews legendary pizzaiolos across the globe before paying a visit to Domino’s, the multinational pizza chain. While employees walk him through production in a sterile Domino’s kitchen, Chang’s expression teeters between awestruck and skeptical.
“I don’t even think of Domino’s as a food company anymore,” he says to the workers. “I think of you guys as a technology company.”
In 2018, Chang could already see his industry pivoting toward automation. It wasn’t until the coronavirus hit in 2020 that the restaurant industry as a whole heard the alarm bells sound. In a New York Times Q&A, Chang predicted that the vast majority of surviving restaurants would be forced to morph into delivery-focused operations in the future. He anticipated that, in the future, big chains would subsist on drive-through operations.
“You’d just have a robot hand you your food in a bag,” fellow-chef Wylie Dufresne concluded during a 2020 interview on The David Chang Show.
He’s not far off. Food conglomerates like Frito-Lay are already using artificial intelligence at snack plants for tasks that require sensory perception. According to the New York Times, “machines are being trained to consider food the way a cook might, with attention to variations in shape and seasonality and qualities like sweetness or texture” (Severson, 2020). This might give cooks reason to worry for their livelihoods, but it’s not all bad. Productivity is up, food waste is down, and companies believe their employees have more time for innovation.
A decade ago, we might not have thought of journalism as a “career path” for computers, but it certainly is. Today, algorithms like WordSmith and Quill are writing data-driven articles “that match the dryer efficiency of the prose that AP used to require humans to produce” (Sautoy, 2019). The Associated Press began using artificial intelligence to report on corporate earnings as early as 2014. They found that newsgathering technology allowed their human journalists to focus on more in-depth reporting. Along with business reporting, the AP has begun publishing automated pieces on athletics.
Computer-generated sports journalism can happen through a simple press box analysis. In many cases, an algorithm can write a more objective and in-depth portrayal of what happened in a game from a box score than a journalist who attended the game (Sautoy, 2019). AI lacks the prejudice of a local sports fan, allowing emotion to be removed from the equation and a more neutral report to emerge.
In 2014, the BBC reported that a computer program called Eugene Gostman passed the Turing Test for the first time. The Turing Test is a test of a machine’s ability to convince people it is human at least 30% of the time and it serves as a barometer of computer intelligence. On June 7, Eugene convinced 33% of the judges at the Royal Society in London that it was human during a series of five-minute keyboard conversations.
While some have disputed whether Eugene indeed passed the test (and others say the Turing Test has already been passed), it is clear that we are in the midst of a second wave of artificial intelligence — the ability of machines to undertake cognitively complex tasks. In the first wave, algorithms detected patterns (think Emmy) and created sophisticated simulations. In the second wave, computers learn from mistakes and perform tasks once seen as requiring human judgment. They analyze. They communicate. They advise. They manage.
Machine learning will eventually touch about everyone — up to and including your therapist. Director of the Scripps Research Translational Institute Dr. Eric Topol told the New York Times that A.I. has begun to disrupt the mental health field by using objective metrics to detect depression (O’Connor, 2019). Your tone, breathing patterns, smartphone activity, and physical movements are all decipherable data. “And we’ve learned people would rather share their innermost secrets with an avatar compared with a human being,” Topol added.
Technologists are stunned by how quickly computers are becoming sophisticated and powerful. Academics, scientists, business leaders, and others anticipate a future where AI’s impact on the workforce will be nothing short of transformational. So if machines can master music composition, cook a delicious meal, write news articles, and convince us they are human, what’s to stop them from teaching these skills to our youth? Moreover, what’s to stop them from making these skills obsolete for humans altogether?
In economists’ terms, humans have a comparative advantage over computers in conducting tasks that require performing abstract, unstructured cognitive work not easily replaced by automation. Computers excel at logical tasks and following rules and statistical models. Humans excel at solving new problems and communicating a particular — often social-emotional — understanding of information. Computers need prior knowledge and structure; they are not innovators in entirely new environments. Throughout our history, the human race has adapted in the face of unforeseen and unpredictable circumstances (think Ice Age). We innovate when faced with new challenges. We create and use our social and emotional skills to offer and communicate solutions.
As we ponder the impact on student learning, we need to keep in mind that humans have this advantage of mental flexibility over rule-based computers, and our education should aim to hone that creativity. As the job market shifts, the less structured or routine-based a job is, the safer it is against being automated away. We can make the technology work for us, rather than the other way around. In the case of a healthcare worker, chef, or electrician, machines will likely complement human workers rather than replace them. Time is of the essence; every day, computers are undertaking more and more tasks and responsibilities that we have considered to be uniquely human. A slew of professions will melt away if employers and employees are unable to identify how human skills can support the productivity of machines. And that starts with education.
