Over the last three years, the number of parameters in AI models have gone up 16,500x and computing power used to train these foundational models has gone up by a trillion times since 2017. Machines have surpassed an average human in most cognitive tasks be it reading, writing, speech recognition, or handwriting recognition.

Unfortunately, education has not kept up with technology’s pace of change. Students are still learning archaic skills like spellings, memorizing parts of the human digestive system, and being asked to describe parts of Rutherford’s alpha-ray scattering experiment. While memory in itself is an important skill, what is being memorized is irrelevant in today’s world, let alone tomorrow’s. At the same time, it is not clear what skills are important for future employability. There is a fear that no matter what we learn, in 10 years machines may do all of it.

The immediate question that colleges are dealing with is whether to allow students to use AI, i.e. Bard or GPT. 

Our favorite attempt at answering this question is via an analogy by Chris Dede – “It is a little like the relationship between Captain Picard in Star Trek: The Next Generation and Data the Android, who looks like a person but who is an AI. Captain Picard uses Data to do things that he can’t easily do. But Data is not in charge of the starship because Picard has a whole set of human wisdom that Data can’t possibly have because he’s an AI.” 

Contrary to common belief among educators that the use of technology in schools will inhibit learning, we believe each generation defines a baseline of technology which allows the next to build and learn on top of it, using their uniquely human creativity. In the 40s, when only vacuum tubes were known, programmers could only write simple programs by inserting wires into tubes. With assembly language programs, we could do more complicated tasks. The world changed with high level languages, and then with open source. We can teach today’s students how to create much more sophisticated software, e.g.identifying biomarkers from an MRI scan, by building on top of existing AI stack and open source libraries.  . Learning how to write a new database is not a required skill today for 99%+ programmers..

We need to look at AI as a redefinition of the existing baseline of knowledge. We need to adapt what we teach over and above what LLMs can do, and not hide our heads in the sand and pretend they do not exist. 

The challenges faced in adapting education to the latest technology stem from the industrialization of the field – we’ve been using the same assembly line techniques in education that were once used in manufacturing (classes, degrees, grades, courses, assessments, and so on). Education has been traditionally viewed from the perspective of educators or institutions, rather than that of the students. While this standardization has made education more scalable, it has come at the hefty expense of personalization for the students. We need to put the learners first, and at the center of the learning process. 

For our education system to harness the advantages of AI, the following needs to change:

  • The future of education should be interdisciplinary. New ideas will come at the intersection of disciplines and not necessarily within them. For example, when you combine AI with robotics and biology, you might be able to create the next gen robotic surgery equipment. We need to change the way higher education today boxes people into “electrical engineering”, “psychology” and “economics” and instead create programs at the intersection of these disciplines.
  • The skills being taught need to evolve. As large language models (LLMs) render the value of memorizing facts futile, what becomes critical is the ability to generalize the learning to new situations and to creatively develop new ways of thinking about an issue. Focus of instruction needs to shift to skills such as curiosity, creativity, problem solving, critical thinking, personal/social awareness, empathy, leadership as they’d become the long-term differentiators of humans against machines. Universities need to significantly up the ante on group projects, presentations, joint exercises with the industry, exchange programs to help students develop these skills.
  • Educators need to learn the new toolset. It is very hard for someone to teach others if they themselves are irrelevant already. It is a must for all educators to know what technology can do, how it is applicable to their domain, what tool sets are available and build familiarity with them. Teachers need to adapt these tools to automate routine tasks, and focus on delivering higher-value instruction. 

The world ahead is both scary and exciting. It will allow many earth citizens to do what they could not do earlier, e.g. a lot more of us will become programmers tomorrow. At the same time, it is going to challenge the relevance of many others as machines will take over what they do. We all need to decide which side we want our children to be on.

Originally published in Times of India: