There are many reasons to be wary of claims that AI will end up being a wholly positive presence in education. Alongside the blatant big ticket ‘AI harms’ and ‘AI wrongs’ that are now being regularly raised in critical discussions are various subtler changes as AI tools become entrenched in classrooms and other educational settings. These are changes to the conditions under which people engage with education – small shifts and adjustments to the character of teaching and learning which, while each seeming relatively minor when seen separately, soon add up to a significant recalibration of how education is provided and experienced.

Many of these changes involve people having to adjust what they do and how they act in educational settings that are infused with AI. One telling example of this are the various ways in which students and teachers might find themselves having to talk in the ‘always listening’ classroom – i.e. learning spaces stuffed full of microphones, cameras, smart speakers, voice recognition and transcription software, and other forms of AI recording. In theory, the mass extraction of voice data vastly increases what is known about the learning process and therefore can enhance the support that students can be given. However, in practice these classroom conditions are arguably far more off-putting and restrictive putting.

For example, one of the obvious features of the always listening classroom is teachers and students having to speak in standardised ways that they know can be picked up and parsed by computers – not speaking too quickly, being mindful of enunciation, concealing regional accents, avoiding dialects and speaking at volumes that can be picked up (or not) by the nearest microphone. Savvy students will learn to ‘prompt hack’ their school’s AI – parroting phrases that systems are programmed to take as evidence of learning (such as “I learned that” and “I conclude”). Savvy teachers might well follow suit and start speaking in similarly algorithmically-appeasing ways – perhaps aping the workplace strategies of call centre workers and gratuitously peppering conversations with declarations of ‘I feel’ and ‘sorry’ to boost automated sentiment analysis ratings. At the same time, students looking to fly under the radar will understandably slip into modes of speech that cannot easily be parsed by AI – forms of algorithmically-obfuscating slang, cant and anti-language.

Having to make these adjustments and alterations to how one speaks might not seem a major inconvenience but replicates the general trend for AI to standardise and homogenise what goes on in classrooms. At the very least, speaking like an AI is likely to take some of the colour, spontaneity and character out of classrooms. More seriously, this particular form of AI-flattening will undoubtedly make classrooms less welcoming and less inclusive places. Indeed, the ways in which we speak are an essential part of our social and collective identities. Tone-policing and micro-aggressions around one’s capacity to speak ‘correctly’ have long been part of how schools marginalise students from cultural and linguistically-diverse backgrounds. Now, having to speak in ways that are algorithmically acceptable further compounds what Basil Bernstein termed the ‘elaborated code’ required for academic success at school and university – modes of communicating that are often unfamiliar to many minoritised students such as those from working-class backgrounds. 

The ways that we speak and the words that we use, are key parts of being able to think, learn and thrive in any educational setting. Restricting the ways that language that can be used in a classroom is therefore clearly disadvantaging for many – an implicitly political way of ensuring that some people are further advantaged over others. If nothing else, the inability of AI to deal with the full range of human expression will inevitably lead to hollowed-out, flat-sounding forms of teaching and learning. Language, dialogue and conversation is at the heart of good teaching and learning. Anything that limits what and/or how things can be said within a classroom needs to be treated with the utmost suspicion.