How AI could make the world fairer for presented kids
A team of researchers from Las Vegas recently developed an AI model that can be used to assess the ability of a human pianist. On the surface, it might not sound like some kind of breakthrough research that could change the world overnight. And it probably isn’t. But it is a crucial component in a stack of machine learning that could make the world fair to bright children.
Education doesn’t work the same for everyone. For example, in the United States, most public school systems have programs to identify so-called “gifted” children. Unfortunately, there is no consensus in academia and government about what exactly constitutes a “gifted child” or how they should be treated. In essence, it is a jack-of-all-trades, with programs often invented at the institutional level and implemented without supervision.
In some cases, children classified as the highest percentile learners are given access to tailored teaching. However, this is far from the norm, especially in impoverished or poorly populated areas. Gifted children are often forced to fit into traditional educational paradigms.
This results in a significant group of children waiting more time for the other children to finish than they are to be educated. The answer, many experts would agree, is more one-on-one teaching. However, budget and staff constraints make it unlikely that we will soon solve the problem of raising gifted children by traditional means.
[Read: How this company leveraged AI to become the Netflix of Finland]
This is why this paper about the above mentioned AI that can judge a human pianist’s abilities is so interesting.
The team wanted to find out if the AI could accurately determine whether or not a human pianist was a skilled player. They found that models trained on both video and audio outperformed models trained on a single input, and then created a novel data set. According to the paper, the team managed an accuracy of about 75% against human judges.
The most interesting thing about this AI project is not the potential to eventually become a piano judge on a human level, but the potential as an information module in a single AI / student teaching paradigm.
With proper hardware configuration, this can be used to assess human piano performance in real time. And it could probably be configured to give instant feedback and even provide spontaneous suggestions for improvement. And all of this should be relatively easy with modern technology. You can even use a language model trained on teacher-student interactions and name it like Piano Teacher Bot 3000.
It’s easy to imagine the evolution of similar AI models for other school activities, such as arts and sports, where human performance is easier to quantify in the real world than in the digital. And for things like math, chemistry, and grammar, an AI that is able to answer their questions in the clear and move them to the next lesson could help faster students avoid constant boredom.
There is nothing like it in the current one-way digital instruction paradigm. The current focus is on developing systems that work for most students and preventing fraud – two issues that are not necessarily that important to the highest percentile learners.
An AI-based real-time observation and feedback system would have the additional advantage of learning from its own data loops. While long-term studies in the US on the effectiveness of a particular educational paradigm are difficult due to general inconsistencies in the educational system, a self-contained AI system could generate enormous amounts of useful data on a relatively small scale of use.
There are, of course, myriad ethical and privacy concerns associated with the idea of using an AI system to monitor and engage students.
But the upward trend could be huge as long as it is done ethically. It is far easier to put AI in classrooms than it is to solve the current budget and staffing crisis that is occurring in US schools across the country.
Published on January 25, 2021 – 21:58 UTC