Last year, in the height of the election season, the Obama administration quietly released a national strategic plan for artificial intelligence (AI) research and development. The plan was the beginning of a national effort to prepare Americans for a future with AI—a future some computer scientist believe our nation is ill-equipped to handle.
AI has become a part of the American fabric for some time. Siri and Alexa are already taking orders, self-driving cars have hit some streets, and the concept of interconnectivity is now a reality through the Internet of Things. But experts assert that in order for the society to fully embrace AI, learning machines should not replace human workers, but complement them. So to prepare the future workforce for a computer coworker, there must be a shift in teaching and learning—a change that should begin in the classroom.
Using research and concepts from several AI experts including Mark Stehlik of Carnegie Mellon and Rand Hindi of Snips, EdSurge put together the following three-step list educators can use to start implementing AI education in schools.
Step 1: Demystifying Artificial Intelligence
Despite what the movies might depict, most computer scientist understand that society is not and will not be in Terminator land anytime soon. But the question remains: “What stage of AI development is our society in?” That is a question students and teachers can use to begin AI exploration in class.
Mark Stehlik, a professor at Carnegie Mellon University’s School of Computer Science (CS), noted in a presentation at SXSWedu that, “People shouldn’t be afraid of AI. They should understand what it can and cannot do.” He believes conversations that build AI understanding should begin with students in the classroom. He also believes that it is important for students to understand that many of the tools they play with right now are built with machine learning capabilities.
Dr. Rand Hindi, CEO of Snips (a machine learning device company), is part of a research group working with the French government to prepare their country for AI. He believes that it is important to understand how much of the economy is vulnerable to automation, but won’t be automated because of other variables. “Even if a job can be automated doesn’t mean it will be,” said Hindi during New York University’s Future of AI Panel. “What you can do technically is not what will happen in practice because in practice you have to take the human variable into consideration.” For Hindi, part of demystifying AI is being able to understand the scope of its capacity, which he believes can affect about 40 percent of our economy.
Step 2: Understand Everyone’s Role, AI is Not Just For Computer Scientists
A common misconception about AI education is that it’s only important for students pursuing computer science. But understanding AI can teach students many different skill sets that help them work alongside machines, soft skills like psychology and communication.
“Some parts of jobs are emotional. You could automate a diagnosis, but do you really want a robot telling you have cancer rather than a doctor,” asks Hindi. “What I think we should teach people is how to be complementary with AI.” This notion of complementary skills training that Hindi brings up is something that higher education institutions are already considering.
Representatives from the University of John Hopkins medical school sat in on an AI panel at SXSWedu to learn how to prepare their medical students with complementary AI skills. “65 percent of students are going to have jobs that don’t exist yet, and this AI stuff is going to be a big player, it already is, jobs are changing,” says Dr. Kimberly Duncan, the director of Innovation at John Hopkins, in an interview with EdSurge. “The same thing is happening with the way we make discoveries in medicine. So we have to think differently about how we prepare people for the new world. What we are doing right now isn’t going to cut it.” For Duncan, it is critical that her medical students learn how to work with machines.
Stehlik also maps out several complementary roles that students in the K-12 space can begin to learn. “Think about a self-driving car that has to decide if it is going to hit the vehicle in front of it because it cannot avoid that or pull off onto the sidewalk. But there are people on the sidewalk,” explains Stehlik. “Who is the car beholden to? Is it beholden to protecting its occupants, or it is obligated to calculating how many people are likely to die and chose the minimum of that? How does the machine decide?” These are the types of AI issues that will affect a broad group of people. This ethical question is something Stehlik believes needs to be parsed out by people who understand ethics. Skills that go beyond programming like psychology, philosophy and design thinking skills can help with answers. “I don’t want some person in Silicon Valley to make that decision,” remarks Stehlik.
Step 3: Ramp Up Teacher Training and Revise the Curriculum
“Computer science programs, while they are great, are band-aids,” says Stehlik who believes that educators in the K-12 space are severely underprepared to teach AI concepts. “You wouldn’t train a physics teacher to go teach biology,” he continues. “What we need to do is get computer science students into computer science teaching.”
For Stehlik, the onus is on technology companies and higher education institutions to prepare K-12 teachers for AI instruction by providing them with curriculums, capacity and continuing education opportunities. He believes that companies like Google should encourage their employees to go into schools and teach classes.
“[Tech companies] have been complaining about H-1B visas, but instead of complaining about capacity, they can be doing something about it,” says Stehlik. “If you paid someone Google salary to teach for two years, they could educate more personnel to work for Google in the future.”
He also believes higher education institutions can support teachers by offering free computer science courses and curriculum to educators. When he is not teaching at Carnegie Mellon, he spends time teaching an AP computer science in a local high school. “The curriculum that was in use at the course that I was teaching was 12 years old, and for 12 years it was not a good course,” says Stehlik. “It was doing things like asking students to calculate Fahrenheit to centigrade conversions, that is not engaging.” He revised the curriculum to mimic the technologies students actively used. “I created a graphics program for C++ that a thousand teachers used, kids like graphics,” he said.
According to Stehlik, certain math concepts that are actively taught like Calculus and Trigonometry are simply not useful in the computer science field. “Analyzing huge data sets statically with machine learning requires math, but not calculus, it’s statistics and probability. The kind of math that our curricula are not attuned to right now.”
For school districts just getting into the Computer Science education, preparing students for a future with something that they might not even live to see seems like an arduous task, but the impetus to begin the process has entered the minds of many experts in the field. For them, it is imperative that educators begin to explore this reality and press students to aspire to be part of this profession.
Stehlik is convinced that technology and medicine are going to drive to U.S. economy forward, but worries about students who aspire to jobs he believes are artifacts of the past. “But why should humans go a mile underground, get black lungs and die at 50 from coal mining? No, let robots do that work,” says Stehlik. “What that really means is we failed that person educationally and aspirationally. Now more than ever we have to teach for the future.”
