Will We Still Need Words?

Here is a question that sounds like science fiction and is really philosophy: if you could share a thought with another person directly, without turning it into words first, would you still want words at all?

I think the answer is yes, but not for the reasons people expect. This essay is the speculative end of the argument I build across the journal, starting from how AI is changing human language.

Words are a compression format

Start with what a word actually does. You have a thought, rich and specific and partly wordless. To share it, you compress it into a small sequence of agreed symbols, send those through the air or onto a screen, and hope the other person decompresses them into something close to your original.

It is a lossy format. Most of what you meant does not survive the encoding. Anyone who has failed to say what they felt knows the loss personally. Seen this way, language is not the thought; it is a heavily compressed transmission of it.

That framing comes straight out of the evolution of language: each stage, from speech to writing to code, was a better way to store and move that compressed signal.

What direct transfer would change

Now imagine the compression step getting much cheaper, whether through AI that helps articulate thought or, far more speculatively, through the brain-computer interfaces now in their earliest medical form.

The gains are obvious: more bandwidth, less loss, less of the gap between what you mean and what you manage to say. The book opens on exactly that gap, because closing it is the headline promise of the whole shift.

The losses are subtler, and this is where I part company with the optimists. The friction of putting a thought into words is not only a cost. It is part of how thinking works. To say something clearly, you have to find out what you actually think. Remove the friction entirely and you remove some of the refinement it forces. A frictionless channel could make us faster and shallower at once.

Why words probably survive

Even granting all of that, I do not think we stop needing language. Three reasons:

1. Compression is a feature, not only a flaw. A lossy summary is often more useful than the full signal. You do not want to transmit your entire mental state; you want the part that matters, made explicit.
2. Words are how we think alone. Even with no one to talk to, language is the tool we reason with. That does not disappear because transmission improves.
3. Shared meaning still needs a shared code. Two minds can only exchange meaning to the degree they share a model of the world. Words are that shared code, hard-won over millennia.

What changes, then, is not that language dies, but that it stops being the only interface. That is a smaller claim than “the end of words,” and a stranger one to live inside.

The real question

So the honest version of “will we still need words” is not yes or no. It is: when words become optional rather than mandatory, what do we keep using them for, and what do we quietly stop saying altogether? The things we stop bothering to put into words are the things we may stop fully thinking.

That is the question at the centre of Building Your First Brain. I do not pretend to settle it. I try to map it honestly, before we arrive. The book is free for the first 1,000 readers.

Further reading

• The deeper background on whether models that wield words actually grasp them: do large language models understand language.
• The llms.txt proposal, a small, real example of language being reshaped for its new, non-human readers.