Another profound consequence of LLMs is that the distinction between literacy and its educational 'poor cousin' oracy is becoming blurred. This is partly as a result of the extraordinary pace of evolution in voice AI such as voice-typing, editing tools and automated voice generation.
The opportunities are great when these technologies make us more productive and efficient in our communications. They are also levelling the playing-field for dyslexic and other neurodiverse youngsters, and for anyone whose talents do not easily align with traditional learning and assessment methods. Pupils previously judged to be less intelligent because they struggle with reading or producing good written output, can now take pride in demonstrating their intelligence and creativity. They are reminding us just how articulate they are.
We should keep in mind that the conversational qualities, phonetic shortcuts and emojis of WhatsApp, X and other social media formats bear strong resemblance to oral communications. More than a decade ago, John McWhorter coined the term 'fingered speech' to make the point that texting and instant messaging are arguably nearer to speaking than to writing in the conventional sense.
Literacy and oracy are two sides of the same coin in the currency that humans need in order to thrive in an AI future. But one risk of the blurring of literacy and oracy is that we spend less time reading and thinking (or at least less patiently, diligently and critically), whilst copying, pasting, and typing/dictating messages much more rapidly. Across electronic media, words are often moving faster than meaning and understanding.
We now have a generation of young adults skilled in fast-paced, informal channels and protocols of communication. This can deliver productivity gains. But it can also be to the detriment of clarity, attention to detail and understanding. Many younger people admit to disliking or struggling with the protocols of more carefully considered and courteous correspondence such as email*.
The big question is: how concerned should we be about this?
Posing a more precise question, with reference to Daniel Kahneman's influential 2011 book Thinking, Fast and Slow: has social media triggered a generational shift towards instinctive, unconscious and emotional thinking and away from more deliberative, logical and analytical thinking? If so, what are the long term consequences for human relations and for human wisdom in the age of AI?
A return to an oral tradition?
When considering the questions above I try to suppress my personal bias and concerns (as a one-time English teacher) about declining standards in long-form reading and writing skills. After all, speech is natural and came first.** We are born with the neural wiring for speech, but reading and writing are less ‘natural’ skills, only acquired through education and through evolutionary technologies such as the alphabet, printing and publishing. As far back as the 4th century BC, Socrates (in Plato’s Phaedrus) tells of the Egyptian God Theuth's debate with King Thamus on the pros and cons of granting the Egyptian people the benefits of writing. On the one hand, the technology of writing extends memory and shares wisdom. On the other hand it will:
‘create forgetfulness in the learners’ souls, because they will not use their memories...they will appear to be omniscient and will generally know nothing.’
This may be the very earliest reference to what is now known as cognitive offloading***: 'why do I need to burden my brain by retaining knowledge when I can easily look things up?'
Hence emerged an enduring but sometimes chafing alliance between documented wisdom and oral eloquence. Does the value of knowledge live in static, unresponsive text, or does it only emerge dialogically from a relationship between active, well-informed human minds through spoken discourse? The answer lies somewhere in between, but precisely where will depend on cultural context and whether we are introverts or extroverts, engineers or imagineers.
In more recent centuries (since the invention of printing and publishing systems) mastering the skills of consuming, re-producing and expanding the 'canon' of curated published text has dominated the academic and professional pathways to success. This is sometimes referred to as the age of the linear literary mind.
But we can now see the beginning of the end of the age of information eco-systems built on such reverence for curated printed text. This is partly because of the proliferation of non-curated and self-published electronic text. We are inundated by hyperlinked documents, web pages, bulleted slides, emails and multiple channels of news, promotions and instant messaging. We are all authors now.
So it has become impossible for us to read everything that appears on our screens, making it much more difficult to differentiate the wheat from the chaff. This goes some way to explaining the enormous popularity of podcasts, vlogs and social media influencers. They are the latest channels of curation, with relatively low barriers to entry and minimal quality-assurance mechanisms.
With such growing reluctance to read long-form text with care, there is now a huge demand for it to be synthesised and summarised by LLMs. This can be easily achieved, including by speaking instructions into our devices. It should be no surprise that there is this desire to develop more productive and dynamic oral interactions with technology, and with 'big data' in particular.
But oracy is not just about speaking. It is the ability to articulate ideas, actively listen, develop understanding and engage in dialogue with others through spoken language. If the processes of human knowledge-generation and learning really are leaning back towards to the oral traditions of our species, we certainly need to 'up our game'. At the very least we need to put oracy on an equal footing with literacy in schools and reflect this parity in our assessment and accreditation systems.
I explore some of the consequences of this trajectory towards a new age of oracy, along with the importance of finding sustainable career nutrients, in the website section "All the World's a Stage" - Securing Human Roles in an AI Future.
The campaign for greater recognition of oracy skills has been running for a long time, supported by many impressive educationalists. The excellent report of the Oracy Education Commission (October 2024) is well worth a read. I am also very impressed by Voice21, the UK's oracy education charity. They have been active for over a decade, working with schools to transform the learning and life chances of young people through talk, and campaigning for oracy to have a higher status in the education system.
Stepping up to the challenge
Our curriculum, assessment and teacher-training methods are terribly slow in adapting to all the evolving aspects of our culture. This puts pressure on individual teachers to figure things out for themselves, and to experiment. Some are resistant to change. Others are embracing generative AI, mostly to produce/adapt learning materials, but also (in more advanced usage) to support the assessment of their pupils' needs and their output. But meaningful and equitable progress in the profession of teaching is hampered when the institutional systems of learning and assessment show no signs of evolving to accommodate the ways in which the world is changing.
The sector must get over its fear and build pro-actively on the increasing evidence that AI augments human teaching rather than replacing it. When developed and applied rigorously, AI and other Edtech not only enhance learning, but can also make teaching a more sustainable and rewarding career-choice.
Technology, especially the tools that can tailor individual learning pathways and provide formative assessment feedback, can free teachers up to focus on the more human aspects of their roles: creativity, ingenuity, empathy and dialogue. 40 years after the term was first coined, perhaps the flipped classroom*** is finally coming of age? With more time and data to identify and nurture the strengths of each child and to respond to their differing learning needs, teachers can better guide them to a more resilient and self-determining future of lifelong learning. In my view this is the most impactful strategy for addressing the teacher recruitment and retention crisis.
Conversely, the greatest risk to the teaching profession is to ignore this evidence, passively allowing learners to develop the habits of using AI more as an 'answer-machine' than as a learning tool, and wasting time trying to figure out whether a student has cheated. Whenever I hear people complain of students using LLMs to cheat, I worry that we are missing the big opportunity and misdirecting our efforts.
We should not delude ourselves that AI in education is primarily a technical problem for the technologists to solve. It is first and foremost a teaching, learning and assessment challenge that the education profession needs to own. Of course there are complex aspects of the challenge which are cultural, political, legal and technological, but we must not use that as an excuse to disown the challenge.
Like any new technology, the worthy and aspirational goals of generative AI will be shaped, and potentially corrupted, by the demands of those investing in it. Or by those influencing its application in everyday life. For all the reasons above, leaders in education have a responsibility to contribute to the development of quality assurance processes and governance structures for the effective and ethical use of AI in the areas over which we have influence.****
Footnotes
* I am referring to the email protocols based on traditional physical correspondence such as letters, though this may be dying out.
** I am indebted to Nicholas Carr's brilliant book The Shallows (2010) for some of the content in this paragraph. His latest book Superbloom - How Technologies of Connection Tear Us Apart (2025) is equally brilliant.
***Described by Professor Dylan Wiliam in 2017 as 'the single most important thing for teachers to know', Cognitive Load Theory is now even more important to build into our teaching, learning and assessment standards.
**** The flipped classroom is the modern version of Socratic teaching. It aims to increase engagement and learning by having pupils read things privately/at home and then build on that learning in the classroom through collaborative problem-solving, interactions and discussion. I have long been an advocate of Century tech, a pioneer of adaptive learning that supports this direction in education. Dan Fitzpatrick's excellent book Infinite Education (2025) also addresses similar themes in more depth than I can cover in this essay.
***** The Institute of Ethical AI in Education (IEAIED) has produced this useful framework.