AI Hype: Smoke and Mirrors

An interesting Twitter thread offers a sobering counter-narrative to the hype around “AI” applications. It is sobering because the media headlines are either dominated by “fear and loathing for AI” or full of hype around apps that are overnight sensations. For example, Lensa, a popular AI portrait app on TikTok, earned $37.5 million in 60 days and has since vanished from the zeitgeist. Lensa isn’t the only shooting star. 

AI features can lure people into signing up, but they have difficulty holding onto them. The author points out that retention problems faced by AI apps are because they have a clunky UX and aren’t focused on simplicity or bringing user delight. There are no network effects, collaborative features, and only short-term value. The success of a solution is directly tied to the frequency of the problem it addresses, and many AI apps target low-frequency issues, diminishing their necessity. No surprise AI apps, at least at present, have short lifespans. 

The elephant in the room is that most of these hot today-gone tomorrow apps are wrappers on OpenAI’s API. As I wrote in my review of MacWhisper, the only way to create longer-term customers is to create a daily usage behavior, and that comes from using AI to enhance a workflow that matters to me, the customer.

Since history is a four-letter word in Silicon Valley, no one learns from the past. The boom-and-bust of these AI apps reminds me of the early days of the iPhone app store when apps like Lighter, IQ Chat, and iFart went up the charts and vanished. It took a while before mega-billion dollars giants such as Instagram and Uber came along. They had one thing, they created workflows that not only solved a problem but created a daily use behavior. 

In 2011, I wrote an essay about what I thought made up for a hit Internet service. A hit internet service has:

1. They have a clear purpose.
2. They are simple to use.
3. They are fun to use or facilitate some type of entertainment or both.

I looked at popular services of the time, the big blow-ups, and the then-ascendant ones. These included Facebook, Instagram, Netflix, Yelp, Skype, Groupon, and Digg. Skype was a fantastic growing app before Microsoft’s acquisition killed the app. Groupon never really figured out its second act. Digg died because it totally messed up. Of the remaining four, three have dominant players in their categories.

In my original piece, I underweighted the importance of network effects, but it wasn’t till the rise of the smartphone that network effects became a crucial differentiator.

Weeks after I published my original piece, I wrote another article, but this time I focused on mobile apps. In my piece, Why some apps work, and some don’t, I pointed out the key reasons why those apps work. 

• Happiness (alternately, enchantment)
• Utility (alternately, solving a problem)

Those reasons are still valid even in this age of “AI.”  And so is the harsh reality — there are no second chances on the Internet. I, for one, am waiting for the Ubers and Instagrams of this new enhanced-software age. 

June 28, 2023. San Francisco