W5 - Artificial Intelligence (AI)
Tags
UX πŸ‘‹
Published Date
Description

The coolest / most interesting use case for artificial intelligence in the near future

"Education is the most powerful weapon which you can use to change the world." - Nelson Mandela
Back in High School I was was always wondering, why we can't just copy this entire generic knowledge we were advised to learn by heart using a USB stick. Why would everyone be required to run through a years long process of learning the same stuff - without actually learning how to learn? When would be the time to develop our individual skills? Who should be in this role of discovering these skills?
Can't we just streamline this entire basic knowledge transfer process? Automated and guided by an AI that learns about the individual student, detects attention spans, adjusts the learning pace accordingly and makes sure every student is on a similar level by providing assignments relevant to each particular student. Assignments that consider real-world problems and let students apply their knowledge in a way which is more meaningful for them.
All that is a challenge, probably no teacher in this world is able to achieve to date. A challenge we as humans can't really control, as there will always be students sitting in the last row, not paying attention or belonging to a rather submissive or silent personality type swallowed either by the extrovert loud mass sitting in the front or the competent ambivert. And maybe that's the right time to reveal something about me: I am also "ambivert" - someone who "exhibits qualities of both introversion and extroversion, and can flip into either depending on their mood, context, and goals."). This already helped me a lot on a personal level and explains why I'm more keen to network with people I feel related to and thus feel more comfortable to talking to. However, whenever I sensed in-confidence about certain content in school, such as presenting results in math, sharing my literature review or individual sports examination, my introvert part came to shine and I deployed measures such as looking at the other side of the room, pretending not to be there at all.
The status quo in education appears to be following some economical methodology. A free unregulated market, where "the winner takes it all". Where the "invisible hand" decides on who wins and who is going to end up being trapped unemployed in an infinite downward spiral of life taking unreflected decisions in life, finance and politics and after all blames the state by sharing conspiracy theories.
πŸ€·πŸ½β€β™‚οΈ
But what if an AI continually learned about individual learners, adapt to their individual needs, being able to detect extraordinary skills and e.g. promote the upper 20% of math class with extra assignments?

Hidden biases: Detecting and omitting racial biases by design

The news that criminal justice algoriths were found to mislabel African-American defendants as 2 times more risky than white defendants went around the world.
But let's look at education again. In the the United States, inequality in education is that big, they invented a scoring model for it. This "hardship score" was created to enable "admissions officers should also consider context like personal essays, teacher recommendations and family background." and was based on the individual's background such as familiy and neighborhood figures.

The purpose of scoring students

This scoring model, which was intended to improve underprivileged students' situation however came with huge criticism regarding privacy / biases and students from privileged families attempting to exploit the system through bribery, which after all enabled them with benefits reserved for students with special needs.
The fact there is an opportunity gap, for instance between white and black students is just one aspect that needs to be taken into account when designing an AI-directed education system. We need to learn from the human's biases we already see in nowadays' teachers, before we can teach an AI to filter out those aspects.
Β 

Remove all the data, that could lead the AI to develop group-biases

"AI can help identify and reduce the impact of human biases, but it can also make the problem worse by baking in and deploying biases at scale in sensitive application areas." - https://hbr.org/2019/10/what-do-we-do-about-the-biases-in-ai
Instead of revealing an individuals' identity, every student would authenticate with an anonymous identifier, might have a customized page for their subjects where personal preferences are mentioned.
Instead of blant notebooks, students would build their own colorful world around every subject. A student indicates to like animals? The AI would create individual math assignments that would follow the curriculum but be spiced up with stories containing animals. This would allow a huge flexibility in tailoring assignments, while maintaining a high level in quality and giving students a more practical purpose for what they are doing. One of your students mentions early excitement for business? The AI would dynamically serve that and provide the students with adequate practise in Excel while recommending relevant courses going beyond the basic education.
Students might still be sitting in one classroom learning for a certain subject, mentored by a former teacher but with the AI serving individual assignments tailored to individual needs, everyone is working on different tasks and needs to prove their creativity in solving them. No cribbing from neighbours anymore. Students work on projects where a AI detected they might appeal to them and present their results to everyone so there can be an active exchange.
Β 
Me must design the AI in a way it omits racial biases and rather works based on the individual learning behaviour. For instance - If reading takes too long, students would be prompted to do a dyslexia test and receive according material. Instead of reading, the same assignment could be based more on podcasts than literature & reading. If the system detects concentration problems, it would provide tips for a better learning environment rather than questioning the family / social enviromnent. There must be no discrimination in grading or opportunity anymore.
Β 
"If knowledge alone becomes less important not enough anymore is the ability to apply that knowledge"
(After watching this, I am again happy being part of a University of Applied Sciences πŸ₯³)

Question: The perfect child

❓
Would you send your child to an AI-dominated school although you knew the experience might be experimental, yet the results held out in prospect are brilliant (e.g. higher learning outcomes in shorter amount of time)
Β 
πŸ“‘UX & Technologies blog