The Economist__High-quality text to train AI will run out by 2026

A recent article published in The Economist states that there is an ongoing “war” for data (to train AI models, which need 1 trillion words, a value that is much greater than the size of Wikipedia which has “only” nearly 4 billion words). The article states that demand for this data is growing so quickly that the stock of high-quality text will run out by 2026 https://www.economist.com/business/2023/08/13/ai-is- setting-off-a-great-scramble-for-data

The (very) positive part of the aforementioned article occurs when it recalls that, for example, Microsoft paid 7.5 billion dollars for a software repository and also that there is an increase in the number of content creators (whose materials were used in the training of AI models) who decided to take legal action to be financially compensated for this unauthorized use and the key phrase is “Holders of information are taking advantage of their greater bargaining power“

The article mentions that AI firms also collect data from users’ interactions with their models, in that context, it is pertinent to revisit a post from July 26th, where I disclosed the fact that over 7 months I had made 624 requests to ChatGPT (almost 100 per month) which resulted in a text history that gives a total of 550 pages (Arial 11). However, taking into account that the new history that I requested from ChatGPT today, refers to a total of 884 requests, which represent 1034 pages, that means that in the last 30 days, I made 260 requests to ChatGPT that translated into 484 pages of text ( Arial 11). In comparative terms, the more than 300 academic publications of which I am the author and co-author represent almost 5,000 pages of text, but that number still lacks another 1,000 pages, corresponding to reviews of several hundred articles, doctoral theses, and research projects.

PS – Not to mention the other thousand pages of text that I have produced for the blogosphere since the end of 2019, in posts that were read by visitors from 150 countries, in which those from Finland and Norway deserve special mention.