Being Human in the World of AI

Being Human in the World of AI

Sir Arthur C. Clarke once said, “Any sufficiently advanced technology is indistinguishable from magic.” And whilst Clarke made this observation long before ChatGPT was even a twinkle in Sam Altman’s eye, one could be forgiven for thinking he was describing GenAI.

In 1980, Seymour Papert, a leader in both education and technology and one of the co-creators of the Logo programming language (remember that turtle?), defined AI as being “concerned with extending the capacity of machines to perform functions that would be considered intelligent if performed by people.”

One can’t help but wonder what these two gentlemen would think about Generative Artificial Intelligence in 2026. This post and my upcoming presentation at the National Education Summit in Melbourne will not be yet more examples of how AI will save education and the world as we know it. Nor a diatribe against AI and how it will doom humanity. Instead, I aim to challenge us to look beyond the magic and to develop an understanding of what it means to be human in the world of AI.

Many of us are aware, at least to some extent, of the environmental impacts of GenAI – the increase in energy and water use, the creation of e-waste. But what has been given less exploration and airtime is ways that these environmental impacts can be somewhat mitigated. Being human in the world of AI means using our combined humanity to minimise the impact on the planet. In terms of water usage, other alternatives to evaporative cooling are starting to be explored, including air economisers, recycled water, non-potable water and even zero-water designs. Shifting to renewable sources such as wind and solar will reduce the overall water and non-renewable energy consumption of AI. Running more intensive jobs at nighttime can also save water, due to cooler ambient air. This is also an area where AI has the potential to help solve the problems of AI, by using AI-powered leak detection so that issues can be identified before millions of gallons are lost in the cooling process.

Currently, the huge expansion of data centres is outpacing the expansion of renewable energy generation. Designing systems so that some computer operations can be performed later, when more of the grid is using renewable energy, can also have a positive impact on a data centre’s carbon footprint. As with leak detection, machine learning could also be used to tackle complex problems such as improving the electrical grid.

Whilst the environmental impacts of AI have been explored somewhat, some of the human ethical concerns around AI use have not received the same attention. I will touch on these briefly here, but these concerns will be explored further in my upcoming presentation during the Summit.

As a teacher, one of my key concerns around AI use is that of equity. When schools (and higher education institutions) use AI-detecting software, they run the risk of unfairly and unintentionally targeting students who come from a non-English-speaking background. Research carried out by Stanford in 2024 found that these AI detectors misclassified more than 60% of writing done by someone who has English as an additional language. 60%. This is because so-called AI detectors don’t actually ‘understand’ writing – they simply analyse patterns of elements such as word predictability.

We have also seen huge growth in AI-driven tutoring systems, and whilst this has some potential benefits in terms of equity and access, it is possible that these systems may also be catering less effectively to students from backgrounds who are not as well represented in the training data for the model. In addition to this, in an education system as stratified as Australia’s, we have the situation where not all students have access to the same LLMs and models and – this is where teachers are key – not all students have been provided with the same education around how to most effectively use AI to improve their learning.

Another key area where we as teachers need to encourage our students’ thinking and exploration is around bias. ChatGPT, the OG of AI platforms and one frequently used by many students we teach, was originally trained on a big dataset that was scraped from 12 years of websites, from the approximate period of 2008–2020. The scraped data includes: public websites, digitised books and academic papers, social media posts and user-generated content, as well as code repositories like GitHub. These datasets can exclude marginalised people and minorities. As most AI platforms that we as educators will be using now have an internet connection, this bias is somewhat mitigated in that information is more up to date. However, of course, that also relies on what is currently published on the internet being free from bias.

Whilst many users of AI may believe that the technology is doing all the hard lifting, the reality is that the ‘magic’ of many AI platforms comes from an invisible workforce, due to a process known as data enrichment, data annotation or data labelling. Think about self-driving cars. For these cars’ systems to be able to distinguish a stop sign from, say, a pedestrian, humans – ‘data enrichment workers’ – have had to manually annotate thousands of images of stop signs. Now think about guardrails that exist in platforms such as Copilot, and what humans will need to be exposed to for these protections to exist for platform users. For the initial ‘machine learning’ to occur, humans had first labelled millions of images, for as little remuneration as one cent per image.

How can we ensure that AI is not built on an invisible, underpaid and unempowered workforce? The first step is awareness. And education. Read up and research these cases. Talk to your colleagues about them and talk to your students. Explore the terms ‘digital colonialism’ and ‘digital sweatshops’ and how workers are resisting and demanding better working conditions. Former US President John F. Kennedy once famously said, “Ask not what your country can do for you—ask what you can do for your country.” The next time you are using AI, perhaps ask not what Copilot can do for you, but what its use enables, shapes and costs. And how can these negative impacts be mitigated? When using AI in class with your students, encourage them to question – from where did this output come? Whose voices may be missing or misrepresented? How can we use technology – including Artificial Intelligence – to improve the environment and the lives of humans? I will be exploring some of these questions and ideas further at the National Education Summit in Melbourne, and I would love as many people as possible to join the conversations so that we are shaping the use of technology for the good of humankind. Not being shaped by technology.

AI transparency: Copilot Researcher Agent was used for initial research.

Claude was utilised for proofreading.