There's no sound in space🪐 [HTC #28]
...but the volume inside of this newsletter is astronomical
Hello, World!
Hold the Code is back for our 28th edition, bringing you a wide range of topics relating to AI, technology, and ethics.
This week, we discuss the sale of NFTs and how it helps and hurts the artists who create them. We also dive into how machine learning can be used to understand the behavior of black holes. Our weekly feature this week discusses the potentially benefits and drawbacks of AI in higher education. With these stories, there’s bound to be something that piques your interest.
Happy reading!
For Sale or Not For Sale
Art for sale, but not quite
NFTs, or “nonfungible tokens”, is a booming technology rooted in blockchain (to learn more about what an NFT is, checkout our 15th edition). In recent years, NFTs have exploded in the arts community because it enables the sale of digital ownership of artworks but not the art itself, giving artists another way to make $$$ from their art (potentially a lot of it!) without giving up their copyright of their work.
Blooming sales… and piracy
While the NTF market is exploding, systems to verify the legitimacy of a seller’s ownership of the NTF are struggling to keep up. Digital pirates — often bots — are stealing digital art online and passing it off as their own to sell. Artists who are not aware of NFTs are the most at risk of this piracy, since they are unaware of the trade and are not actively looking out for unauthorized sale of their artwork’s digital ownership.
Easy steals, difficult saves
Currently, the onus to report stolen art rests on the artist. On OpenSea, one of the most popular NFT trading platforms at the moment, sellers do not need proof of ownership or use their real name to sell, but artists must provide proof and use their real name to file copyright claims - which itself is an extremely cumbersome process.
With thousands of malicious bots looking to steal artworks for unauthorized sale, how can we develop systems to safeguard against this new type of piracy?
Astronomical Algorithms
The problem with black holes
Black holes are a big mystery for physicists. We know that two orbiting black holes can merge into one massive black hole — this cataclysmic event releases gravitational waves into space, which we can detect. However, we can’t see these mergers, so we don’t know exactly how they happen. Scientists must rely on complicated, error-prone mathematical models to predict how merging black holes behave.
A new method
However, a team of researchers may have developed a better approach: a machine-learning algorithm that can take gravitational wave data from any merger and create a new model for the black holes’ motion. Trained on a dataset of existing models, the algorithm can infer the physics governing how the black holes will move and output a simple equation with the same accuracy as models that take weeks to run on supercomputers. In fact, researchers found that the algorithm’s models can be applied to more complex black hole systems than any existing models are able to handle.
"We have all this data that relates to more complicated black hole systems, and we don't have complete models to describe the full range of these systems," said Brendan Keith, one of the researchers behind the algorithm. "Machine learning will tell us what the equations are automatically. It will take in your data, and it will output an equation in a few minutes to an hour, and that equation might be as accurate as something a person had been working on for 10 to 20 years."
The researchers will need to test these models further before their algorithm can be used, but they have hope that this new approach could change the way scientists study deep space phenomena. In fields where humans are largely in the dark, machine learning could help us uncover the mysteries of the universe.
Weekly Feature: Student, Teacher, Robot
AI in higher education is not new. Many of the administrative decisions surrounding finances, investing, student recruitment, and operations are handled by AI. However, many research projects are currently looking at how AI can facilitate learning in higher education — which also raises questions around how to eliminate bias in the algorithms responsible for this learning.
Assessing Impact
Many of these research projects center around adaptive learning, in which AI acts as the orchestrator between the student and the material they are trying to learn. The AI, in theory, can help the student learn more efficiently by adapting to their needs and goals. Many of these use assessments to characterize a specific user and then can guide them to appropriate materials. The algorithm will try to show the student concepts they have not yet mastered, while omitting material that the student has already learned.
Susan Fourtané writes in Fier Education:
“The potential and impact of AI on teaching have prompted some colleges and universities to take a closer look at it, accelerating its adoption across campuses…By 2028, the AI market size is expected to gain momentum by reaching over $360 billion, registering a growth rate of 33.6 percent between 2021 and 2028, according to a research firm Fortune Business Insights’ report. ”
Current Projects
Here’s a list of some of the promising systems impacting higher education:
Jill Watson: Developed by Ashok Goel of Georgia Tech, Jill Watson is basically a computerized graduate assistant. This system also saw recent improvements in its applicability to different courses. It is now much quicker to create a ‘Jill Watson’ for a particular course, only requiring a syllabus and a Q&A with the instructor.
Oli: Oli is a chatbot used by the Common App last year to guide students through the college application process. Beyond that, Oli also encouraged students to take care of themselves during the COVID pandemic by listening to music, talking to friends, or taking deep breaths.
AI tutor platform from Google: Last month, Google announced a new AI tutor platform that personalizes assignments, feedback, and help for each user. This system works by first collecting data from educators on learning objectives and then generating activities based on these concepts, such as short-answer or multiple choice quizzes.
Assessing Ethics
All of the buzz around these new projects has also raised questions about the ethics surrounding the use of these systems. AI is notorious for often containing bias in the ways they are implemented, trained, and applied. Especially in education, it's imperative that these systems be built for all users, no matter their background or identity.
Additionally, some have concerns around chatbots being used in education. The systems need to protect the privacy of students. Learners should know that they are not communicating directly with their instructors, but with a computer system, which may change the questions they pose and the information they disclose.
This leads us to ask: how are our institutions using AI in our classrooms? And who is included in the processes of developing and applying these systems to higher education?
Written by Larina Chen, Michelle Zhang, and Molly Pribble
Edited by Molly Pribble