EdTech | Machinelearning | DeepGrade
The objective of this article is to provide readers on how technology [such as machinelearning], AI based solutions [For Example: AI-powered assessment and learning system] and Data are inter-related and how problems such as Biased AI can be solved or avoided in the context of adopting AI in Education.
Machine learning (ML) is the scientific study of algorithms and statistical models that computer systems use to perform a specific task without using explicit instructions, relying on patterns and inference instead. It is seen as a subset of artificial intelligence. Machine learning algorithms build a mathematical model based on sample data, known as “training data“, in order to make predictions or decisions without being explicitly programmed to perform the task. Source – Wikipedia.org
Machine learning is all about learning from Data and machine finds patterns from the data. ML is all about Data and Algorithms, but Data is key. ML offers various model, but it depends on the volume of data one has. Based on the volume of data simple to complex models can be chosen.
Adopting AI-powered EdTech solutions are on rise and will play a major role in next few years, lot of new innovations are on the way towards how to improve AI in EdTech. While this growth and transformation is good for the education industry there is a need to address several factors as they come in. We are going to touch upon one such factor which is Biased AI and how this can be solved.
EdTech platforms and solutions are on the rise and examples to name a few, AI-powered assessment and learning system, AI tutoring, Personalized learning, Experimental learning and AI assisted education or AI led education. Companies globally (countries US, Europe, China and India) are all investing heavily in EdTech. Analysts report that AI usage in U.S. education will grow 47.77% from 2018-2022. The EdTech platforms are going to rely on the data as one of the key source of input to be fed into their machinelearning system. There is a known problem here.. AI systems learn and amplify human biases the problem is when an AI-powered EdTech system was trained on data annotated by humans, the humans’ biases tainted the data, which in turn infected the algorithms, which in turn produced biased outcomes. We call this as Biased AI.
There are two types of bias in artificial intelligence and machine learning: algorithmic/data bias and societal bias. Our focus is more towards algorithmic/data bias (training the AI with biased data) alone for this article. The data bias differs from product to product. Let us take an example of AI based assessment and learning system and see how this can be solved by another AI based solution such as DeepGrade. Consider applying a AI based assessment and learning system which tries to consider the assessment reports which are manually evaluated (which is the process in today’s educational system majorly) to be fed as a input data into machinelearning algorithms. Then there is high possibility of getting Biased AI reports. Primary reason for biasing is varying patterns of manual evaluation when fed into machinelearning system then it tends to learn from errored data and tends to generate biased report.
The AI Bias can be explained with various examples, above is very simple example considering evaluation or grading as a source where the data gets generated which is given as input to machinelearning to generate models.
AI based grading/evaluation solution such as DeepGrade comprising machine learning algorithms can add value to the overall grading or evaluation practice, they could not just replace hodgepodge grading practices but avoid biased AI practice as well.
Reference
