AI Fatigue

If you’re like me and have opened Twitter lately, you probably stared into the abyss of never-ending AI news.

”Search is going to change forever”

“Chat GPT replaces writers”

“Investors have secretly been using AI over the last 12 months”

“Food critics love the new pizza made by AI”

Ok, the last one is made up, but the intensity is real. Type “AI” into Google News, and you see a tirade of scary or mind-blowing headlines. Yes, I published content about AI as well. I take full responsibility and promise to do it again.

Anyway.

How can you not feel overwhelmed by that? Well, you can’t unless you refrain from social media or block anything with the word “AI” in it. And yet, we’re bound to feel overwhelmed because we also feel FOMO, the fear of missing out. We don’t want to be late when an industry-changing tool or technology comes on the scene. And I argue we shouldn’t.

The fact that two of the most valuable companies in the world, Google and Microsoft, are sprinting - not running - to win the AI race shows this is not a crypto fad. Don’t let me tell you how to live your life, but completely tuning anything out that has to do with AI is likely a mistake. The likelihood AI is going away, and everything is back to pre-Chat GPT times in a year is low.

The question then is: how do you stay informed without breaking your brain in the process?

Innovation jumps

Technology with a wide impact on people's lives always overwhelms humans. We’re feeling it harder this time because the internet is a firehose with a very high noise-to-signal ratio. We’re drinking beer with too much foam out of Oktoberfest steins instead of 12 FL OZ Miller Lite bottles. But the iPhone and Twitter aren't the main culprits here (though they don’t help). It's the sudden introduction of new technology.

Inventions typically nest and brood somewhere in the world for years, and then all of a sudden turn into innovation and hit the masses out of left field. That's where the saying "the future is already here, it's just not evenly distributed" comes from.

It has always been this way. Some of the biggest inventions of humankind took years before reaching the public, as Morgan Housel writes [link]:

• It took 3 years for the first flight of the Wright brothers to be recognized

• Western Union told Alexander Graham Bell that the telephone had “no value” and “too many shortcomings to be seriously considered as a practical form of communication” when he first tried to sell it

• Congress hated Ford’s introduction of the car and threatened to ban the “horseless carriages propelled by gasoline”

• 3D printing has been around since 1984

But once they hit, they hit. Within 50 years, the phone reached 50 million global "users" - not bad for 1875 - 1925. The automobile did it in half the time.

How does that cure our AI fatigue, though?

How to cope with AI Fatigue

I don’t expect the pace of innovation to slow down. We’ll go through waves of highs and lows, but unless we all refrain from the internet or Twitter finally implodes, we’ll keep drinking from steins. There is a lot we can do to better manage how we keep track of new tech. Here is how I approach it.

A lot of the news is speculative, few articles describe what works today. Whenever an article theorizes about how something could be used, don’t read it. Speculation is interesting and entertaining (heck, I speculate quite a lot), but you only really learn from content that explains how something works right now. Replace content about theory with content about application and experience. Filter out the buzz, focus on the essential.

Miller’s Law says humans can keep 7 things in their short-term memories, plus or minus 2. It has since been relativized, but the point that our memory is finite still holds. The best way to outsource short-term memory is by writing a list. Here is what I do when getting overwhelmed: I make a list of all the content I want to read/watch/listen to (in Pocket or Notion) and then put that list out of sight and out of mind. I might never come back to it, but I feel more at peace for having it. There is probably a good psychological principle out there for that, but if not, I offer to call it the Kevin list. Here is where it gets worse, though. Many times, when I saved an article and read it again weeks or months later, it wasn’t as good as I thought.

Timeboxing is how much I spend on trying out new things has proven to be an efficient way to defer overwhelm to later. One hour a week is often enough to try new tools.

All information is not the same. Books offer better insights because they’re typically more thoroughly researched than a Tweet. The older a book, the more insightful it typically is. If I really want to learn something, I read a book about it.

Nothing beats first-hand experience. Trying something out is an even better way to learn than reading. Most new AI toys tools offer free trials or are completely free for a limited amount of time. Use that window of opportunity to play with tools and see what’s really behind them. Many times, tools sell themselves better than they perform.

Technology moves slower than you think

I seek comfort from AI Fatigue and Innovation Burnout in the realization technology always seems to move faster than it actually does. Google holds the third-most patents in AI and invented the Transformer (a key part of large language models, not Optimus Prime), but Search still relies on backlinks. E-readers didn’t kill books. 3D printers haven’t printed cars or houses yet.

Technological process has always been more slow and dispersed than we think. Instead of a single ground-breaking product launch like the iPhone, the steam engine made its way through many iterations around the globe. However, just like the iPhone, the first commercial steam engine combined many innovations and concepts.

An early concept of a steam engine was invented between 30 and 15 BC in Roman Egypt, but it wasn’t capable of doing meaningful work. Over the next one and a half millennia, innovators around the world iterated on the idea of using steam to produce force. Over 100 years before Thomas Newcomen patented the first commercial steam engine, Taqi al-Din described a steam turbine in 1551 Egypt. Giovanni Branca came to the same conclusion in 1606 in Italy. In 1606, Jeronimo de Ayanz y Beaumont received 50 patents to steam-powered inventions, and Denis Papin used the first piston in 1690. Then, in 1712, Necomen brought the steam pump to a level that made it broadly usable. It took another 50 years for James Watt to make critical improvements that kicked off the Industrial Revolution. Was the steam engine invented in a year or over 200 years?

Humans have always been skeptical and overwhelmed by new technology. Some predicted the printing press would never last because they thought handwriting was superior to mechanical printing. When the phone reached critical mass, the press claimed it might keep people from going to concerts because they would just call in. When electric push buttons were invented, people thought they caused human skills to atrophy.

When thinking about the fact that AI has been researched for over 60 years and the Transformer was invented in 2017, I feel AI refreshed.