The Augmented Human: When Does a Tool Become Part of Our Thinking?

In a famous thought experiment, a man called Otto has Alzheimer's disease. He carries a notebook in which he writes what he wants to remember and consults it whenever he needs to, and with this notebook he finds his way to the museum just as a woman called Inga finds hers using her memory. From this example the philosophers Andy Clark and David Chalmers posed a question in their 1998 paper "The Extended Mind": when is the notebook part of memory, rather than merely an aid to it?

An old augmentation

The story does not begin with artificial intelligence. In the Phaedrus, Plato tells how King Thamus warned the inventor of letters that they would weaken people's memory rather than strengthen it. In Understanding Media (1964), Marshall McLuhan argued that every medium is an extension of some bodily function, so that the wheel extends the foot, and that every extension carries a kind of "amputation" of that function. None of this stopped people from writing or riding wheels, but it taught us to ask what recedes when something else grows stronger.

What is new?

What is new is that the tool is no longer inert. One can distinguish three layers: devices we wear that supply information, translation and reminders; large language models with which we share the drafting of ideas and the arranging of arguments; and interfaces that connect the brain to a computer. The third seems the most distant, yet its therapeutic value is documented. In a 2023 Nature study, a woman with amyotrophic lateral sclerosis used a neural implant that reads her attempts to speak; the average decoded speech was 62 words per minute, with a word error rate of 9.1% on a 50-word vocabulary and 23.8% on a 125,000-word vocabulary. Here augmentation returns to a person some of her ability to communicate, among the noblest uses of technology. The line between treatment and optional "enhancement" remains a question for competent authorities and for ethical and Sharia guidelines, not for companies alone.

What might we lose?

The risk in the second layer is subtler because it is comfortable. In four experiments published in Science in 2011, participants remembered where information was saved better than the information itself: about 49% for the folder against about 23% for the statement. The researchers concluded that we remember less by knowing the information and more by knowing where to find it. Caution is needed here: these were laboratory experiments on simple statements, and they do not show that using tools weakens the mind in itself; it may free attention for what matters more. They do warn that what we do not practise we may abandon, and that convenience tempts us to delegate more than we intend.

Who owns the extension?

If we accept Clark and Chalmers's idea that tools we rely on regularly and with trust may be part of the machinery of cognition, a question follows that they did not ask: what if the notebook belongs to a company that decides what is shown in it and what is hidden? That is our inference, not their text, but it explains why transparency matters, as does the user's right to choose their tools and delete their data. The principles of transparency and accountability are discussed in the article "Responsible Artificial Intelligence: Four Principles for Governance and Trust".

Questions to ask of every tool

1. Who controls this tool and decides what it suggests to me?
2. Could I do the task myself if it were unavailable?
3. Does it replace a skill I want to keep alive? If so, set aside time when the tool is not used.

The mind is a blessing, and we give thanks for it by using and maintaining it well. Useful augmentation is what increases our capability while keeping responsibility for decisions in our hands, not what changes who decides. Our reliance on tools is likely to grow; what remains ours is to set the terms and limits of that reliance.