Automated Education: A Design Fiction

Academic case study imagining futures with automation; 2018

How will education and admission to higher education change with the advent of more pervasive or normalized AI assistants? This project explores how AI assistants might subtly hack the realities of teenagers entering higher education, surfacing issues of free will, personal exploration and expression, future of education, and future of work.

The process of getting into top colleges, heretofore has been rather opaque. With the shift in higher education to support rapid changes in industry due to the rise of AI, people are competing for admittance into industry-focused, accelerated bachelors degrees at top universities. These degrees are highly competitive, and while there are many micro and nano degree options, these accelerated bachelors from top tier universities are tailored to yield high paying job in competitive and emerging job markets.

To help high school students get into these competitive programs, AI Innovations created College Companion, a personal AI that helps guide students into their ideal college programs.

With so much data available about people’s activities and behaviors, College Companion aggregates data and behaviors from people who are admitted into top-tier college programs and patterns that information into actionable steps which guide the behavior of students seeking admission. College Companion works as a sort of life experience optimizer, or LEO, guiding students on to the best possible paths towards their program of choice. It helps them achieve things and secure opportunities that place them them at the top of the ranking algorithms that help decide who is admitted to these competitive universities.

Meet Rachel. She wants to study biodata engineering in an accelerated bachelor's program at The Georgia Institute of Technology, in order to become a design engineer for DNA data storage networks. This program only admits 5% of applicants and Rachel needs every advantage she can get.

Several weeks before her Junior year of high school begins, her parents give her a College Companion to optimize her life and greatly increase her chances of admittance to Georgia Tech, the school of her dreams.

When she installs and syncs with college companion, it greets her with an analysis of her skills based on data it has collected about her from her transcripts, extracurriculars and community activities, social media and other personal online histories.

“Rachel! So pleasant to meet you! You have demonstrated a great aptitude in many things, but it seems you are drawn to biology and really excel in that field. However, to reach your goals you could use more experience in data structures and math. I know you want to get into Georgia Tech’s Biodata Engineering School, I’m here to help you by accentuating your strengths and strengthening your weaknesses. Shall we begin?”

The next week school starts. College Companion helps Rachel pick her classes by looking at patterns of classes taken by former Georgia Tech admitted and their average GPAs as well as noting which classes are required by Georgia Tech for admittance into the biodata engineering program. College Companion uses those patterns to suggest classes to Rachel, informing her of the minimum grade point average she must achieve in each class to be a successful applicant.

Compared to the people enrolled in Georgia Tech’s Biodata Engineering program, Rachel is a little lacking in volunteer experience so College Companion finds and suggests a local volunteer opportunity that matches Rachel’s interests and goals at a local youth organization that explores science through art.

Winter Quarter of her Junior year, Rachel doesn’t meet her minimum GPA in AP Applied Physics. College Companion quickly runs an analysis of what makes people with slightly lower GPAs still competitive enough to get into Georgia Tech. College Companion suggests that Rachel enroll in a biodata structures MOOC (massive open online course) taught by a biotechnologist at Georgia Tech. College Companion has observed that Rachel’s competition and many of the previous applicants to Georgia Tech have enrolled in extracurricular micro degrees while still in highschool to make themselves more competitive.

In addition to the extra online classes, in order to correct her slight GPA imbalance, College Companion notes that the application algorithm to get into Georgia Tech gives some weight to applicant’s social circles, slightly favoring applicants with high-performing academic friends. College Companion noticed that Rachel’s friend Mara’s grades are slipping and there are pictures of her in detention in her social media feed. College Companion has a heart-to-heart with Rachel about how if she really wants to go to Georgia Tech she should start distancing herself from Mara. Rachel slowly does this.

College Companion knows that most accepted applicants to the accelerated biodata engineering degree at Georgia Tech have some industry experience while still in high school. During the summer before her senior year College Companion secures Rachel an e-learning internship with the same Georgia Tech professor who taught her MOOC earlier that year. This internship also helps Rachel begin to build a relationship with a hopefully future professor.

During her Senior year, College Companion helps Rachel study for her SATs and quizzes her until she gets a score that slightly exceeds scores of other Georgia Tech admitted. It also helps he write a memorable statement of purpose that has the same type of heartfelt message as other applicants which is tied it to her academic interests.

Rachel applies to Georgia Tech’s accelerated biodata engineering bachelor's program and is accepted thanks to all the hard work and strategic planning she did alongside College Companion. She’s never been happier to have something like College Companion in her life.

Problem Space:

In this design fiction, I explore a hypothetical future where AI assistants are used to optimize the chances of high schoolers to get into desired universities using data about previously admitted students. This AI organizes and guides prospective high school students through their final two years of their high school education. If the student follows the AI assistant’s advice scrupulously, they will have the highest chances of getting into the highly competitive, industry-focused accelerated bachelors programs that prepare students to work at the few remaining high-skill, high-pay jobs in an economy where much of the workforce has been replaced by automation.

I arrived at this space after going through several stages of ideation and iteration starting from a place of asking how ‘reality hacking’ might take place in the future. Although ‘reality hacking’ started in a more obtuse form (see my process of refinement below), it slowly morphed into this more subtle (and more likely) scenario that comments on the influence of AI and Algorithms in agency in day-to-day life.

Click through the slide show above to read the entire story.

ANALYSIS

I synthesized trends in higher education and continuing education, the future of work, trends in AI and some trends in technology, visualized below.

Mind map of relationships between trends in higher and continuing education, future of work, future of tech, and AI.

My World Building Process

This was my first go at world building, which meant it got messy. I followed a chain of worlds and interventions to my final world above, but the main theme that helped keep them all together was the idea of reality hacking. Below I detail my process of shifting between world and solution to new world until I reached my final, reality hacking world where teens are being subtly hacked by AI assistants to receive competitive advantages in higher education.

Iteration 1

Reality Hacking in a World of ‘Unbiased’ AI Generated News:
In a post-truth era, AI written news from primary sources has become standard. However, people have begun hacking reality with 3D printed prosthetics and adversarial images (like a bunch of 3D printed carrots look like a gun to an AI reading a video feed), staged violence, fake blood, etc. I then discuss possible methods for solving this problem: biohack back, self-aware AI, and AI lead discourse about truth in the news.

Iteration 2

Ways to detect Reality hacking:
My first iteration was an attempt at having more automated conversations around where news was coming from and it’s credibility. I created gifs that very roughly explored how in a ‘futuristic’ newsfeed on social media of some sort, one could be reminded that not all news is created equal. These conversations or reminders were guided by an AI, either through the help of an AI assistant, or through a type of symbol and mapping of verifiability.

Scrolling through a news feed,
Clicking on a dubious news story,
Exploring its roots and why it might be compromised.

Scrolling through a news feed,
Being intercepted by an AI assistant
who warns about nefarious news.

ITERATION 3:

Hyper-Subtle Reality Hacking through Topographic Data Analysis
& AI Targeting

After a conversation with my professor, we realized the potential of Reality Hacking could be much more subtle. How might automation or big data allow people to shift reality through subtle interjections/intersections. To this end, I began to imagine scenarios where reality hacking was accomplished through probability, topographic data analysis, subtle placement of objects and personalized, automated news surfacing algorithms. In this iteration I became interested in assemblage theory, topographic data analysis, and hyper-subtle systems-influencing methods.

Iteration 3:

The Forensics of Solving Reality Hacking
I had intended to take the hyper-subtle methods of reality hacking and explain ways that detectives might unravel these instances through pattern analysis. These gifs and images are evidence of that line of thought. The carousel below explores ideas such as machine learning expression reading for hints of dishonesty, tracking heart rates to see if patterns of excitation match patterns of action (was their heart racing before they even instigated the ‘crime’), and antagonistic 3D printing of things like a bunch of carrots that are read as an automatic weapon by computer vision. The images below also represent a visual language for the story I was about to tell that has been synthesized in my final reality hacking world and story at the top of the page.

However, I had a last minute realization that reality hacking is, to some extent, already happening with algorithmic suggestions, advertisements and etc. What if this was translated into a story where those types of algorithms began to erode basic human values like freedom exploration in education, freedom in adolescence, liberal arts education, etc. I therefore shifted from this reality hacking storyline to a storyline closer to our every day lives where data and automation help college students get into prestigious programs.