The Decades Long Quest for A Digital Aristotle
Aristotle was probably the best tutor in the world and the most knowledgeable person of his times. But still his student, Alexander the Great, went on to conquer half of the world. Being smart it seems, doesn’t automatically translate into being peaceful. The same fears echo today with the rising power of artificial intelligence.
Aristotle is also the name of the essay that a technologist Daniel Hillis published on Edge.org in 2000. The essay received many reactions from top scholars, like Seymour Papert, the founder of constructivism in education who co-authored Logo programming language and inspired e.g. Lego Mindstorms robots and Scratch programming language.
In his essay, Daniel Hillis, imagines a digital assistant that would know people intimately, and was able to teach them effectively — based on their interests and styles of learning. The essay was viewed as very creative and ambitious, but the response from experts was somewhat skeptical. People asked — how can computers know us, and understand us?
Around the same time, at the dawn of a new millennium, a new set of Millennium Development Goals (MDGs) was adopted by the global community. And by the end of that decade, countries like Kenya made their primary and secondary education tuition-free. Also during those years, Wikipedia and Google changed the way how we search for information and create knowledge.
In our decade, since 2010, we saw a tremendous progress in some areas of artificial intelligence, like machine learning and deep learning, based on the exponential growth of processing power of Graphics Processing Units. With even more fascinating progress in neuroscience, based on fMRI technology and combined with machine learning, scientists are getting closer to literally “reading minds and visualizing memories and dreams”.
Could digital assistants help children learn more effectively and escape poverty?
From 2016 there is a new set of global goals — 17 so called Sustainable Development Goals, and a broader Agenda 2030 that was adopted by the UN, and aims to be universal. Meaning, it is not just rich countries helping poorer countries to develop faster — which was the paradigm of MDGs. Sustainable Development Goals imply that all countries are somewhat “developing countries” in their own ways and with their own challenges.
The goal SDG4 is to improve the quality of education and associated targets, are relevant for rich and poor countries alike. There are pockets of generational poverty also in the richest developed world. And education is broadly seen as a means for upward social mobility. But often children from marginalized backgrounds fall behind. Mostly because they lack the creative and motivated environment, and skilled tutors and mentors, that kids from the richer backgrounds have.
One way how to help students learn and teachers teach better, could be the use of smart digital assistants powered by artificial intelligence in the classrooms and homes. But the path towards digital assistants like Aristotle seems littered by failed startups and grand ambitions, that so far didn’t materialize. Still today we can see some encouraging signs — there is a strong trend in EdTech startups and some innovative approaches like Global Learning XPrize competition.
We also saw a success of companies like Duolingo, that bring free language courses on finger tips via mobile apps. And there is a solid research of what is effective in “learning to learn” — for example summarized in the most popular course on Coursera by Barbara Oakley and called Learning How To Learn.
Aristotle today is called Jill Watson or Wolfram|Alpha
Personal attention to each individual student by skilled teachers is crucial and AI might never truly replace real teachers. But digital and exponential technologies might rapidly reduce the cost of personalized teaching and replying questions that students might have.
For example, Ashok Goel, professor at Georgia Institute of Technology, have created an automated AI teacher assistant Jill Watson (on IBM Watson platform) for students taking his courses online.
Google, Amazon and Apple have their AI based digital assistants. Amazon’s Alexa and Apple’s Siri both use calculations in their answers that are provided by Wolfram|Alpha. A computational search engine, named after Stephen Wolfram, creator of the Mathematica software and Wolfram Language.
Similar tools help children to calculate and visualize math problems and reply questions given in plain language.
Currently, there is no easy way for students to go back to concepts they learned over the years, and to see where the gaps in their knowledge and skills are. For example, biology connects to math. “Are there paths that unravel blockages over the time?” This is what Aditi Avasthi, CEO of an Indian company Embibe.com asked.
And her company created a knowledge graph to map the whole education concepts in Indian curriculum. She found out that the key to remove the blockages is so called grit — the ability of students to persevere among difficulties and continue with learning. (Another important and related concept is “a growth mindset versus fixed mindset” researched by a psychologist Carol Dweck.)
Embibe found out that AI can help parameterize human behavior and help students in learning according to their individual needs and style. For example, their software tracked parameters like “first-look accuracy” and “careless mistakes”, and gave students feedback what to improve.
At the end of her TEDx talk, the Emibe CEO, Aditi Avasthi, said: “No child shall be left behind, and no child shall be labeled.”
This is a worthy goal and AI for sure can help to reduce costs of individualized learning. Maybe it can even help to motivate children and keep them on track of their personal journey in effective learning. But AI is built by humans, and it mirrors biases of its creators. Digital technologies are just tools and no complete replacements to dysfunctional institutions or politics and outdated concepts like teacher-centered frontal instruction with “talk and chalk”.
But surely one can imagine also innovative peer-learning groups that use digital assistants in their collaboration and solve real-life problems. Therefore, a cautious optimism is warranted. What can be automated, probably will be automated by AI. But people all around the world shall have a say in how exactly this will be done and to what extent — keeping a good measure and balance between performance, privacy, freedom and sustainability. Sustainable Development Goals are also an attempt to strike such balance between prosperity and sustainability, between current generations and future ones. Between globalization and protection of unique cultural heritage.
Aristotle said that “Every virtue carried to the extreme, is a vice”. In our age, when researchers are pursuing the creation of “Safe AI”, this quote seems relevant as ever.
