Predictive text is turning 6 years old

In The Devil Wears Prada movie (2006), the protagonist's dad asks her why she is not working as a writer, to which the protagonist answers:

"I wrote those emails!"

Note: Everything I write here is written by me without AI help. Hence, any grammar mistakes are mine.

Google introduced Smart Compose in May 2018, exactly 6 years ago.

By 2019, it was already saving 2 BILLION keystrokes per week.

In only 6 years:

There's predictive text everywhere: when you are texting on your phone, on LinkedIn, Meta and Microsoft apps, you name it.

ChatGPT officially launched, replacing GPT-2 which was essentially predictive text on extremely powerful steroids.

Gemini replaced Smart Compose in Google products.

We have gotten used to clicking on Tab to finalize the sentence without having to type it that we do not notice we are doing it.

We have gotten used to opening Gemini, ChatGPT or Claude and have a small conversation, requesting them to write a poem in the style of Hemingway, and to ask questions instead of searching on Google.

Universities and schools are still working to put a lid on students using any of those tools to write their essays.

All of this in only 6 years.

To spin your wheels: 2030 is also 6 years away. We are at the midpoint between those two, and I can't wait to see the advances that predictive text and LLMs will have.

To spin your wheels even further, and as of the time of this publication, both OpenAI's GPT4o and Google's Gemini 1.5's context window is 128 THOUSAND tokens.

• For an easy explanation of what Tokens are, check link in the comments
• Side note for context (pun intended): the context window is the amount of text or tokens that the model considers when generating or predicting the next word. A longer context window allows the user to have much longer conversations before the LLM starts to forget or hallucinate things. Although forgetfulness, retrieval precision and hallucination are still an issue in all current models, but perhaps that will not be an issue in 6 years.

What does that say about human's uniqueness and creativity as individuals? Ultimately it's also a question of ethics: do students 'fess up' when they have used Gemini, ChatGPT, Claude or any other AI to 'enhance' their writing? do people say "I wrote you that email using an LLM?"

One thing is clear: Had the Devil Wears Prada been filmed in 2024, Andy (the protagonist) would not be able to tell to her dad to his face "I wrote those emails!" without batting an eyelash.

I have always wanted to be a writer, and for what’s worth, my teammates have told me that I write beautiful emails -if "beautiful" has a place in business strategy.

I have a 300-page fiction book manuscript in a drawer that someday, someday I will edit, without the help of AI.

In the meantime: "I wrote those emails, (and this newsletter)!"

The devil wears a Large Language Model...

PS. The new episode of The Leadership, Strategy+AI podcast is live! Check it out, link in the comments.