Time Is a Torrent: How to Catch the AI Wave Without Drowning

Sword or torrent?

People often say that "time is like a sword: if you do not cut it, it cuts you". A sword is a tool in your hand, and you decide when to draw it. A torrent is a collective force that does not wait for your decision: it carries those who board its current and passes by those who hesitate on the bank. Is AI a torrent? There is some truth in the comparison, but it deserves to be tested against figures before we turn it into a slogan.

What do the figures say?

In August 2024 a US study surveyed more than five thousand people aged 18 to 64 and found that 39.4% had used generative AI. That was about two years after ChatGPT's launch, roughly double the rate of personal computer use three years after the IBM PC's launch in 1981, and also higher than the internet at a comparable stage. Workplace use, however, was close to that of the PC (28% of workers after two years against 25% after three), so most of the difference came from use at home. The 2026 edition of the Stanford index adds that generative AI reached 53% of the population within three years, faster than the computer and the internet.

The figures have limits: they are American, and they measure use, not depth or benefit. Trying a tool is one thing; changing how you work with it is another.

Kodak's story as it happened

Kodak's tale is usually told as that of a company that slept on its own invention. The facts are more complicated. In 1975 the engineer Steve Sasson built the first early digital camera. It weighed about 3.6 kilograms, captured a black-and-white image of 0.01 megapixels on a cassette tape, and took 23 seconds per picture. Sasson recounts that the executives were convinced nobody would want to look at their pictures on a television set. Kodak then dominated the American photography market and was unwilling to undermine its film business.

Yet it did not neglect digital afterwards. In 2005 it ranked first in the US digital camera market for the second year running, with a 24.9% share according to IDC, shipping 7.05 million units to retailers. Then it filed for Chapter 11 bankruptcy on 19 January 2012. The lesson is not that it was ignorant of change, but that knowing about it, and even leading a new market, did not shield it from the disruption of a business model built on film. That is an interpretation open to discussion; what the sources establish are the facts above.

Five steps for someone who fears being late

You do not need to learn everything. Here are five small steps you could try this week:

1. Swap teaching: agree with a colleague or relative who loves technology to exchange expertise: you learn one tool a week from them, in return for your experience in your field.
2. Narrow the field: choose a specific task you love and are good at, and learn how the tool serves that alone.
3. Start from the problem: ask what drains most of your time at work, then look for a tool for that, not the reverse.
4. Be a bridge: explain what you tried in plain language to colleagues or family; teaching others fixes what you have learned.
5. Set a trust checkpoint: do not enter sensitive data, and review the tool's output before relying on it (see the articles "Three Myths About Artificial Intelligence: Fluency, General Intelligence and Neutrality" and "Privacy in the Age of AI: How Your Secrets Are Inferred, and How to Protect Them").

Beware of haste

The feeling of a torrent can drag you into chasing every new tool, which is a kind of drowning. The praiseworthy speed is to begin now with a small, regular step whose benefit you measure, taking the means and relying on God for the rest. God willing, this widens your ability to tell what benefits from what merely distracts.

The race is not with others but with our older version of knowledge: to leave behind what sufficed yesterday so that we may learn what tomorrow requires.