Does ✨AI Powered✨ Matter?
Ben Thompson at Stratechery1 wrote an article in 2013 why the proprietary nature of the iPhone was defying Clayton Christensen’s “Theory of Disruption.”
Christensen’s theory is based on examples drawn from buying decisions made by businesses, not consumers.
The reason this matters is that the theory of low-end disruption presumes:
Buyers are rational
Every attribute that matters can be documented and measured
Modular providers can become “good enough” on all the attributes that matter to the buyers
All three of the assumptions fail in the consumer market, and this, ultimately, is why Christensen’s theory fails as well.
This makes sense, businesses, generally, have processes in place to force rational decision making and finance departments to push for functional and good enough tools. On top of that, avoiding vendor lock-in and being able to pit vendors against each other are high priorities for enterprise customers2.
But consumers don’t have to play by those rules. They are a buying committee of one that can pay extra for luxury, status, brand, process, mission, or just because they want to. Apple’s success also shows that consumers will pay a premium for ecosystems that “just work.”
With all of that, will ✨AI Powered✨ be the next big push of consumer products?
Maybe not…
Joe Reis recently wrote in his newsletter:
I think it’s essential for practitioners and product managers to understand how to use data and AI to create better products. But I don’t believe the end customer cares whether a product is “AI-powered.” People want products that work. (emphasis added)
Thinking about that I realized two things
✨AI Powered✨ may be the new “Bluetooth enabled”
Enterprises and consumers have switched places
The new Bluetooth
Bluetooth is a great technology, but it is in a lot of products that doesn’t make sense3. Great Bluetooth products get rid of all the wires in our lives: headphones, speakers, microphones, and keyboards. Bad Bluetooth products solve problems no one has or in more complicated and expensive ways.
GenAI has many of the same risks. The most prominent examples of consumer GenAI are all chatbots. But I don’t need a chatbot in most parts of my life. An improved Siri or Alexa? Definitely! A pin on my shirt? No thanks.
A friend and former colleague of mine would always remind me that when it comes to Data Science/ML/AI the best implementation fade into the background, you don’t sell on the process you sell on better results. This is a tough lesson for a lot of people working in data because they care about the process.
“Bluetooth enabled” is selling the process. ✨AI Powered✨, in most current products, is the same thing. “Pay more because of how we do it,” but does it work better?
Corporate status symbols
So with all of that you would think that ✨AI Powered✨ would hold no sway over business and enterprise customers, and you would be wrong. The mad rush to make everything ✨AI Powered✨ is not only coming from vendors it is being pushed by enterprises. The current FOMO and investor demand that everyone have an “AI strategy” is forcing vendors to stick GenAI somewhere in their product so they can answer the question “How are you using AI to make the product better?”4 Combine that with “unlimited” budgets for AI and you have a recipe for the corporate version of status purchases.
While there are great pronouncements about LLM and GenAI there is still very little evidence it is currently changing much on anything but corporate buying patterns.
Is GenAI is completely disrupting software development? Or only improving efficiency 20%5?
Maybe the hype will last long enough for the tech to catch up, but unlike consumers, companies can’t continue to make status purchases. Eventually rational heads will take over and start asking “why are we paying 20% for this tool?” “What are we getting for this?”
It is impossible to tell when that correction will happen6, but it will. It may be the rare occasion when consumers are more rational than corporations.
https://arc.net/l/quote/ogjxgind
Also how to measure coding efficiency is a slippery thing to nail down. So even if you see “50% increase” read the fine print. That can anything from “how fast the task was completed” to “how quickly PR were merged”
This may last longer than you would think because CFO seem to be in on the hype this time around.