AI Ethics and Data Protection for Learning

The recent resignation of Dr Geoffrey Hinton, the so-called Godfather of AI, has once again highlighted the complex ethical issues surrounding artificial intelligence. Dr Hinton resigned from his role as the head of Google Brain with dire warnings about the dangers ahead. ‘It’s hard to see how you can prevent the bad actors from using it [AI] for bad things,’ he declared.

Ethics and data protection are at the top of the list of concerns for many. As advanced as AI systems are, they do have their shortcomings. AI tools rely on data to learn and make decisions. Many of us have been free and easy with our personal data, handing it over to organisations without much thought. And with that comes the risk of hacking and the misuse of personal data.

At its core, the Cambridge-Analytica-Facebook scandal involved the misuse of personal data. The data science company used AI to collect information from Facebook profiles and used targeted adverts to try and influence users’ choices at the ballot box. The fallout from the scandal caused severe harm to the reputations of both companies. And Cambridge Analytica was forced to shut down soon afterwards.

Furthermore, AI models trained on incomplete or biased data will produce results that are also biased. An embarrassing example of when it can all go horribly wrong came in 2015. Google was forced to apologise when its image recognition software tagged photos of Black people as ‘gorillas’.

This post examines the topic of ethics and data protection in learning environments. We look at the burning issues and suggest ways L&D providers can ensure they act responsibly and ethically.

L&D systems collect vast amounts of personal student data about behaviour, preferences and performance. And while AI models can put this data to good use with personalised student learning and development plans, data collection comes with responsibilities. Protecting the privacy and security of users’ personal information is vital to prevent unauthorised access or misuse.

Ways you can safeguard confidential data include robust encryption and secure data storage. Moreover, limit access to information to only those that need it and keep a log of who uses data and when. That way, you have an audit trail if any security breaches occur.

As the use of artificial intelligence has become more widespread in learning and development, the amount of data collected and processed has similarly exploded. And so, too, has the potential for bias and unfair treatment of groups or individuals.

AI is only as good as the data and algorithms that power it. If these are biased, AI can perpetuate the existing biases in society, including race, gender or socioeconomic status.

Say, for example, an AI learning system is trained to rely on written text. This might adversely impact those students with English as a second language. Or an AI system that uses facial recognition might disadvantage those with different features. Indeed, Amazon got into hot water when its facial recognition system misidentified several members of Congress as criminals.   

You can take steps to safeguard the fairness of your learning system. The first is to ensure that data sets are diverse and representative. Developers should also consider the potential effect of their AI systems on different user groups and take action to address any issues. Finally, regularly audit and monitor AI-driven L&D models to ensure fairness and stay ahead of any unintended impacts.

Another challenge with AI-driven learning systems is that they can be difficult to understand and interpret. After all, these systems use complex algorithms that can be gobbledygook to anyone without a PhD in computing.

AI researchers are working on developing more transparent and easy-to-understand learning systems to address these issues. These tools will clearly explain decisions, recommendations and predictions, improving trust and accountability.

In the meantime, administrators are responsible for ensuring they understand how their AI-powered learning systems work. L&D professionals can then make informed decisions about how the system is applied and used in practice.

The discussion on transparency leads to our next issue of user control and consent. Until recently, organisations and users have been guilty of complacency around how data is acquired and used. However, high-profile data breaches like the Cambridge-Analytica-Facebook scandal have changed that view. Users are now far more aware and concerned about how organisations use their personal information.

L&D professionals need to be mindful of this. Give students clear explanations about why the data is being collected and how it will be used within the learning system.

Furthermore, make sure you obtain learners’ informed consent before collecting and using their personal information. And consider giving students the option to opt out of data collection and provide ways they can modify their profile.

Artificial intelligence is most effective when humans retain oversight and direction. AI should be seen as a tool to enhance human capabilities rather than replace them altogether. After all, people are needed to develop the data sets and algorithms driving the technology. And they are also required to review and correct outputs to improve the accuracy and fairness of AI systems.

In L&D contexts, AI should complement educators and L&D professionals. A collaborative approach helps maintain the right balance between automation and the emotional intelligence and creativity of people.

AI learning systems offer fantastic opportunities to enhance the student experience for learners of all needs and abilities.

It’s already being used to detect a student’s disability and apply an adapted form of learning. For example, artificial intelligence can create content that is more accessible to students with disabilities by using closed captions and audio descriptions for students. And AI-driven learning plans can also be trained to be inclusive and accessible, providing the student with a supportive, personalised pathway.

However, as we have seen, AI relies on data to learn patterns and make decisions. Every learner is different, and L&D professionals should ensure their AI-powered learning systems reflect students’ different styles, abilities and backgrounds. Diverse, representative data sets are critical.

Get it right, and there’s a real opportunity to bridge the digital divide and promote equal opportunities for all learners.

Responsible data usage and retention in AI learning systems are crucial aspects that affect quality, fairness and accountability. 

Data fuels artificial intelligence and needs to be collected, processed and analysed in ways that respect the individual’s privacy, security and rights.

Top of the list is ensuring compliance with data protection guidelines like the EU’s General Data Protection Regulation. This legislation gives users the right to delete their data profiles and opt out of data collection.

When it comes to learning systems, data should only be collected to improve student learning outcomes and experiences. Therefore, think about the data you are collecting and ask yourself whether it’s really needed for that purpose.

If the data is relevant, use data anonymisation techniques so that individual students cannot be identified. And once the data is no longer needed, it should be deleted to prevent it from being misused.

Educators and L&D professionals should develop clear guidelines and policies on data usage and retention. Doing so will provide the accountability and transparency that is so critical. And it will go a long way towards demonstrating the ethical and responsible use of personal information.  

The good news is that artificial intelligence can potentially transform and enhance the learning process. However, it also poses significant challenges and risks. Ethics and data protection concerns have become urgent as AI becomes more prevalent in learning environments. We can no longer stick our heads in the sand and hope for the best.

Here’s a summary of the pressing issues for L&D professionals:

• Establishing clear and consistent guidelines for data collection, usage, retention, and sharing in AI learning systems, ensuring compliance with relevant regulations

• Designing and developing AI learning systems with the input and feedback of diverse and representative groups of users

• Monitoring AI learning systems for fairness, inclusivity and accessibility and avoiding or mitigating unintended biases or discrimination against certain groups or individuals

• Providing transparency around data sources, algorithms, and outcomes of AI learning systems and allowing users to access, correct, or delete their information.