Everybody’s talking about AI. It’s everywhere, TV and media subscriptions, satellite navigation, social media platforms, online shopping, you name it, AI is integrated, and its presence is growing.

AI is a game-changer for learning and online training providers:

• It can gather and create training material in seconds
• It can deliver that training in effective ways
• It can identify skills gaps and fill the gaps with relevant training material
• It can create individualised training plans for people

To do all this it needs the necessary data. And therein lies an issue – AI needs data to deliver value.

If the data available presents only one side of the picture, or there is insufficient data to offer a full picture, then that is what AI will present. Without the data, AI systems will encourage biases and give incorrect information.

If this is true, there are three questions to address:

1. Can we trust that the education being delivered is comprehensive and unbiased? Can we trust that the training plan created for each member of staff is what they need?

In the human world of education, we recognise subject matter experts (SMEs), experts, tutors, trainers, etc.

We trust them to deliver education that is correct, fair, inclusive and unbiased. We have processes in place to police them. Learning plans are created by Line Managers and L&D Managers with experience, and who know, and have worked with, the person they are building the plan for.

Can AI systems access enough data to offer trusted advice? How can we regulate its advice?

2. Does it matter if the education provided by AI is wrong?

People learn from their mistakes. Can we learn from the mistakes that AI delivers?

A recent piece of research, carried out by the BBC, demonstrated that AI got its facts wrong in about 20% of questions on current affairs.

For example, ChatGPT stated that Nicola Sturgeon was still First Minister of Scotland in February this year – the fact that there have been 2 First Ministers since Sturgeon may be a learning for some.

Errors are not limited to current affairs. A recent AI search stated that Poland was once under British rule. The subsequent discovery that it was not, arguably offers a more significant learning experience than had the original answer been correct.

How correct do the answers provided by AI actually need to be to educate effectively?

3. AI only uses existing data. Who owns it? What are your company and supplier policies around allowable use of AI?

The ethics around the use of AI are a legal minefield, and as a relatively new technology, rules and regulations around its use are still evolving.

AI Usage Policies that ensure education and learning processes are free of bias, that guidelines for use exist to ensure it is used responsibly and correctly, and that data is protected, will be essential to establish credibility and promote online safety.

This will become more and more relevant as the Government’s AI plans gather momentum with a ‘bumper’ Bill to regulate the technology expected next year.

It cannot be denied that the development of artificial intelligence is revolutionising the online training arena – its positive uses are countless and growing – but, as with all new technologies, we have to be careful.

AI is not alive, it cannot replicate human empathy, emotion or even intelligence, so we need human intervention and policy to regulate it. As Nick Clegg, former Deputy-Prime Minister and Meta Executive, said in a recent article in The Times, non-synthetic, non-AI content “is the precious content for truth and for human connection that we will need to hold on to.”

While AI is an exceptionally useful innovation, and its evolution is exciting, it will enhance, not replace, humans in the world of education no matter how sophisticated it becomes.