When Vinod Khosla asks “Will We Need Teachers or Algorithms?,” of course the answer is both.  Really, TechCrunch’s title is misleading because Khosla doesn’t argue that algorithms should replace teachers.  I had the pleasure of meeting Khosla when Mike Arrington was interviewing him as part of StartupWeekend EDU in Seattle, and he comes across as passionate about seeing technology’s potential to change education.

Many of Khosla’s points I agree with wholeheartedly—our system of fixed class time for graduation credit needs replacing, online learning has the clear potential to change the nature of teaching, gamification holds possibilities, and big data has much to teach us about student behavior.  I also love that Khosla sees the potential for each of these new online systems to become “a customizable playground for low-cost experimentation.”

Khosla’s issues with our traditional “fixed time, variable learning” model, for example, as opposed to a more flexible “fixed learning, variable time” model are on track.  Algebra is a good example—some students can learn algebra in a semester while others need two years. Instead of providing slower instruction for these students, we often make them take the course twice. Not an effective method of learning. Similarly, we’re holding back students who could fly through the material and move on to more advanced work.  Currently all diplomas are certainly not equal.  Four years of high school can look very different from one student to another, but they’re all required to sit through the same amount of credit hours.

I’m fortunate to know several folks who are working through these big data problems to understand how students learn better, and I do see them as becoming real game changers in education.  Recognizing how long students remain working on specific problems or reading particular passages can tell us more about where students get stuck. We can begin to find patterns that will enable us to support students more quickly. I can only imagine what tomorrow will bring in terms of better understanding student behavior and learning patterns.

Still, this article shows a lack of full understanding of what it takes to be a skilled teacher, aspects that cannot be replaced by algorithms and online programs. While Khosla does acknowledge the need for human interaction between a teacher and a student—and we know that a single adult, such as a teacher, can make the difference between a student’s success or failure, the teacher relationship goes further, or any caring adult could serve the same role, not only trained teachers.

Being a great teacher does not mean being a great lecturer. Khosla seems to believe that most teachers lecture, which is not the case for good teachers. Sure, there are some teachers with a gift for explaining topics clearly. Programs like Khan Academy have the potential to be able to offer multiple explanations of concepts with the hope that a student will find one that makes it all click. However, the role of a teacher should not be seen as a conveyer of knowledge—if a teacher practices this way—“sage on a stage,” then he or she is not a great teacher.

One of the greatest skills a strong teacher has is to be able to guide a discussion or activity so that students discover what they think, believe and want to know. Great teachers can also provide students opportunities to learn empathy and compassion. Skills a computer can’t teach. A skilled teacher can push student thinking further than most computer programs.

Twentieth-century skills such as collaboration, creativity and innovation can’t simply be learned by playing an online game or earning a badge. Different skills are involved in collaborating online than collaborating in person, for example.  Can a computer program teach innovation? Creativity? Inquiry goes beyond researching online.

The one advantage that most teachers would jump on– the potential for help with the time-consuming task of grading—Khosla doesn’t even mention.  Many great teachers become administrators or leave education entirely because of the sheer volume of grading. Though I worry that a machine can’t replicate a teacher’s encouraging comments, artificial intelligence could certainly screen for content or math problems. A computer could comment on a student’s code in a computer science class, as another example.

Khosla and I both agree that online learning has the potential to change the nature of the classroom and the role of the teacher, but algorithms and computer programs will never replace skilled teachers.  With the advent of better targeted online programs a teacher can be allowed to spend more time teaching her students how to be critical readers, writers and thinkers, prepared for the challenges of the 21st century.