How to use AI in learning: A responsible approach

by Tess Robinson

Working in digital learning is never dull. There have been so many advances in technology since LAS was established nearly 20 years ago, that we’ve lost count. This constant state of technological flux is what makes it an exciting and interesting area to work in.?

AI is, without doubt, the most seismic technology change in years, but there’s a noticeable change in the narrative around AI in the digital learning industry in recent months. It’s not (yet) proving to be the total revolution that we first thought and it certainly isn’t looking like it’ll replace humans wholesale any time soon. We seem to be reaching the peak of the Gartner Hype Cycle ?and maybe even starting to head into the ‘trough of disillusionment’. Perhaps this is a good time to take a pause and reflect on what our relationship with AI can and should be?

As a purpose-driven organisation, we’ve had to think carefully about our stance and policy on AI, particularly given the environmental impact of data centres and the way that some LLMs amplify bias. Given the potential impact on society and the environment, it’s important that we’re aware of the negative implications, as well as the potential positive impacts so that we can use it responsibly.

The Green Web Foundation published their ‘ Thinking of Using AI’ report in October - it’s long but worth a read. They identified some key actions for businesses who are considering using AI. Here’s my top two:

1. Question your use of AI

Is AI going to solve your business problem better than something else? Or are you just jumping on the bandwagon? AI isn’t always the answer, even though a multitude of vendors will tell you that it is. Does the increased productivity that you want to achieve by using AI, outweigh, for example, increasing your carbon footprint through its use??

Knowing when to use AI and only using it for the purposes it’s best suited to is important. At the recent LN Connect conference in London, Toby Harris and Ash Mehta talked about the danger of AI generated learning content becoming like junk food - ubiquitous and not very good for us. AI can be a transformative tool, but only if we use it in a considered and purposeful way to enhance our human efforts and experience, rather than detract from it.

2. Optimise AI use and use it responsibly

Once you’ve determined that AI is the right choice for your task, choose your supplier carefully. Not all AI is created equal in terms of social responsibility. In my experience of digging into their data usage policies, gen-AI suppliers are not always the most transparent. It can be really hard to find information on where the data is hosted, how it’s used, what the carbon footprint is and what safeguards they have in place to mitigate bias - but that doesn’t mean that we shouldn’t try to find out. The more people who demand it, the more likely this information is to be released. Regulations and legislation such as the Corporate Sustainability Directive (CSRD) and Energy Efficiency Directive (EED) will also mean that large providers will have to make this data available. This will help us in making informed choices on who to use.?

Alongside the two takeaways from the Green Web Foundation, here are some other things to consider when using AI with purpose:

1. Establish ethical guidelines and principles

If you’re developing an AI product yourself, make sure you establish your own clear ethical guidelines and principles for its development and deployment. Increasingly, your customers will be asking for this anyway, so it will give you a competitive advantage if you’ve already got this in place

2. Audit for bias

Regularly check any AI generated content for bias. Look at it through different lenses before publishing it, to ensure that you’re not inadvertently perpetuating harmful stereotypes. For example, a recent LinkedIn post from Ashley Sinclair at MAAS Marketing highlighted how gen-AI image generation is still creating gender-biased images, where women are unnecessarily sexualised - we need to be alert to things like this.?

3. Be hot on data privacy

Know where your data is going and what it’s being used for. Many organisations just haven’t got a handle on this yet and it’s a huge security risk. Make sure you have consent from your customers, or individuals whose data you want to use in AI powered applications, before you use it.

4. Keep the human in the loop

Incorporating human-centred design principles into the development of AI products or the use of gen-AI in content generation will ensure that they are user-friendly, accessible, and beneficial to all stakeholders. AI is increasingly being viewed as a tool to enhance human creativity, rather than something to replace it - keeping a human in the loop helps us to get the most out of AI, in the most responsible and inclusive way.

5. Be transparent

Communicate openly and transparently with your stakeholders about your use of AI, including its benefits, limitations, and ethical considerations. This also opens channels to address any concerns and build trust.

The world of gen-AI is fast moving. It’s easy to get caught up in all the excitement. This can make it hard to establish a healthy relationship with gen-AI and make sure you’re using it appropriately and responsibly. As the hype is starting to cool, this is a good opportunity for us to reevaluate our use of AI and make sure it aligns with our organisation’s mission, values and commitments to social and environmental responsibility.?