Will AI get it wrong for rights?
BCR Law Family Law Legal Assistant Aurore Talazac, who is currently studying for a Double Degree at the Insitute of Law Jersey, recently took part in the Institutes Academic Conference where she researched and presented on the impact of Artificial Intelligence on Human Rights.
Aurore’s presentation was extremely well-researched and received positive feedback. Here we share with you her presentation.
INTRODUCTION TO AI
Artificial intelligence (known as “AI”) has developed significantly over the past few years. It has gradually infiltrated aspects of our daily life, such as through the use of facial recognition on our phones, or by determining what information (and adverts) to present to us on our social media feeds based on the history of our online interactions.
AI is a technology created to automate tasks that a human normally does, it is supposed to imitate human intelligence by using cognitive processes. Marvin Minsky declared in 1968 that AI is ‘the science of making machines do things that would require intelligence if done by man’.
While it increases its presence in our daily life, we can also start to see it more in specific areas such as the legal field or the armed forces.
AI AND CONCERNS WITH HUMAN RIGHTS
The pace at which AI is developing, and the multiplicity of the uses to which it is put, raises a number of geopolitical and human rights issues. As a consequence, it puts an emphasis on the importance of the interpretation and development of Human Rights in order to meet some of the challenges that AI presents. It is necessary to assess the role of Human rights in relation to Artificial Intelligence and the way its challenges can be resolved.
In particular, because the development and deployment of AI does, or at least has the potential to, affect almost all human rights, from the right to live with the use of AI in the medical field, which could help detect conditions such as cancer early stage to the right to political participation, in the areas of voter engagement and turnout for example where it can be used to analyse voter data and predict who is most likely to vote.
There are three “hot topics” of concern in relation to AI that frequently pop up in public discourse on this subject, namely:
- Discrimination, for instance in the area of hiring, AI-powered systems can be used to analyse job applications and resumes, with the aim of identifying the best candidates for a given position, however, if the data used to train the AI system is biased or reflects discrimination hiring practices of the past, the system may perpetuate these biases, leading to discriminatory outcomes.
- Automatised arms used in recent conflicts such as ‘killer drones’, this has been seen in the recent conflict between Ukraine and Russia where a lot of people were killed by those killer drones and the reality of it is that civilians have been hit in strike after strike as targets are misidentified
- The right to education, is important in the context of AI, in particular, because it has the potential to exacerbate existing inequalities in access to education.
This article will focus on the impact of AI in the world of education. The right to quality education is one of the 30 Human Rights contained in the UN’s Universal Declaration of Human Rights (Article 26).
WHY THE RIGHT TO EDUCATION IS IMPORTANT, AND ITS RELATIONSHIP WITH AI
First, it is necessary to consider what the right to education means and why it is important.
The right to education is a right guaranteed by international human rights law. The Universal Declaration on Human Rights, adopted in 1948, proclaims in Article 26: ‘everyone has the right to education’.
According to UNICEF, education is a cornerstone to reducing poverty and child labour. It is important socially and culturally. It helps promote democracy, peace, and tolerance.
Education is what forms the future generation, it helps students to be more independent, to learn how to work as part of a team, how to research and analyse data, and how to think critically.
According to article 29 of the Convention on the Rights of the Child, education provides a child with a place to develop his personality and talents as well as his mental and physical abilities.
It also teaches respect for human rights and fundamental freedoms and ensures an open-mindedness to the world by preparing him for a life in a free society.
Essentially, what all of this means, is that education will shape tomorrow’s minds, it will shape future generations and will create new values and ideas, this is why we need to make sure it is not jeopardised by the introduction of AI into education and education systems. That is why it is essential to address AI in education.
Returning to Marvin Minsky’s definition, outlined earlier that AI is ‘the science of making machines do things that would require intelligence if done by man,’ it can include solving calculus problems or understanding short stories, learning new concepts, and even diagnosing diseases. Usually, AI is better for a technical task rather than reasoning or perception of things such as understanding the language but there is no doubt that AI can have a profound effect on the experience of education in a range of ways
AI already has a significant influence on the right to education. For instance, it is used by some universities during the admissions process. This can run the risk of discrimination if the data upon which the AI relies in order to make its decisions is itself biased. The use of previous data relating to university enrolment might lead an AI program to perpetuate discrimination which was, in origin, human.
For example, if most previous students admitted to the university were affluent white males, and that is the data which the AI uses to make decisions, the AI will be biased and discriminatory. An example of this in action was the Microsoft ‘chat bot’ called Tay.
Tay was set up by Microsoft on 23 March 2016. It was a computer-generated personality designed to simulate the online musings of a teenage girl. After a number of tweets by online ‘trolls’ the AI behind Tay reflected the offensive messages of real humans to pour forth AI-generated hate. In the 24 hours it took Microsoft to shut Tay down, the chatbot had abused President Obama, suggested Hitler was right, called feminism a disease and delivered a stream of online hate.
AI, EDUCATION AND CHAT GPT
“There’s always been this concern that technologies will do away with what people do best, and the reality is that people have had to learn how to use these technologies to enhance what they do best,”.
We can see AI introduced through our educational system via diverse programs. One launched recently, in November 2022, has fast become one of the well-known; ChatGPT and in only 4 months it has had a huge impact on education.
ChatGPT is a brand-new language processing tool driven by AI technology, it can answer questions and can assist with tasks such as writing e-mails, essays, and codes. It was updated recently with GPT4, this new technology has been the focus of a lot of media attention, in particular in the educational field. A number of other platforms such as Copilot on Microsoft Teams 365 are currently testing their own versions of ChatGPT. Its popularity cannot be doubted, it reached 1M followers in 5 days when it took Instagram 2.5 months to reach that.
So, does AI risk affecting the quality of education through the development and use of programs like ChatGPT?
Some argue that ChatGPT will be the end of education. Others consider that it can be a massive tool that will enhance major access to education and to resources to further it.
The University of Georgia in 2019 thought that over the course of the next five to ten years, it would not be possible to get a strong AI matching higher-order human abilities such as abstract reasoning, concept comprehension, flexible understanding, general problem-solving skills and the broad spectrum of other functions that are associated with human intelligence. Only 3 years later, Chat GPT was launched. This met almost all those abilities. GPT4, an advanced version, has already proven its abilities by getting a 163 on the LSAT and scoring into the top 10 percentile of the Bar exam.
CHAT GPT AND THE RIGHT TO EDUCATION
Now that I have briefly explained the main concepts this article focuses on, we should get into the core of the subject which is how will ChatGPT impact the right to education.
As a quick reminder, the right to education is the right of all individuals to receive a quality education without discrimination or exclusion, which includes the right to access education and to receive equal treatment and opportunities in education.
It is essential for the development of individuals and society as a whole by enabling individuals to acquire knowledge, skills, and competencies essential for personal growth.
POSITIVE IMPACT OF CHAT GPT ON EDUCATION
With that in mind, we can already see how ChatGPT can have a very positive impact on the right to education in relation to its technical aspect, such as the access to education, in particular with the use of online platforms when the access is limited by factors such as geography, infrastructure, and resources.
There are some potential benefits of using AI language models like ChatGPT in education.
Firstly, it can enhance learning, ChatGPT and other AI language models can assist individuals by providing personalized and adaptive learning experiences to students and catering to their specific needs and learning styles. The Khan Academy, a non-profit educational organisation offering online resources and with whom ChatGPT has already created a partnership with the organisation has used AI to help students learn a variety of subjects.
By being available 24/7, students can have access to learning resources and get help at all times. This can be extremely helpful for students who might need extra help outside of school. It also provides interactive learning experiences engaging students more in the learning process.
It is also arguable that multi-modal learning incorporating multiple modalities such as audio or visual content can help students learn more effectively and engage more in the learning process. Moreover, students can access instant feedback on their progress and performance.
Secondly, it can assist students and teachers with research by providing access to a vast amount of information and assisting them in data analysis. Students can find relevant information quickly and efficiently, saving a significant amount of time and effort (it can also help analyse patterns in the data).
Moreover, it can translate research materials from one language to another giving wider access to resources to students and educators.
Finally, it can also be argued that ChatGPT and AI language models can fill the gap between minorities in relation to the right to education by giving everyone access to the same resources and fighting against discrimination.
Overall, AI language models like ChatGPT have the potential to enhance the educational experience for students as well as educators by providing personalised learning experiences, research assistance and accessibility features.
NEGATIVE IMPACT OF CHAT GPT ON EDUCATION
Nonetheless, there are also significant limitations to what can be achieved with AI like ChatGPT.
Firstly, AI language models are trained on large datasets, which can introduce biases and inaccuracies into their responses, also ChatGPT data-based stopped in September 2021 and therefore it will not be helpful for research on recent events.
Some fear that an overreliance on ChatGPT and AI language models will reduce critical thinking and analytic skills in students as well as reduce their creativity and innovation.
In relation to critical thinking and analytic skills, if students use ChatGPT as a shortcut to generate ideas or arguments without thinking, it could inhibit their ability to develop these skills, therefore it is important to address that ChatGPT is not a substitute for critical thinking and analysis skills.
It can also have a negative impact on writing and reading skills. Again an overreliance on ChatGPT will lead to a lack of development of skills such as spelling, grammar and sentence structure. Some studies have addressed the subject, such as a 2020 study published in the Journal of Writing Research, which found that using digital writing tools was associated with decreased motivation for writing among high school students, compared to the traditional use of pen and paper.
ChatGPT may also have a negative impact in relation to research skills. By having access to a huge amount of resources in only a few seconds, students will (it is claimed) always choose the easy way out and over time, it’s predicted that they will not be able to search on the Internet properly themselves or to filter information to identify what is relevant.
Confidence is also something that needs to be addressed, I strongly believe that academic achievement is a huge factor in giving a student confidence in their life and their future work life. By relying on a tool like ChatGPT during their studies, a student will never experience the pride of having spent hours researching on a subject and getting a good grade for it, nor will they be able to share their view on the concepts researched. Without that sense of achievement might it not be argued that the confidence which flows from it will be lost and that young adults entering the world of work will experience ‘imposter syndrome’ and question their abilities.
Also, even though it can democratise student resources that were only available to a few, it can also have the opposite effect and create an even larger gap in the right to education. The obvious pitfall here is that ChatGPT and AI generally require the ability to access technology, power and the internet. Those that do not have that access will become materially more disadvantaged compared to their peers who do. This potential issue would be more severe if ChatGPT became a fee-based service. This is something to think about as it had been mentioned by Open AI.
And obviously, one of the biggest concerns is the dependency on technology it might create. Early exposure to technology in general can have a negative impact on student development such as speech delays, reduced attention and increased risk of obesity. Moreover, it can impact social and emotional development by decreasing time spent in face-to-face social interactions. It can also raise concerns about cheating and plagiarism.
Overall, all of those negative aspects are breaching the right to education in a lot of different ways.
The risk of having biased data is clearly contrary to the right to have quality education. As mentioned before, education helps students to be more independent and teaches them how to research and analyse data as well as think critically. We have seen that one of the biggest concerns around AI is the issue of it inhibiting students researching skills or analysing skills which undoubtedly threatened the right to education and in particular Article 29 of the CRC. Moreover, the risk of AI language models like ChatGPT. becoming a fee-based service will jeopardise access to education for a lot of people and violate article 30 of the UN Charter.
NEXT STEPS/ REGULATION/MY RECOMMENDATIONS
Overall, technology and AI language models such as ChatGPT can provide many benefits. There are, however, serious risks. I strongly believe that the use of ChatGPT should be supervised and strictly controlled.
This was already done in New York, the New York City’s Department of Education restricted access to ChatGPT on school networks and devices, worrying about the negative impact on students learning. They are already trying to develop methods to detect the use of AI in the education system by recognising patterns and lack of emotional intelligence in the answers.
So far, however, other countries have not followed suit and France has even accepted its use during an exam without declaring it cheating. In Lyon, a professor gave a grade of 11,5/20 to the students which used ChatGPT and Alain Gay, coordinator of the plagiarism committee declared that the use of ChatGPT cannot be compared as plagiarism, it is more relevant to identify it as ghost-writing.
I don’t see how we could forbid its use completely and do not think that we should. It is indeed a very useful tool or has the potential to be so, which can save a lot of time and stress to students, educators and researchers as well as provide support outside of the classroom to those who may not receive that support from any other source.
Nonetheless, there is a lot of drawbacks which cannot be overlooked, ChatGPT can have a very negative impact over time and because of that, it needs to be regularised for exam seating and graded work so the student will all be held to the same standards without raising the issue of cheating or plagiarism.
At the end of the day, it is up to the students and individuals to resist the temptation of using ChatGPT to draft their work and to use it as what it was designed for, a tool helping them to be more efficient and saving them time. Nonetheless, it is possible that it may interfere more with the right to education of disadvantaged students who are more intensely trying to catch up with their peers who may have had access to better education before. Those individuals might be more tempted to use AI language models such as ChatGPT to do their assessments and if so, the right to education is likely to be interfered with for that group of people.
I would like to end this article with this quote from Elettra Bietti who said that “Work is more than the output of a task, it is how humans learn and find fulfilment. You can use AI to learn, but it is important not to forget that delegating research to a system means removing the opportunity for students or junior lawyers to learn from doing the research. Those people will not be learning the same way anymore.” She clarifies, “They’re going to be learning a very different skill, which is how to work with machines to produce outputs, and how to achieve some fulfilment in the process.”