Many have hailed LLMs as excellent tools for learning but is this really the case? Dr Philippa Hardman has done some interesting research to find out if these AI tools can genuinely impact the learning process. And her findings were quite revealing.
She fed one of her own research papers into ChatGPT 4o, Claude 3.5 and Google’s NotebookLM and prompted the AI to summarise it, pull out the key concepts and ensure that a reader could get a clear understanding of the main research and findings. She listed what she found in her LinkedIn post here.
One of the main points that emerged from her analysis was that whilst the AI tools do a good enough job of increasing access to information because they can summarise inputs, this process of synthesising can also act to adapt the tone, form and sometimes even the meaning of the information that has been submitted. She says:
‘…while generative AI tools undoubtedly enhance access to information, they also actively “intervene” in the information-sharing process, actively shaping the type and depth of information that we receive, as well as (thanks to changes in format and tone) its meaning.’
She also makes an interesting point about how, through the act of synthesising complex information, AI tools may actually neglect the nuance and challenge that is required for true learning. Whilst they may be a good first step…:
‘…in order for AI to *truly* support human learning, we need to build tools which deliver the complex cognitive, emotional & psychological conditions that are required to drive meaningful change in how we think and behave.’
The big mistake here, she notes, is that in thinking of them as learning tools we are actually falling foul of a common misconception that learning = access to information. Whilst the improving access to information is useful, the real work of learning remains largely in human hands.
I wrote something over on LinkedIn last week about the experience of using a Custom GPT which I’ve built and trained on my three books and eighteen years of blog posts. I noted that whilst using the tool has proved to be useful in informing the process of writing blog posts (research, inspiration, making connections to other things) I’m steering well clear of asking the Custom GPT to write full blog posts for me. And this is for a simple reason – writing is how I think. The process of writing is how I understand a topic, make connections only I can make, and decide what I think about it. Asking an AI to write for me means that I miss out on all of this.
This feels similar to the point that Dr Hardman makes. The AI tools can be fantastically useful at assimilation, synthesis and helping us to explore, but the real job of understanding and learning still requires that human sensibility.
I write a weekly Substack of digital trends, transformation insights and quirkiness. To join our community of thousands of subscribers you can sign up to that here.
To get posts like this delivered straight to your inbox, drop your email into the box below.
Photo by Kelly Sikkema on Unsplash
Leave a Reply